Updated on 2025/06/19

写真b

 
HAN Xianhua
 
*Items subject to periodic update by Rikkyo University (The rest are reprinted from information registered on researchmap.)
Affiliation*
Graduate School of Artificial Intelligence and Science Master's Program in Artificial Intelligence and Science
College of Science Department of Mathematics
Graduate School of Artificial Intelligence and Science Doctoral Program in Artificial Intelligence and Science
Title*
Professor
Degree
工学博士 （ 琉球大学 ）
Research Interests

  • Machine Learning, Pattern Recognition, Computer Vision

    • Campus Career*
    • 9 2023 - Present 
      Graduate School of Artificial Intelligence and Science   Master's Program in Artificial Intelligence and Science   Professor
    • 9 2023 - Present 
      Graduate School of Artificial Intelligence and Science   Doctoral Program in Artificial Intelligence and Science   Professor
    • 9 2023 - Present 
      College of Science   Department of Mathematics   Professor
     

    Research Areas

    • Informatics / Intelligent informatics

    • Informatics / Perceptual information processing

    Research History

    • 9 2023 - Present 
      Rikkyo University   Graduate School of Artificial Intelligence and Science   Prof.

      ▶ More details

    • 3 2017 - 8 2023 
      Yamaguchi University   Associate Professor

      ▶ More details

    • 4 2016 - 2 2017 
      AIST   Senior Researcher

      ▶ More details

    • 4 2013 - 3 2016 
      立命館大学 グローバルイノベーション研究機構 研究教員（准教授）

      ▶ More details

    • 4 2010 - 3 2013 
      立命館大学 グローバルイノベーション研究機構 ポストドクトラルフェロー

      ▶ More details

    • 4 2007 - 3 2010 
      立命館大学 総合理工研究機構 ポストドクトラルフェロー

      ▶ More details

    • 2 2006 - 3 2007 
      中国中南林業大学 情報学科 講師

      ▶ More details

    ▼display all

    Education

    • 10 2002 - 9 2005 
      琉球大学大学院

      ▶ More details

    • 9 1999 - 7 2002 
      中国山東大学大学院

      ▶ More details

    • 9 1995 - 7 1999 
      中国重慶大学

      ▶ More details

    Awards

    • 5 2012  
      日本私立学校振興・共済事業団  学術研究振興基金(若手研究者奨励金) 

      ▶ More details

    Papers

    ▼display all

    Misc.

    • Detection of Wheat Heads by CenterNet with Atteniton Module

      井上源太, HAN Xian-Hua

      電子情報通信学会大会講演論文集(CD-ROM)2022   2022

      ▶ More details

    • Robust Hyperspectral Image Super-Resolution via Constrained Non-negative Sparse Representation

      韓先花, 山口大, BoXin Shi(AIST, Yinqiang Zheng(NII, Toru Kouyama(AIST, Atsunori Kanemura(AIST, Ryosuke Nakamura(AIST

      第20回 画像の認識・理解シンポジウム   8 2017

      ▶ More details

    • Diagnosis of Alzheimer's disease by structural MRI-Validation of efficiency of AI-derived Alzheimer's disease score

      Shiino Akihiko, Iwamoto Yutaro, Han Xianhua, Chen Yen-Wei

      Cerebral Blood Flow and Metabolism (Japanese journal of cerebral blood flow and metabolism)28 ( 2 ) 303 - 308   2017

      ▶ More details

      Language：Japanese   Publisher：The Japanese Society of Cerebral Blood Flow and Metabolism  

      <p>Voxel-based morphometry (VBM) uses structural MRI data to investigate brain region volumes in a voxel-wise manner, not unlike computing z-scores in SPECT using eZIS or iSSP. Recently, we added artificial intelligence (AI) to our software "BAAD" (Brain Anatomical Analysis using Diffeomorphic deformation) that was originally developed to support diagnosis of Alzheimer's disease (AD). The AI combines support vector machine (SVM) with a radial basis function (RBF) kernel, and cost functions and slack variables were optimized using data from the ADNI database (314 cases, 386 healthy controls). The probability of AD is computed by BAAD from the set of all regions of interest and is shown as an AD score (ADS). The accuracy and post-diagnostic odds ratio using BAAD AD scores were assessed at 89.6% and 134.1, respectively. We used the AIBL database (72 AD cases, 447 healthy controls) as an application phase for validation by comparison to results from VSRAD (voxel-based specific regional analysis system for AD) software. The accuracy and the post-diagnostic odds ratio for AD scores were 86.1% and 47.9 for BAAD but 84.8% and 14.9 for VSRAD. This suggests that the BAAD approach more fully exploits the potential of structural analysis to support AD diagnosis.</p>

      DOI： 10.16977/cbfm.28.2_303

      CiNii Article

      researchmap

    • A NOVEL AND FAST CONNECTED COMPONENT-COUNTING ALGORITHM BASED ON GRAPH THEORY

      ICIC express letters. Part B, Applications : an international journal of research and surveys7 ( 12 ) 2625 - 2632   12 2016

      ▶ More details

      Language：English   Publisher：ICIC International  

      CiNii Article

      researchmap

    • HEp-2 cell classification using K-support Spatial Pooling in Deep CNNs (医用画像)

      Han Xian-Hua, Chen Yen-Wei

      電子情報通信学会技術研究報告 = IEICE technical report : 信学技報116 ( 225 ) 1 - 6   16 9 2016

      ▶ More details

      Language：English   Publisher：電子情報通信学会  

      CiNii Article

      researchmap

    • HEp-2 cell classification using K-support Spatial Pooling in Deep CNNs (ヘルスケア・医療情報通信技術)

      Han Xian-Hua, Chen Yen-Wei

      電子情報通信学会技術研究報告 = IEICE technical report : 信学技報116 ( 224 ) 1 - 6   16 9 2016

      ▶ More details

      Language：English   Publisher：電子情報通信学会  

      CiNii Article

      researchmap

    • A preliminary study on tensor codebook model for multiphase medical image retrieval (ヘルスケア・医療情報通信技術)

      WANG Jian, Han Xian-hua, Xu Yingying, Lin Lanfen, Hu Hongjie, Jin Chongwu, Chen Yen-Wei

      電子情報通信学会技術研究報告 = IEICE technical report : 信学技報116 ( 224 ) 47 - 50   16 9 2016

      ▶ More details

      Language：English   Publisher：電子情報通信学会  

      CiNii Article

      researchmap

    • A-2-3 Liver lesion retrieval based on multiphase medical volumes using sparse coding based codebook model

      WANG Jian, Han Xian-hua, Xu Yingying, Lin Lanfen, Hu Hongje, Jin Chongwu, Chen Yen-Wei

      Proceedings of the IEICE Engineering Sciences Society/NOLTA Society Conference2016   33 - 33   1 3 2016

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • D-12-16 Food Image Recognition with WLTP

      Fujita Kazuya, Sasano Syota, han Xian-Hua, Chen Yen-Wei

      Proceedings of the IEICE General Conference2016 ( 2 ) 85 - 85   1 3 2016

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • D-12-17 Food image recognition using Data-driven models

      Sasano Shota, Han Xian-Hua, Chen Yen-Wei

      Proceedings of the IEICE General Conference2016 ( 2 ) 86 - 86   1 3 2016

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • Multi-Volume Super Resolution for Mouse MR Images

        115 ( 459 ) 35 - 39   22 2 2016

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Deep Learning Based 3D Medical Volume Super Resolution

        40 ( 6 ) 29 - 34   2 2016

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Multi-Volume Super Resolution for Mouse MR Images

        40 ( 6 ) 35 - 39   2 2016

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Content-based retrieval of focal liver lesions using sparse representations of multiphase contrast-enhanced CT images (医用画像)

      WANG Jian, Han Xian-hua, Xu Yingying

      電子情報通信学会技術研究報告 = IEICE technical report : 信学技報115 ( 401 ) 171 - 176   19 1 2016

      ▶ More details

      Language：English   Publisher：電子情報通信学会  

      CiNii Article

      researchmap

    • Food Image Recognition using Deep Convolutional Neural Network

        115 ( 224 ) 67 - 72   14 9 2015

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Food Image Recognition using Sparse Coding and Hybrid Pooling methods

        26 ( 6 ) 1 - 7   6 2015

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Robust Point Correspondence Using ICP and Sparse Low Rank Decomposition

        115 ( 22 ) 23 - 28   14 5 2015

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Kernel Estimation using Normalized Sparsity Measure and its Application to Video Restoration

        115 ( 22 ) 77 - 82   14 5 2015

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Image Recognition based on Data-driven Model

        26 ( 5 ) 14 - 19   5 2015

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • D-11-34 Two-step Image Hallucination and its Application to 3D Medical Image Super-resolution

      Kondo Yuto, Han Xian-Hua, Chen Yen-Wei

      Proceedings of the IEICE General Conference2015 ( 2 ) 34 - 34   24 2 2015

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • D-12-19 3D facial shape analysis using local features

      Nakatsu Misae, Han Xian-Hua, Kimura Ryosuke, Chen Yen-Wei

      Proceedings of the IEICE General Conference2015 ( 2 ) 71 - 71   24 2 2015

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • D-16-9 Bayes Model for Liver Tumor Enhancement

      Konno Yu, Han Xian-Hua, Wei Xiong, Chen Yen-Wei

      Proceedings of the IEICE General Conference2015 ( 2 ) 149 - 149   24 2 2015

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • B-20-9 3D facial shape analysis using Partial Least Square regression

      Nakatsu Misae, Han Xian-Hua, Kimura Ryosuke, Chen Yen-Wei

      Proceedings of the Society Conference of IEICE2014 ( 1 ) 421 - 421   9 9 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • B-20-11 Computer Aided Diagnosis of Liver Cirrhosis using PLS Regression

      Uetani Mei, Tateyama Tomoko, Kohara Shinya, Han Xian-hua, Kanasaki Shuzo, Inoue Akitoshi, Furukawa Akira, Chen Yen-Wei

      Proceedings of the Society Conference of IEICE2014 ( 1 ) 423 - 423   9 9 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • Sparse and Low Rank Matrix Decomposition for Cirrhosis Diagnosis based Local Morphological Analysis

      Deng Junping, Han Xian-Hua, Chen Yen-Wei

      IEICE technical report.114 ( 200 ) 33 - 37   2 9 2014

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Cirrhosis liver is a terrible disease which is threatening our lives. Meanwhile, cirrhosis will cause significant hepatic morphological changes. While it is well known that the livers from different subjects have similar global shape structure which means liver shape ensemble should be low-rank. However the deformation which caused by cirrhosis can be considered as sparse compared with the whole liver. Therefore, in this study, we proposed to apply spare and low-rank matrix decomposition to partition the local deformation part (sparse error matrix E) from the global similar structure (low-rank matrix A) using the input liver shape D, which is the landmark coordinates of liver shapes and already have been aligned by the current rigid registration methods firstly. And then sparse matrix E is used for diagnosis. In common sense, the normal liver should have less local deformation than that of abnormal liver, which means that the norm of sparse matrix E for normal liver is smaller than the norm for abnormal one. Thus, we proposed a method which found a threshold classifier to classify normal and abnormal livers using the norm of E for these two categories. The proposed method is evaluated by a liver database and compared with statistical shape model(SSM) based methods. The experimental results of proposed method is better than those of SSM-based methods.

      CiNii Article

      researchmap

    • Stacked Fisher Network for HEp-2 Staining Pattern Recognition

      Han Xian-Hua, Chen Yen-Wei

      IEICE technical report.114 ( 200 ) 21 - 26   2 9 2014

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      This study addresses the recognition problem of the HEp-2 cell using indirect immunofluorescent(IIF) image analysis, which can indicate the presence of autoimmune diseases by finding antibodies in the patient serum. Generally, the method used for IIF analysis remains subjective, and depends too heavily on the experience and expertise of the physician. This study aims to explore an automatic HEp-2 cell recognition system, in which how to extract highly discriminate visual features plays a key role in this recognition application. In order to realize this purpose, our main efforts include: (1) a transformed excitation domain instead of the raw image domain, which is based on the fact that human perception for disguising a pattern depends not only on the absolute intensity of the stimulus but also on the relative variance of the stimulus; (2) a simple but robust micro-texton without any quantization in the excitation domain; (3) a data-driven coding strategy with a parametric probability process, and the extraction of not only low but also higher-order statistics for image representation called fisher vector; (4) the stacking of the fisher network into deep learning framework for more discriminate feature. Experiments using the open HEp-2 cell dataset used in the ICIP2013 contest validate that the proposed strategy can achieve a much better performance than the state-of-the-art approaches, and that the achieved recognition error rate is even very significantly below the observed intra-laboratory variability.

      CiNii Article

      researchmap

    • Improved Interactive Medical Image Segmentation using Graph Cut and Superpixels (Medical Imaging)

      Kitrungrotsakul Titinunt, Dong Chunhua, Han Xian-Hua, Chen Yen-Wei

      IEICE technical report.114 ( 103 ) 17 - 20   24 6 2014

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Interactive image segmentation is a useful method for selecting object of interest in image. The variations of intensity and shape in medical images (organs) limits their ability to precisely localize object boundaries, computation time of segmentation and therefore lack of accuracy in the segmentation object. The popular interactive segmentation method is Graph Cut. The computation time of each cut is a key to make interactive image segmentation useful in real application usage. The generally of medical images are larger than 2D image. The lack of computation time will be occur if we try to apply segment out the object in the medical images using only Graph Cut. This paper presents a method for combining Graph Cut with SLIC (Simple Linear Iterative Clustering) to adapt to medical image. To be precise, our method is initialized by design superpixels with SLIC super pixels. With SLIC superpixels, we can increasing the accuracy and also boost up computation time of Graph Cut. The experiments show segmentation results with our method is significantly better than only using Graph Cut in term of accuracy and computation time.

      CiNii Article

      researchmap

    • Organ Bounding Box Annotation based on Adaptive Selection of Bone References (Medical Imaging)

      DONG Chunhua, FORUZAN Amir H., HAN Xian-hua, TATEYAMA Tomoko, CHEN Yen-wei

      IEICE technical report.114 ( 103 ) 27 - 32   24 6 2014

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Accurate segmentation of abdominal organs is a key step in computer-aided diagnosis (CAD) system. To accurately segment a tissue, a volume of interest (VOI), in which the organ was located, was defined by a user. However, user interaction usually makes the task laborious and time-consuming. Hence, we propose a method that finds the VOI bounding box of the organ automatically based on the bone reference, which can be easily extracted from CT volumes. The basic idea of our proposed method is to achieve anatomical localization according to the statistical geometric location of organs within the bone reference. How to choose the bone as the reference was a difficult task, because the available abdominal volume have large variation during imaging procedure. With taking these into consideration, we prepared four different bone references. Using the adaptive selection of appropriate reference bone, the extracted bone from the input image is registered to the reference. After registration of all tissues of the training images, we find the VOI of the ensemble of tissues and use it as the organ bounding box. For the test images, according to the adaptive selection of bone reference, the candidate organ region is extracted based on this organ bounding box. We demonstrated the effectiveness of our method by finding the bounding box of kidney, liver and spleen.

      CiNii Article

      researchmap

    • Multi-touch Based Medical Interactive Visualization System

      Wang Jian, Tu Hua-Wei, Han Xian-Hua, Tateyama Tomoko, Chen Yen-Wei

      Technical report of IEICE. PRMU114 ( 41 ) 31 - 35   22 5 2014

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Medical imaging plays a central role in many healthcare practices. With the development of medical imaging devices, the medical image data can be achieved in higher and higher definitions, and then how to analyze and visualize the acquired large-amount and complex data is a desiring demand in medical education and clinical fields. Therefore, this paper develops a multi-touch based medical image analysis and visualization system for analyzing medical images, displaying the whole body's organs of a subject and the interested organ and region to explore more detailed structures. The designed system includes three modules: (1) Image analysis module, which will enable user segment a desired organ interactively; (2) Visualization module, which can manifest not only the detailed information and global structures of an organ in different view-points but also the correlated location relations between different organs; (3) Control module, which can enable the user to easily interact with the system by multi-touch based control. The designed system is able to achieve a detail visualization of the integrated components, and can be applied to education, doctor-training and clinical sites.

      CiNii Article

      researchmap

    • Multi-touch Based Medical Interactive Visualization System

      Wang Jian, Tu Hua-Wei, Han Xian-Hua, Tateyama Tomoko, Chen Yen-Wei

      IEICE technical report. Signal processing114 ( 39 ) 31 - 35   22 5 2014

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Medical imaging plays a central role in many healthcare practices. With the development of medical imaging devices, the medical image data can be achieved in higher and higher definitions, and then how to analyze and visualize the acquired large-amount and complex data is a desiring demand in medical education and clinical fields. Therefore, this paper develops a multi-touch based medical image analysis and visualization system for analyzing medical images, displaying the whole body's organs of a subject and the interested organ and region to explore more detailed structures. The designed system includes three modules: (1) Image analysis module, which will enable user segment a desired organ interactively; (2) Visualization module, which can manifest not only the detailed information and global structures of an organ in different view-points but also the correlated location relations between different organs; (3) Control module, which can enable the user to easily interact with the system by multi-touch based control. The designed system is able to achieve a detail visualization of the integrated components, and can be applied to education, doctor-training and clinical sites.

      CiNii Article

      researchmap

    • Hybrid Aggregation of Sparse Coded Discriptors for Food Image Recognition

      KUSUMOTO Riko, HAN Xian-Hua, CHEN Yen-Wei

      IEICE technical report.114 ( 42 ) 1 - 5   22 5 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In Recent year, with the increasing of unhealthy diets which will threaten people's life due to the various resulted risks such as heart stroke, liver trouble and so on, the maintaining for healthy life has attracted much attention and then how to manage the dietary life is becoming more and more important. In this research, we aim to construct an auto-recognition system of food images and keep the daily food-log records which will contribute to manage dietary life. In order to achieve the acceptable recognition performance of the food images, we propose to apply a sparse model, instead of vector quantization in the widely used Bag-of-Features (BOF) model, to code local descriptors extracted from the food images for reducing information loss, and use the spatial pyramid matching for integrating spatial information into the image representation vector. Furthermore, a hybrid aggregation strategy called as top-ranked average pooling (TRAP), instead of average or max pooling, is explored for constructing the compact vector for image representation. Experiments validate that our proposed framework can significantly improve the recognition performance compared to the conventional BOF on two food datasets.

      CiNii Article

      researchmap

    • 2-step Image Hallucination and its Application 3D Medical Image Super-resolution

      Kondo Yuto, Nojima Yusuke, Han Xian-Hua, Chen Yen-Wei

      IEICE technical report.114 ( 42 ) 61 - 66   22 5 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In medical diagnosis, high resolution (HR) images are indispensable for giving more correct decision. However, in order to obtain high resolution MR images from MR, It is necessary to impose long-time and high-dosage radiation exposure to patient, and then leads to heavy burden to the patient. Therefore Super Resolution technique, which can generate high resolution images from low resolution images using machine learning techniques, attracts hot attention recently. Therein, image hallucination is one of widely used super-resolution methods in image restoration field. However, the conventional image hallucination generally cannot recover high frequency information. Therefore, this paper integrates a further learning step into the conventional method, and proposes a 2-step image hallucination, which is prospected to recover most high frequency information lost in the available low-resolution input. Furthermore, we apply the proposed strategy to generate the high-resolution Z-direction data using self-similarity among different direction for 3D medical MR images. Experimental results show that the proposed strategy can reconstruct promising HR coronal or sagittal plane by using available LR and HR data pairs in axial plane.

      CiNii Article

      researchmap

    • 3D facial shape analysis based on statistical learning methods

      NAKATSU Misae, HAN Xian-Hua, KIMURA Ryosuke, CHEN Yen-Wei

      IEICE technical report.114 ( 42 ) 67 - 72   22 5 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Recently, the relationship between gene and the facial morphology attracts substantial attention. A generic framework for analyzing facial morphology variation using scanned 3D landmarks was proposed in our previous work, which mainly includes three steps: registration and landmark correspondence, statistical analysis of 3D facial shape, ancestral classification using facial features. However, the used landmark corresponding method in the previous work was based on non-rigid transformation, which is very complicated and has high computational-cost, and the classification accuracy was less than 70%, which still has large space to be improved. This study proposes a simple and fast landmark corresponding strategy using cylindrical transformation; it easily adjusts the dimension of the shape representation vector. Furthermore, beside the shape variation features extracted by Principle Component Analysis (PCA), we propose a novel discriminated feature using mean hyperplane, and normalize the feature vector for reducing scale affect. Experiments on ancestral classification show that our proposed strategy can significantly improve the recognition performances compared to the conventional work.

      CiNii Article

      researchmap

    • Non-parametric Probability Model for Liver Tumor Enhancement

      Konno Yu, Han Xian-Hua, Chn Yen-Wei

      IEICE technical report.114 ( 42 ) 73 - 78   22 5 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Automatic tumor enhancement and detection has an essential role for the computer-aided diagnosis of liver tumor in CT volume data. This paper proposes a novel tumor enhancement strategy by exploring the existing probability of tumor for any voxel. However, the tumor prototypes in a test liver volume from a specific patient or common tumor prototypes are extremely difficult to achieve due to requirement of full-searching and large variation of tumor tissues in different liver volumes. Therefore, this research investigates a tumor-training-data free strategy by only constructing the healthy liver and vessel prototypes, which can be extracted from any slice of a liver volume, and then applies a non-parametric framework for calculating the existing probability of liver or vessel. Finally, the existing probability of tumor can be deduced from that of liver or vessel.

      CiNii Article

      researchmap

    • Generalized Super-Vector coding for image classification

      Nakajima Motoki, Han Xian-Hua, Chen Yen-Wei

      IEICE technical report.114 ( 42 ) 113 - 117   22 5 2014

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Semantic understanding of images remains an important research challenge in machine intelligence and statistical learning. This study mainly aims to explore a generalized feature extraction framework motivated by the popularly used Bag-of-feature (BOF) and super-vector coding using local descriptor (such as SIFT), which is intuitively time-consuming for computation. Compared to the uniformly quantized strategy such as the conventional color histogram, the proposed framework can represent the image more faithful and compact, and then lead to more discriminant representation for images. With the extracted adaptive statistics, a simple linear support vector machine (SVM), which is especially efficient for large-scale database, can be effectively utilized for achieving acceptable recognition performances. Our proposed strategy can achieve much better recognition performances than the conventional and the state-of-the art methods.

      CiNii Article

      researchmap

    • 遺伝子関連研究のための顔面形態解析 第2報-非剛体位置合わせ不要な対応付け法-

      中津美冴, 韓先花, 瀬尾昌孝, 健山智子, 木村亮介, 陳延偉

      電気関係学会関西連合大会講演論文集(CD-ROM)2013   2013

      ▶ More details

    • Sparse Dictionary Representation and Propagation for MRI Volume Super-Resolution

      HAN Xian-Hua, CHEN Yen-Wei

      Technical report of IEICE. PRMU112 ( 37 ) 13 - 18   10 5 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      This study addresses the problem of generating a high-resolution (HR) MRI volume from a single low-resolution (LR) MRI input volume. Recent researches have proved that sparse coding can be successfully applied for single-frame super-resolution for natural images, which is based on good reconstruction of any local image patch with a sparse linear combination of atoms taken from an appropriate over-complete dictionary. This study adapts the basic idea of sparse code-based super-resolution (SCSR) for MRI volume data, and then improves the dictionary learning strategy in the conventional SCSR for achieving the precise sparse representation of HR volume patches. In the proposed MRI super-resolution strategy, we only learn the dictionary of the HR MRI volume patches with sparse coding algorithm, and then propagate the HR dictionary to the LR dictionary by mathematical analysis for calculating the sparse representation (coefficients) of any LR local input volume patch. The unknown corresponding HR volume patch can be reconstructed with the sparse coefficients from the LR volume patch and the corresponding HR dictionary. We validate that the proposed SCSR strategy through dictionary propagation can recover much clearer and more accurate HR MRI volume than the conventional interpolated methods.

      CiNii Article

      researchmap

    • Generalized N-Dimensional Independent Component Analysis Based Multiple Feature Selection and Fusion

      AI Danni, DUAN Guifang, HAN Xianhua, CHEN Yen-Wei

      Technical report of IEICE. PRMU112 ( 37 ) 81 - 86   10 5 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      We proposed a multilinear independent component analysis framework called generalized N-dimensional ICA (GND-ICA) by extending the conventional linear ICA based on multilinear algebra. Unlike the linear ICA that only treats one-dimensional data, the proposed GND-ICA treats N-dimensional data as a tensor without any preprocess of data vectorization. We furthermore introduce two types of GND-ICA solutions and analysis their efficiency and effectiveness. As an application, the GND-ICA can be used for multiple feature fusion and representation for color image classification. Many features extracted from a given image are constructed as a tensor. The feature tensor can be effective represented by GND-ICA. Compared with conventional linear subspace learning methods, GND-ICA is capable of obtaining more distinctive representation for color image classification.

      CiNii Article

      researchmap

    • A Novel Components Selection Method in Statistical Shape Model for Computer-aided Diagnosis : And it's Application for Assisting Cirrhosis Diagnosis

      LUO Jie, HAN Xian-Hua, KOHARA Shinya, TATEYAMA Tomoko, FURUKAWA Akira, KANAZAKI Shuzo, CHEN Yen-Wei

      Technical report of IEICE. PRMU112 ( 37 ) 33 - 38   10 5 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In the field of Computer-aided Diagnosis, organs' shape variability have been found to be important for understanding their inherent structures. It's also well known that Statistical Shape Model (SSM) can effectively capture the shape variation of given training set. Consequently, diseases that cause organ deformations have the possibility to be diagnosed with the aid of computer by SSM based organ shape analysis. However, it's difficult to decide which modes are selected as indicators to assist the diagnosis. Hence, applications of applying SSM for assisting diagnosis haven't widely concerned. In this study we propose a novel modes selection method and test it in case of assisting Cirrhosis diagnosis. Experimental results validate the effectiveness of our method and its potential for assisting the clinic diagnosis.

      CiNii Article

      researchmap

    • A Novel Components Selection Method in Statistical Shape Model for Computer-aided Diagnosis : And it's Application for Assisting Cirrhosis Diagnosis

      LUO Jie, HAN Xian-Hua, KOHARA Shinya, TATEYAMA Tomoko, FURUKAWA Akira, KANAZAKI Shuzo, CHEN Yen-Wei

      IEICE technical report. Image engineering112 ( 36 ) 33 - 38   10 5 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In the field of Computer-aided Diagnosis, organs' shape variability have been found to be important for understanding their inherent structures. It's also well known that Statistical Shape Model (SSM) can effectively capture the shape variation of given training set. Consequently, diseases that cause organ deformations have the possibility to be diagnosed with the aid of computer by SSM based organ shape analysis. However, it's difficult to decide which modes are selected as indicators to assist the diagnosis. Hence, applications of applying SSM for assisting diagnosis haven't widely concerned. In this study we propose a novel modes selection method and test it in case of assisting Cirrhosis diagnosis. Experimental results validate the effectiveness of our method and its potential for assisting the clinic diagnosis.

      CiNii Article

      researchmap

    • Generalized N-Dimensional Independent Component Analysis Based Multiple Feature Selection and Fusion

      AI Danni, DUAN Guifang, HAN Xianhua, CHEN Yen-Wei

      IEICE technical report. Image engineering112 ( 36 ) 81 - 86   10 5 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      We proposed a multilinear independent component analysis framework called generalized N-dimensional ICA (GND-ICA) by extending the conventional linear ICA based on multilinear algebra. Unlike the linear ICA that only treats one-dimensional data, the proposed GND-ICA treats N-dimensional data as a tensor without any preprocess of data vectorization. We furthermore introduce two types of GND-ICA solutions and analysis their efficiency and effectiveness. As an application, the GND-ICA can be used for multiple feature fusion and representation for color image classification. Many features extracted from a given image are constructed as a tensor. The feature tensor can be effective represented by GND-ICA. Compared with conventional linear subspace learning methods, GND-ICA is capable of obtaining more distinctive representation for color image classification.

      CiNii Article

      researchmap

    • Sparse Dictionary Representation and Propagation for MRI Volume Super-Resolution

      HAN Xian-Hua, CHEN Yen-Wei

      IEICE technical report. Image engineering112 ( 36 ) 13 - 18   10 5 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      This study addresses the problem of generating a high-resolution (HR) MRI volume from a single low-resolution (LR) MRI input volume. Recent researches have proved that sparse coding can be successfully applied for single-frame super-resolution for natural images, which is based on good reconstruction of any local image patch with a sparse linear combination of atoms taken from an appropriate over-complete dictionary. This study adapts the basic idea of sparse code-based super-resolution (SCSR) for MRI volume data, and then improves the dictionary learning strategy in the conventional SCSR for achieving the precise sparse representation of HR volume patches. In the proposed MRI super-resolution strategy, we only learn the dictionary of the HR MRI volume patches with sparse coding algorithm, and then propagate the HR dictionary to the LR dictionary by mathematical analysis for calculating the sparse representation (coefficients) of any LR local input volume patch. The unknown corresponding HR volume patch can be reconstructed with the sparse coefficients from the LR volume patch and the corresponding HR dictionary. We validate that the proposed SCSR strategy through dictionary propagation can recover much clearer and more accurate HR MRI volume than the conventional interpolated methods.

      CiNii Article

      researchmap

    • Image Super-Resolution using Manifold Learning with Vector Quantization

      TANIGUCHI Kazuki, HAN Xian-Hua, IWAMOTO Yutaro, SASATANI So, CHEN Yen-Wei

      IEICE technical report.112 ( 38 ) 19 - 24   10 5 2012

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Image Super-Resolution (SR) is to recover the lost high-frequency information from several or only one available image. Single-Frame SR, one of hot topics in SR research fields, can generate a high-resolution image from only one low-resolution image by using the prior prepared database. Therein, the example-based and neighborhood embedding-based SR are the very popular single-frame SRs to infer the lost information in the LR input with the known corresponding relations between LR and HR images in database which has to be prepared in large-scale for having most varieties of image, and then take a lot of computational time for inferring. Therefore, this study proposes to first obtain some prototypes from the prepared LR and HR images using vector quantization such as k-means clustering method, and the achieved prototypes are as the training database for inferring the lost information of any LR input. Then, the amount of corresponding LR and HR data in training database can be greatly reduced, which guarantee much less computational time. Experimental results also show that our proposed strategy can achieve higher quality high-resolution image and lower computational time than conventional methods.

      CiNii Article

      researchmap

    • Performance Comparison of SPHARM Based Statistical Shape Model and Point Distribution Model

      TATEYAMA Tomoko, OKEGAWA Megumi, KOHARA Shinya, HAN XianHua, FURUKAWA Akira, KANASAKI Shuzo, CHEN Yen-wei

      IEICE technical report.112 ( 38 ) 39 - 44   10 5 2012

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In medical imaging research, the three-dimensional (3-D) shape representation and analysis of anatomic structures using only a few parameters is an important issues, and can be applied to computer assisted diagnosis, surgical simulations, visualization, and many other medical applications. This paper proposes the representation of the shape surfaces of a simply connected 3-D object using spherical harmonic (spharm) functions, which can provide an approximate global description of the shape using only a few spharm parameters. Spharms are obtained from several partial differential equations used in physics, such as the Laplace, Helmholtz, and Schrodinger equations, in which spherical coordinates are used. In this study, we aim to develop a 3D representation of human anatomy, such as the liver and spleen, with fewer parameters by using spharm functions. Furthermore, the experimental results show that the shape of the liver and spleen can be accurately represented with only a few spharm parameters.

      CiNii Article

      researchmap

    • D-12-22 Visual Impression Estimation of Clothing Fabric Images Using Machine Learning Methods

      Chen Dingye, Han Xianhua, Huang Yang, Huang Xinyin, Chen Yen-wei

      Proceedings of the IEICE General Conference2012 ( 2 ) 116 - 116   6 3 2012

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • 球面調和関数を用いた人体肝臓3次元形状表現と統計形状モデル構築

      健山智子, 上谷芽衣, 田中英俊, 桶川萌, 小原伸哉, 韓先花, 金崎周造, 佐藤滋高, 古川顕, 陳延偉

      電気関係学会関西連合大会講演論文集(CD-ROM)2012   2012

      ▶ More details

    • 球面調和関数を用いた人体脾臓の3次元形状モデリングと統計形状モデルの構築

      健山智子, 小原伸哉, 田中泰史, 桶川萌, 古川顕, 金崎周造, 若宮誠, 韓先花, 陳延偉

      画像の認識・理解シンポジウム(MIRU2011)論文集2011 ( 2011 ) 255 - 262   20 7 2011

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • Independent Components Selection of Color SIFT Descriptors for Image Classification

      AI Dan-ni, HAN Xianhua, DUAN Guifang, RUAN Xiang, CHEN Yen-Wei

      IEICE technical report111 ( 48 ) 25 - 30   12 5 2011

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      This paper addresses the problem of ordering the color SIFT descriptors in the independent component analysis for image classification. Component ordering is of great importance for image classification, since it is the foundation of the feature selection. To select the distinctive and compact independent components of the color SIFT descriptors, we propose two ordering approaches based on local variation, named as the localization-based ICs ordering and the sparseness-based ICs ordering. We evaluate the performance of proposed methods, the conventional ICs selection method (global variation based components selection) and original color SIFT descriptors on object and scene databases, and obtain the following two main results. First, the proposed methods are able to obtain the acceptable classification results in comparison with original color SIFT descriptors. Second, the highest classification rate can be obtained by using the global selection method in the scene database, while the local ordering methods give the best performance for the object database.

      CiNii Article

      researchmap

    • Joint Kernel Equal Integration of Visual Features and Textual Terms for Biomedical Imaging Modality Classification

      HAN Xian-Hua, CHEN Yen-Wei

      IEICE technical report111 ( 47 ) 51 - 56   12 5 2011

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In this paper, we describe an approach for the automatic modality classification in medical image retrieval task of the 2010 CLEF cross-language image retrieval campaign (ImageCLEF). This work is focused on the process of feature extraction from medical images and fusion the different extracted visual feature and textual feature for modality classification. To extract visual features from the images, we used histogram descriptor of edge, gray or color intensity and block-based variation as global features and SIFT histogram as local feature, and the binary histogram of some predefined vocabulary words for image captions is used for textual feature. Then we combine the different features using normalized kernel functions for SVM classification. The proposed algorithm is evaluated by the provided modality dataset by ImageCLEF2010.

      CiNii Article

      researchmap

    • Independent Components Selection of Color SIFT Descriptors for Image Classification

      AI Dan-ni, HAN Xianhua, DUAN Guifang, RUAN Xiang, CHEN Yen-Wei

      IEICE technical report111 ( 47 ) 25 - 30   12 5 2011

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      This paper addresses the problem of ordering the color SIFT descriptors in the independent component analysis for image classification. Component ordering is of great importance for image classification, since it is the foundation of the feature selection. To select the distinctive and compact independent components of the color SIFT descriptors, we propose two ordering approaches based on local variation, named as the localization-based ICs ordering and the sparseness-based ICs ordering. We evaluate the performance of proposed methods, the conventional ICs selection method (global variation based components selection) and original color SIFT descriptors on object and scene databases, and obtain the following two main results. First, the proposed methods are able to obtain the acceptable classification results in comparison with original color SIFT descriptors. Second, the highest classification rate can be obtained by using the global selection method in the scene database, while the local ordering methods give the best performance for the object database.

      CiNii Article

      researchmap

    • Super-Resolution of Medical Images Based on Hallucination Methods

      IWAMOTO Yutaro, HAN Xian-Hua, OHASHI Motonori, SASATANI So, TANIGUCHI Kazuki, CHEN Yen-Wei

      IEICE technical report111 ( 47 ) 57 - 62   12 5 2011

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In Medical imaging, the data resolution is usually insufficient for accurate diagnosis in clinical medicine. Especially in most case, the resolution in the slice direction (Z direction) is much lower than that of the in-plane resolution (XY direction). Therefore it is difficult to construct isotropic voxels, which is very important in 3-D visualization systems, such as surgical system. In this paper, we propose a method for improving resolution in the slice direction for medical volume images based on Hallucination method, which is one of the learning based super-resolution techniques. The experimental results verify the effectiveness of the proposed method, and we compared with the conventional interpolation methods.

      CiNii Article

      researchmap

    • Auto Recognition of food images using SPIN feature and Food-Log system

      WAZUMI Minami, TSURUGAI Youhei, HAN Xian-Hua, CHEN Yen-Wei

      IEICE technical report111 ( 47 ) 93 - 98   12 5 2011

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Recently, with the increasing of unhealthy diets and the attracted attention for healthy life, how to manage the dietary life is becoming more and more important. In this paper, we aim to construct a system, which can auto-recognize the menu contents from food image taken by mobile phone. As we know that the viewpoints can be varied in any direction when taking food images, and then, rotation-robust features for image representation are very important. Therefore, in this paper, we propose to extract rotation invariant features using circle-segmentation called SPIN for food recognition, and construct a Food-Log system, which records the contents of food menu, calories and nutritional value for management of the dietary life.

      CiNii Article

      researchmap

    • High Frequency Compensated Face Hallucination Method

        2010 ( 5 ) 6p   2 2011

      ▶ More details

      Language：Japanese  

      CiNii Article

      researchmap

    • High Frequency Compensated Face Hallucination Method

      SASATANI So, HAN Xian-Hua, OHASHI Motonori, IWAMOTO Yutaro, CHEN Yen-Wei

      IEICE technical report110 ( 382 ) 323 - 328   13 1 2011

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Face Hallucination method is one of learning-based super-resolution techniques, which can reconstruct a high-resolution image from only a single low-resolution image based on machine learning methods. However, it is difficult to reconstruct the detailed high-frequency components. In this paper, we proposed a high frequency compensated face hallucination method. The proposed method is a three-step method: (1) reconstruct a high-resolution image by using conventional face hallucination method; (2) calculate a low-resolution residual image between the input low-resolution image and the downsampled reconstructed high-resolution image and then reconstruct a high-resolution residual image by using the conventional face hallucination method; (3)compensate the high-resolution residual image to the reconstructed high-resolution image in step 1.

      CiNii Article

      researchmap

    • Image Super-Resolution using RIPOC-based registration

      Taniguchi Kazuki, Han Xian-Hua, Iwamoto Yutaro, Sasatani So, Ohashi Motonori, Chen Yen-Wei

      ITE Technical Report35 ( 0 ) 33 - 36   2011

      ▶ More details

      Language：Japanese   Publisher：The Institute of Image Information and Television Engineers  

      Multi-Frame Super-Resolution is a technique which attempts to reconstruct high-resolution image by fusing a number of low-resolution images. Therein, high-accuracy registration for the input low-resolution image set is required for reconstruction a clear high-resolution image. The conventional registration approaches for Multi-Frame Super-Resolution such as Iterative Back Projection and LK optical flow methods, can obtain accurate parameters for tiny translation or rotation movement between images. However, for large movement especially in rotation, there are still no efficient methods for reconstructing high-resolution image. Therefore, in this paper we proposed a new method using Rotation Invariant Phase Only Correlation, which can estimate large rotation parameter, and then follow the Iterative back Projection method for reconstructing high-resolution image. Experimental results show that the reconstructed high resolution images by our proposed approach are much better than those by interpolation techniques.

      DOI： 10.11485/itetr.35.23.0_33

      CiNii Article

      researchmap

    • Scene Image Recognition with Multi Level Resolution Semantic Modeling

      TANAKA Yoshiyuki, OKAMOTO Atsushi, HAN Xian-Hua, RUAN Xiang, CHEN Yen-Wei

      IEICE technical report110 ( 28 ) 169 - 174   6 5 2010

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In this paper, we propose a multi-level resolution semantic modeling for automatic scene recognition. The basic idea of the semantic modeling is to classify local image regions into semantic concept classes such as water, sunset, or sky, and use occurrence frequency of local region's semantic concepts for global image representation. However, how to decide size of the local image regions is a trial problem. The optimized region size would be dynamically changing for different scene or concept types. Therefore, this paper proposed a dynamical region size(Multi-level resolution)of local image regions for semantic concept model, and fusion the probabilities to scene types of several resolutions for final recognition of a scene image. Experimental results show that the recognition rate using our proposed algorithm is much better than that using the conventional semantic modeling method for scene recognition.

      CiNii Article

      researchmap

    • Hierarchical Classifier with Multiple Feature Weighted Fusion for Scene Recognition

      KIKUTANI Yae, OKAMOTO Atsushi, HAN Xian-Hua, RUAN Xiang, CHEN Yen-Wei

      IEICE technical report110 ( 28 ) 175 - 179   6 5 2010

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Recently, scene recognition is becoming an additional functional in digital camera. Automatic scene understanding is a highest-level operation in computer vision, and it is a very difficult and largely unsolved problem. The conventional methods usually use global features(such as color histogram, texture, edge)for image representation and recognize scene types with some classifiers(such as Bayesian, Neural Network, Support Vector Machine and so on). However, the recognition rate still cannot satisfy the requirement of real applications. In this paper, we proposed to use weighted fusion of global feature(Color histogram)and local feature(Bag-Of-Feature model)for scene image representation, and use hierarchical classifier according the visual feature properties of scene types for scene recognition. Experimental results show that the recognition rate with our proposed algorithm can be improved compared to the state of art algorithms.

      CiNii Article

      researchmap

    • Image Registration Using both Principal Component Analysis and Gradient Method for Super-resolution

      SASATANI So, HAN Xian-Hua, CHEN Yen-Wei

      IEICE technical report110 ( 28 ) 193 - 198   6 5 2010

      ▶ More details

      Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Super-Resolution is the technique enhances resolution of image, and one of its techniques is multi-frame method is an approach to reconstruct a high-resolution(HR)image from some of low-resolution images. There are many image registration techniques used in this method, but they have some fault each other and it is possible to improve the HR image quality by resolving it. In this paper, we proposed image registration technique using usual methods and show an experimental result.

      CiNii Article

      researchmap

    • An adaptive color SIFT descriptor based on independent component analysis for image classification

      AI Danni, HAN Xianhua, RUAN Xiang, CHEN Yen-Wei

      IEICE technical report109 ( 306 ) 1 - 6   19 11 2009

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      In this paper, we present a novel color independent components based SIFT descriptor (termed CIC-SIFT) for object/scene classification. We first learn an efficient color transformation matrix based on independent component analysis (ICA), which is adaptive to each category in a database. The ICA-based color transformation can enhance contrast between the objects and the background in an image. Then we compute CIC-SIFT descriptors over all three transformed color independent components. Since the ICA-based color transformation can boost the objects and suppress the background, the proposed CIC-SIFT can extract more effective and discriminative local features for object/scene classification. The comparison is performed among seven SIFT descriptors and the experimental classification results show that our proposed CIC-SIFT is superior to other conventional SIFT descriptors.

      CiNii Article

      researchmap

    • Multi-class Object Recognition by fusion of image descriptors : Classification evaluation of PASCAL VOC Challenge database

      HAN Xian-Hua, CHEN Yen-Wei, RUAN Xiang

      IEICE technical report109 ( 306 ) 103 - 108   19 11 2009

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Category recognition is important to access visual information on the level of objects. Over the past several years, substantial performance gains on challenging benchmark datasets have been reported in the literature. This progress can be attributed to two developments: the design of highly discriminative and robust image features and the combination of multiple complementary features based on different aspects such as shape, appearance. In this paper we extract different appearance descriptors for several properties of local keypionts obtained by grid sampling, Harris and Difference of Gaussian (DoG), and a shape descriptor-Pyramid Histogram of Orientation Gradient (PHOG). The extracted descriptors are evaluated by PASCAL VOC database with nonlinear SVM classifiers, and they also are combined for evaluation. The experimental results show that the accuracy rate with fusion of different image descriptors can be improved.

      CiNii Article

      researchmap

    • A supervised LPP and Neural Network based Scene classification with Color Histogram and Camera Metadata

      HAN Xian-Hua, CHEN Yen-Wei, FUJITA Hideto, KOYAMA Kan-Ichi, TAKAYANAKI Wataru

      Technical report of IEICE. PRMU108 ( 484 ) 257 - 262   6 3 2009

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Scene classification (e.g., landscape, sunset, night-landscape, etc.) is still a challenging problem in computer vision. Scene classification based only on low-level vision cues has had limited success on unconstrained image sets. In other hand, camera metadata related to capture conditions provides cues independent of the captured scene content that can be used to improve classification performance. Analysis of camera metadata statistics for images of each class revealed that some metadata fields are most discriminative for some classes. So, in this paper, we proposed to use the combined feature of scene color histogram and camera metadata, and then using supervised Locality preserving projection (LPP) for feature space transformation and dimension reduction, and finally, adapt Probabilistic neural network for scene classification. Experimental results show that the classification accuracy rate can be improved compared with using PCA (Principal Component Analysis) subspace learning method, and are also better than that with only the low-level vision feature (color histogram).

      CiNii Article

      researchmap

    • A supervised LPP and Neural Network based Scene classification with Color Histogram and Camera Metadata

      HAN Xian-Hua, CHEN Yen-Wei, FUJITA Hideto, KOYAMA Kan-Ichi, TAKAYANAKI Wataru

      IPSJ SIG Notes. CVIM2009 ( 29 ) 257 - 262   6 3 2009

      ▶ More details

      Language：Japanese   Publisher：Information Processing Society of Japan (IPSJ)  

      Scene classification (e.g., landscape, sunset, night-landscape, etc.) is still a challenging problem in computer vision. Scene classification based only on low-level vision cues has had limited success on unconstrained image sets. In other hand, camera metadata related to capture conditions provides cues independent of the captured scene content that can be used to improve classification performance. Analysis of camera metadata statistics for images of each class revealed that some metadata fields are most discriminative for some classes. So, in this paper, we proposed to use the combined feature of scene color histogram and camera metadata, and then using supervised Locality preserving projection (LPP) for feature space transformation and dimension reduction, and finally, adapt Probabilistic neural network for scene classification. Experimental results show that the classification accuracy rate can be improved compared with using PCA (Principal Component Analysis) subspace learning method, and are also better than that with only the low-level vision feature (color histogram).

      CiNii Article

      researchmap

      Other Link： http://id.nii.ac.jp/1001/00061527/

    • D-12-36 A Spatial Weighted Edge Autocorrelogram for Image Annotation

      Ai Danni, Duan Guifang, Han Xianhua, Ruan Xiang, Chen Yen-Wei

      Proceedings of the IEICE General Conference2009 ( 2 ) 145 - 145   4 3 2009

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • D-12-18 PCA-CSIFTによる自動Annotation付与(D-12.パターン認識・メディア理解A(パターンメディアの認識・理解・生成),一般セッション)

      岡本 充史, 韓 先花, 阮 翔, 陳 延偉

      電子情報通信学会総合大会講演論文集2009 ( 2 ) 127 - 127   4 3 2009

      ▶ More details

      Language：Japanese   Publisher：一般社団法人電子情報通信学会  

      CiNii Article

      researchmap

    • D-12-17 高倍率画像を生成するための階層的Super Resolution(D-12.パターン認識・メディア理解A(パターンメディアの認識・理解・生成),一般セッション)

      大橋 基範, 韓 先花, 陳 延偉

      電子情報通信学会総合大会講演論文集2009 ( 2 ) 126 - 126   4 3 2009

      ▶ More details

      Language：Japanese   Publisher：一般社団法人電子情報通信学会  

      CiNii Article

      researchmap

    • Robust face recognition based on modified architecture of Independent Component Analysis

      HAN Xian-Hua, CHEN Yen-Wei, YAMADA Akihiko, FUJITA Hideto

      IPSJ SIG Notes. CVIM164 ( 82 ) 79 - 84   29 8 2008

      ▶ More details

      Language：English   Publisher：Information Processing Society of Japan (IPSJ)  

      A number of current face recognition algorithms use face representations found by statistical subspace learning methods. Therein, Principal Component Analysis (PCA), Independent Component Analysis (ICA) and Linear Discriminant Analysis (LDA) are a high-focused research topic in this field. LDA can get good results for supervised face recognition. However, it is limited for some face recognition system, where no or only one sample face can be obtained for training. In the other hand, ICA can obtain acceptable recognition result. However, the applicability of ICA to high-dimension pattern recognition tasks such as face recognition often suffers some problems. The most important one is real-time problem, another is the computer memory. The mentioned two problems make ICA classifier unsuitable and inapplicable in real system. In this paper, we propose a modified ICA architecture to deal with the two problems. Firstly, we use eigenface method to calculate the eigenvector and eigenvalue for the training samples, and whitening the face images. Finally, the independent coefficients of image factorial code can be obtained by ICA analysis. Experimental results show that the proposed method can not only obtain high-speed but also get acceptable accuracy rate.

      CiNii Article

      researchmap

    • Robust face recognition based on modified architecture of Independent Component Analysis

      HAN Xian-Hua, CHEN Yen-Wei, YAMADA Akihiko, FUJITA Hideto

      IEICE technical report108 ( 199 ) 79 - 84   29 8 2008

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      A number of current face recognition algorithms use face representations found by statistical subspace learning methods. Therein, Principal Component Analysis (PCA), Independent Component Analysis (ICA) and Linear Discriminant Analysis (LDA) are a high-focused research topic in this field. LDA can get good results for supervised face recognition. However, it is limited for some face recognition system, where no or only one sample face can be obtained for training. In the other hand, ICA can obtain acceptable recognition result. However, the applicability of ICA to high-dimension pattern recognition tasks such as face recognition often suffers some problems. The most important one is real-time problem, another is the computer memory. The mentioned two problems make ICA classifier unsuitable and inapplicable in real system. In this paper, we propose a modified ICA architecture to deal with the two problems. Firstly, we use eigenface method to calculate the eigenvector and eigenvalue for the training samples, and whitening the face images. Finally, the independent coefficients of image factorial code can be obtained by ICA analysis. Experimental results show that the proposed method can not only obtain high-speed but also get acceptable accuracy rate.

      CiNii Article

      researchmap

    • D-16-2 ICA Based Ring Artifacts Reduction in Cone-Beam CT Images

      DUAN Guifang, CHEN Yen-wei, HAN Xianhua, FUJITA Akinori, HIROOKA Ken, UENO Yoshihiro

      Proceedings of the IEICE General Conference2008 ( 2 ) 269 - 269   5 3 2008

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      CiNii Article

      researchmap

    • Reduction of Ring Artifacts in Cone-Beam CT images

      DUAN Guifang, CHEN Yen-wei, HAN Xianhua, FUJITA Akinori, HIROOKA Ken, UENO Yoshihiro

      IEICE technical report107 ( 461 ) 227 - 233   25 1 2008

      ▶ More details

      Language：English   Publisher：The Institute of Electronics, Information and Communication Engineers  

      Cone-beam CT (CBCT) scanners are based on volumetric tomography, using a 2D extended digital array providing an area detector [1]. Compared to traditional CT, CBCT has many advantages, such as less X-ray beam limitation, high image accuracy, Rapid scan time, etc. However, In CBCT images there are always some ring artifacts that appear as rings centred on the rotation axis. Due to the data of the constructed images are corrupted by these ring artifacts, qualitative and quantitative analysis of CBCT images will be compromised. Post processing and application such as image segmentation and registration also turn more complex as the presence of such artifacts. In this paper, a method with a novel region of interest (ROI) building technique is presented. It deals with the reconstructed CBCT image and can effectively reduce such ring artifacts.

      CiNii Article

      researchmap

    • 高速点火核融合実験におけるX線画像計測-ICA Shrikageフィルタによるポアソンノイズの除去-

      陳延偉, 韓先花, 健山智子

      大阪大学レーザーエネルギー学研究センター共同研究成果報告書2005   2006

      ▶ More details

    ▼display all

    Books and Other Publications

    • Residual Sparse Autoencoders for Unsupervised Feature Learning and Its Application to HEp-2 Cell Staining Pattern Recognition

      Xian-Hua Han, Yen-Wei Chen（ Role： Contributor ,  Chapter 11）

      Deep Learning in Healthcare, Chapter 11, Yen-Wei Chen and Jain, Lakhmi C. (Eds.), [Online First], Springer, 2020. DOI: 10.1007/978-3-030-32606-7_11  2020 

      ▶ More details

    • Hyperspectral Image Super-Resolution Using Optimization and DCNN-Based Methods

      Xian-Hua Han（ Role： Contributor ,  One chapter）

      Processing and Analysis of Hyperspectral Data, Jie Chen (Eds.), [Online First], IntechOpen, DOI: 10.5772/intechopen.89243.  2020 

      ▶ More details

    • High-Order Statistics of Micro-Texton for HEp-2 Staining Pattern Classification in 「Artificial Intelligence in Decision Support Systems for Diagnosis in Medical Imaging」

      Xian-Hua Han, Yen-Wei Chen（ Role： Contributor ,  Chapter 4）

      Springer 出版  1 2018 

      ▶ More details

    • Subspace Methods for Pattern Recognition in Intelligent Environment・Sparse Representation for Image Super-Resolution

      Xian-Hua Han, Yen-Wei Chen（123-150）

      Springer  6 2014  （ ISBN:9783642548505 ）

      ▶ More details

    • Adaptive Noise Reduction and Edge Enhancement in Medical Images by using ICA?Computational Intelligence in Biomedical Imaging

      Xian-Hua Han, Yen-Wei chen

      Springer  6 2013  （ ISBN:9781461472445 ）

      ▶ More details

    • Multilinear Supervised Neighborhood Preserving Embedding Analysis of Local Descriptor Tensor?Principal Component Analysis

      Xian-Hua Han, Yen-wei Chen（91-106）

      INTECH  3 2012  （ ISBN:9789535101956 ）

      ▶ More details

    ▼display all

    Presentations

    • 正規化スパース性制約を用いたぼけカーネル推定および動画像復元への応用

      野島優輔, 韓先花, 陳延偉

      信学技報, vol. 115, no. 24, PRMU2015-1 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • Robust Blood Vessel of Liver Extraction Using Deep Convolution Network

      Titinunt Kitrungrotsakul, Xian-Hua Han, Yutaro Iwamoto, Yen-Wei Chen

      信学技報, vol. 116, no. 225, MI2016-48 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • 深層Residual Attention ネットワークを用いたスナップショット撮影からハイパースペクトル画像の復元

      寄元康平, 韓先花

      第19回情報科学技術フォーラム(FIT2020), H-006 

      ▶ More details

      Event date： 1 9 2020 - 3 9 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • Deep Attention ConvLSTM-UNetを用いた画像の降雨ノイズ除去

      山道航平, 韓先花

      第19回情報科学技術フォーラム(FIT2020), H-028 

      ▶ More details

      Event date： 1 9 2020 - 3 9 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • 注目領域の自動学習ネットワークによる画像キャプションの自動生成

      寄元康平, 韓先花

      第23回 画像の認識・理解シンポジウム (MIRU2020) 

      ▶ More details

      Event date： 3 8 2020 - 5 8 2020

      Presentation type：Poster presentation  

      researchmap

    • 知覚損失学習を用いた単一画像の雨筋除去

      山道航平, 韓先花

      第23回 画像の認識・理解シンポジウム (MIRU2020) 

      ▶ More details

      Event date： 3 8 2020 - 5 8 2020

      Presentation type：Poster presentation  

      researchmap

    • Multispectral and Hyperspectral Image Fusion with Deep Learned Prior

      Zhe Liu, Xian-Hua Han

      ▶ More details

      Event date： 3 8 2020 - 5 8 2020

      Presentation type：Poster presentation  

      researchmap

    • Deeply Supervised ConvLSTM-Unetによる医療用画像の分割

      方文昊, 韓先花

      第23回 画像の認識・理解シンポジウム (MIRU2020) 

      ▶ More details

      Event date： 3 8 2020 - 5 8 2020

      Presentation type：Poster presentation  

      researchmap

    • 段階的逆投影ネットワークによるハイパースペクトル画像の高解像度化

      吉武大輝, 韓先花

      第23回 画像の認識・理解シンポジウム (MIRU2020) 

      ▶ More details

      Event date： 3 8 2020 - 5 8 2020

      Presentation type：Poster presentation  

      researchmap

    • 画像のCNN特徴とLSTMのデンス結合構造を用いた画像キャプションの生成

      寄元康平, 韓先花

      2020年電子情報通信学会総合大会, D-12-42 

      ▶ More details

      Event date： 17 3 2020 - 20 3 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • 深層知覚損失ネットワークを用いた画像の降雨ノイズ除去

      山道航平, 韓先花

      2020年電子情報通信学会総合大会, D-12-59 

      ▶ More details

      Event date： 17 3 2020 - 20 3 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • 改良したSSDを用いた航空写真からsmall scare物体の検出と分類

      中井克啓, 韓 先花

      2020年電子情報通信学会総合大会, D-12-61 

      ▶ More details

      Event date： 17 3 2020 - 20 3 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • Mask R-CNNを用いた細胞核の検出とセグメンテーション

      藤田 世哉, 韓 先花

      2020年電子情報通信学会総合大会, A-15-5 

      ▶ More details

      Event date： 17 3 2020 - 20 3 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • Deep supervision U-netとconvLSTMの統合による医療画像の分割

      方 文昊, 韓 先花

      2020年電子情報通信学会総合大会, A-15-6 

      ▶ More details

      Event date： 17 3 2020 - 20 3 2020

      Presentation type：Oral presentation (general)  

      researchmap

    • Hyperspectral Residual Component Reconstruction with Deep ResNet

      Xian-Hua Han, Hiroki Yoshitake, Yen-Wei Chen

      第22回 画像の認識・理解シンポジウム (MIRU2019)  1 8 2019 

      ▶ More details

      Language：English   Presentation type：Poster presentation  

      researchmap

    • 敵対的生成ネットワークを用いたハイパースペクトル画像の高解像度化

      吉武 大輝, 韓 先花

      第22回 画像の認識・理解シンポジウム (MIRU2019)  31 7 2019 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • Region CNNを用いた細胞核の検出

      藤田 世哉, 韓 先花

      第22回 画像の認識・理解シンポジウム (MIRU2019)  31 7 2019 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • Computer-Aided Diagnosis of Liver Cancers Using Deep Learning with Fine-tuning

      Weibin WANG, Dong LIANG, Lanfen LIN, Hongjie HU, Qiaowei ZHANG, Qingqing CHEN, Yutaro IWAMOTO, Xianhua HAN, Yen-Wei CHEN

      電子情報通信学会PRMU研究会, 福岡工業大学  2018 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • 深層学習技術を用いたCT画像からの肝臓腫瘍候補の検出

      轟佳大, 韓先花, 岩本祐太郎, Lanfen Lin, Hongjie Hu, 陳延偉

      第８回横幹連合コンファレンス, P14-S, 京都  2 12 2017 

      ▶ More details

    • Deep Convolutional Neural Networkを用いた食事画像認識

      佐藤亮輔, 韓先花, 岩本祐太郎, 陳延偉

      平成29年電気関係学会関西連合大会  2017 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • Automatic Vessel Segmentation Using A combined Deep Network

      Titinunt KITRUNGROTSAKUL, Xian-Hua Han, Yutaro Iwamoto, Yen-Wei Chen

      第36回日本医用画像工学会大会(JAMIT2017).  2017 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • Learning an overcomplete codebook of tensor local structures for multi-phase medical image retrieval

      Jian Wang, Xian-Hua Han, Yingying Xu, Lanfen Lin, Hongjie Hu, Chongwu Jin, Yen-Wei Chen

      第36回日本医用画像工学会大会(JAMIT2017).  2017 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • スパース自己符号化器を用いてHEp-2 細胞画像認識システム

      韓 先花, 陳 延偉

      第36回日本医用画像工学会大会(JAMIT2017).  2017 

      ▶ More details

      Presentation type：Oral presentation (general)  

      researchmap

    • Robust Hyperspectral Image Super-Resolution via Constrained Non-negative Sparse Representation

      Xian-Hua Han, Boxin Shi, Yinqiang Zheng, Toru Kouyama, Atsunori Kanemura, Ryosuke Nakamura

      画像の認識・理解シンポジウム (MIRU2017).  2017 

      ▶ More details

      Presentation type：Poster presentation  

      researchmap

    • 深層学習を用いた肝臓腫瘍候補の検出

      轟佳大,TitinuntKitrungrotsakul, 岩本祐太郎, 韓先花, 陳延偉

      平成28年度電気関係学会関西連合大会  2016 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • A preliminary study on tensor codebook model for multiphase medical image retrieval

      Jian Wang, Xian-Hua Han, Yingying Xu, Lanfen Lin, Hongjie Hu, Chongwu Jin, Yen-Wei Chen

      信学技報, vol. 116, no. 225, MI2016-58  2016 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • HEp-2 cell classification using K-support spatial pooling in Deep CNNs

      Xian-Hua Han, Yen-Wei Chen

      信学技報, vol. 116, no. 225, MI2016-47  2016 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • BoVWとBoTWを用いた肝腫瘍CT画像の検索

      Yingying XU, Lanfen LIN, Hongjie HU, Dan WANG, Yitao LIU, Jian WANG, Xianhua HAN, Yen-Wei CHEN

      35回日本医用画像工学会大会  2016 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • Extraction of Liver Vessel using Graph Cut based Submodular on 3D data

      Titinunt Kitrungrotsakul, Xian-Hua Han, Yen-Wei Chen

      第35回日本医用画像工学会大会  2016 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • Combined Bags of Color and Texture Featuresを用いた食事画像認識

      笹野翔太, 韓先花, 陳延偉

      第19回画像の認識・理解シンポジウム（MIRU2016）,  2016 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • 三次元脳MR画像の超解像処理とセグメンテーションへの応用

      岩本祐太郎, 韓先花, 椎野顯彦, 陳延偉

      第19回画像の認識・理解シンポジウム（MIRU2016）,  2016 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • Improved Spatial Pyramid Pooling in Deep Convolutional Network for Remote Sensing Image Classification

      Xian-Hua Han, Ryosuke Nakamura, Yen-Wei Chen

      第19回画像の認識・理解シンポジウム  2016 

      ▶ More details

      Language：English   Presentation type：Poster presentation  

      researchmap

    • Super-Resolution for 3D Anisotropic Low-resolution MR Images Using Multilayer Convolutional Neural Networks

      Qiaochu Zhao, Yutaro Iwamoto, Xian-Hua Han, Yen-Wei Chen

      第19回画像の認識・理解シンポジウム（MIRU2016）  2016 

      ▶ More details

      Language：English   Presentation type：Poster presentation  

      researchmap

    • A preliminary study on multilinear sparse coding for medical image retrieval

      Jian Wang, Xian-Hua Han, Yingying Xu, Lanfen Lin, Hongjie Hu, Chongwu Jin, Yen-Wei Chen

      第19回画像の認識・理解シンポジウム（MIRU2016）  2016 

      ▶ More details

      Language：English   Presentation type：Poster presentation  

      researchmap

    • 2段階Hallucination法による医用画像の高解像度化と3次元医用画像への応用

      近藤佑斗, 韓先花, 陳延偉

      2015年電子情報通信総合大会，D-11-34, 滋賀県草津  2015 

      ▶ More details

    • Sparse and Low Rank DecompositionとICPを用いた頑健な点群位置合わせ法

      趙きょうそ, 韓先花, 陳延偉

      信学技報, vol. 115, no. 24, PRMU2015-5  2015 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • Liver Segmentation Using Superpixel-based Graph cuts and Shape Constraints

      TitinuntKitrungrotsakul, Xian-Hua Han, Yen-Wei Chen

      第18回画像の認識・理解シンポジウム  2015 

      ▶ More details

      Language：English   Presentation type：Poster presentation  

      researchmap

    • ベイズモデルを用いたCT画像からの腫瘍候補の検出

      今野悠, 韓先花, Xiong Wei, 陳延偉

      第18回画像の認識・理解シンポジウム  2015 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • Sparse and Low Rank DecompositionとHuber ICPを用いた頑健な点群位置合わせ法

      趙きょうそ, 韓先花, 陳延偉

      第18回画像の認識・理解シンポジウム  2015 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • 3次元顔面形態の局所的統計解析

      中津美冴, 韓先花, 木村亮介, 陳延偉

      第18回画像の認識・理解シンポジウム  2015 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • データ駆動モデルを用いた食事画像識別

      笹野翔太, 中島基輝, 韓先花, 陳延偉

      第18回画像の認識・理解シンポジウム  2015 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    • Deep Convolutional Neural Network による食事画像認識

      韓先花, 陳延偉

      信学技報, vol. 115, no. 224, PRMU2015-77  2015 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • Generalized Super-Vectorを用いた一般画像分類

      中島基輝, 韓 先花, 陳 延偉

      信学技報, vol. 114, no. 41, PRMU2014-21  2014 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • ノンパラメトリック確率モデルを用いたCT画像からの腫瘍候補の検出

      今野 悠, 韓 先花, 陳 延偉

      信学技報, vol. 114, no. 41, PRMU2014-14  2014 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • 統計学習法を用いた3次元顔面形態解析

      中津美冴, 韓 先花, 木村亮介, 陳 延偉

      信学技報, vol. 114, no. 41, PRMU2014-13  2014 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • 2段階Hallucination法および3次元医用画像の高解像度化への応

      近藤佑斗, 野島優補, 韓 先花, 陳 延偉

      信学技報, vol. 114, no. 41, PRMU2014-12  2014 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • Multi-touch Based Medical Interactive Visualization System

      Jian Wang, Hua-Wei Tu, Xian-Hua Han, Tomoko Tateyama, Yen-Wei Chen

      信学技報, vol. 114, no. 41, PRMU2014-７  2014 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • スパースコーディングとハイブリッドプーリングを用いた食事画像認識

      楠元理子, 韓 先花, 陳 延偉

      信学技報, vol. 114, no. 41, PRMU2014-1  2014 

      ▶ More details

      Language：Japanese   Presentation type：Oral presentation (general)  

      researchmap

    • Organ Bounding Box Annotation based on Adaptive Selection of Bone References

      Chunhua Dong・Amir, H. Foruzan, Xian-hua Han, Tomoko Tateyama, Yen-wei Chen

      信学技報, vol. 114, no. 103, MI2014-27  2014 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • Improved Interactive Medical Image Segmentation using Graph Cut and Superpixels

      Titinunt Kitrungrotsakul・Chunhua Dong, Xian-Hua Han, Yen-Wei Chen

      信学技報, vol. 114, no. 103, MI2014-25  2014 

      ▶ More details

      Language：English   Presentation type：Oral presentation (general)  

      researchmap

    • Data-Driven Model of Weber Local Descriptors for Visual Recognition

      Xian-Hua Han, Yen-Wei Chen, Gang Xu

      MIRU2014 第17回画像の認識・理解シンポジウム  2014 

      ▶ More details

      Language：English   Presentation type：Poster presentation  

      researchmap

    • Generalized Super-Vector

      中島 基輝, 韓 先花, 陳 延偉

      MIRU 2014第17回画像の認識・理解シンポジウム  2014 

      ▶ More details

      Language：Japanese   Presentation type：Poster presentation  

      researchmap

    ▼display all

    Teaching Experience

    • 4 2023 - 8 2023 
      応用機械学習特論 （ Yamaguchi University ）

      ▶ More details

    • 4 2023 - 8 2023 
      Machine Learning （ Yamaguchi University ）

      ▶ More details

    • 4 2023 - 8 2023 
      パターン認識 （ 山口大学 ）

      ▶ More details

    • 4 2022 - 8 2023 
      Data science technology II （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 8 2023 
      Advance Machine Learning （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 8 2023 
      プログラミング演習Ⅰ （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 8 2023 
      データ科学と社会Ⅱ （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 8 2023 
      データ科学と社会Ⅰ （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 8 2023 
      物理と情報のための基礎数学Ⅲ演習 （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 8 2023 
      基礎セミナー （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 3 2023 
      情報処理特論 （ Yamaguchi University ）

      ▶ More details

    • 4 2017 - 3 2021 
      グラフ理論 （ 山口大学 ）

      ▶ More details

    • 4 2017 - 3 2020 
      自然科学Ⅱ （ Yamaguchi University ）

      ▶ More details

    ▼display all

    Professional Memberships

    Research Projects

    • 弱教師深層学習と計算解剖モデルの統合による肝臓がんの診断支援

      日本学術振興会  科学研究費助成事業 

      陳 延偉, 岩本 祐太郎, 韓 先花, 古川 顕, 金崎 周造, 西川 郁子

      ▶ More details

      10 2020 - 3 2026

      Grant number：20KK0234

      Grant amount：\18720000 （ Direct Cost: \14400000 、 Indirect Cost：\4320000 ）

      2021年度は深層学習を用いた肝臓がんの診断支援を目的に以下の国際共同研究成果が得られた。
      (1) 自然画像に比べ、学習に利用できるアノテーションされた医用画像が非常に少ない。少ない学習データでも高精度に医用画像解析ができるように、人体の解剖知識と臓器の解剖モデルを先験知識として人工知能モデルに組み込むDeep Atlas Prior法を提案し、肝臓や脾臓などのセグメンテーションに高い精度を実現した。その成果は医用画像のトップ学術誌IEEE TMI (IF 10.094)で発表した。また、グラフ深層学習ネットワークを用いた半教師学習による高精度な臓器セグメンテーション法と高解像度を維持した３次元臓器セグメンテーション法を開発した。それぞれの成果は、医用画像解析分野のトップ国際学会MICCAI2021で発表した。
      (2) 弱教師学習による高精度な肝臓腫瘍検出法を開発した。従来の肝臓腫瘍検出の学習に、膨大な腫瘍れベールのアノテーションが必要であり、時間と労力がかかる。本提案法は、画像レベルのアノテーションだけ、高精度な肝臓腫瘍検出ができるようになった。本研究生かは医工学分野のトップ国際学会IEEE EMBC2021で発表した。
      (3) Phase AttentionとUncertain Region Inpaintingを用いた高精度な肝臓腫瘍のセグメンテーション法を開発した。それぞれの成果は、国際学術誌Medical Physics (IF 4.21)と医用画像解析分野のトップ国際学会MICCAI2021で発表した。
      上記の共同研究成果以外に、国際連携交流として、２ヶ月に一度三大学（立命館大学（日本）、浙江大学（中国）、インド工科大学（インド））によるオンライン国際連携研究交流会を開催している。毎回各大学から学生１名が最新の研究成果を紹介する。三大学の教員が共同指導する。

      researchmap

    • Weakly-Supervised Differentiation of Focal Liver Lesions Using Phase-Attention Net

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research 

      ▶ More details

      4 2021 - 3 2025

      Grant number：21H03470

      Grant amount：\17160000 （ Direct Cost: \13200000 、 Indirect Cost：\3960000 ）

      researchmap

    • Integrated Model-Driven and Data-Driven Framework for Hyper-Spectral Image Super-Resolution

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (C) 

      ▶ More details

      4 2020 - 3 2023

      Grant number：20K11867

      Grant amount：\4290000 （ Direct Cost: \3300000 、 Indirect Cost：\990000 ）

      researchmap

    • Tensor Sparse Coding for Temporal and Spatial Feature Extraction and Classification of Liver Lesions in Multi-phase CT Images

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) 

      ▶ More details

      4 2018 - 3 2021

      Grant number：18H03267

      Grant amount：\15600000 （ Direct Cost: \12000000 、 Indirect Cost：\3600000 ）

      researchmap

    • テンソルスパース表現による多時相CT画像の時空間特徴抽出と肝腫瘍性病変の診断支援

      日本学術振興会  基盤研究B 

      陳 延偉

      ▶ More details

      4 2018 - 3 2020

      Grant type：Competitive

      researchmap

    • Study on Image Representation Learning and Understanding based on Human's Perception Principle and Deep Statistical Analysis

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (C) 

      Han Xian-HUa

      ▶ More details

      4 2015 - 3 2018

      Grant number：15K00253

      Grant type：Competitive

      Grant amount：\4550000 （ Direct Cost: \3500000 、 Indirect Cost：\1050000 ）

      This study aimed at learning compact and inherent image representation for high-level vision probelms, and developed advanced image recognition and understanding methods. Our main achievements are three-fold: 1) Based on human’s perception principle, we transformed the raw-image domain into differential excitation domain and proposed to use the micro-texton as local descriptors for retaining all information, which would be distinguishable even for the subtle difference in image structures. 2) We proposed a novel middle-level image representation learning framework via exploring the deviation statistics of local descriptor set on the learned GMM model; 3) We stacked several layers of the middle level representation extraction framework, and proposed multiple-layer fisher network architecture for high-level feature learning. We applied our proposed image representation learning strategy for several image recognition applications, and proved much better performances can be achieved.

      researchmap

    • Localization and Alignment of Non-rigid Organs by Exhaustive Search of Position, Orientation,Scale and Deformation Parameters

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (C) 

      Xu Gang

      ▶ More details

      4 2014 - 3 2017

      Grant number：26330212

      Grant amount：\4810000 （ Direct Cost: \3700000 、 Indirect Cost：\1110000 ）

      This research shows that against the intuition of impossibility of exhaustive search of a relatively large number of parameters, by a number of techniques to reduce computational cost, we succeeded in localization and alignment of livers in MR volume data by exhaustive search of 15 parameters including 3 for position, 3 for orientation, 1 for scale, and 8 for deformation. Deformation is represented as a linear combination of principal components of a set of some 50 aligned 3D sample models.
      Using GPU for acceleration, we were able to find livers withour any human intervention at a speed of 1.5 minutes, which is now manually done by human operators at a speed of half a day.

      researchmap

    • 変形の伴う3次元形状間の全自動アラインメント

      日本学術振興会  基盤研究C 

      徐剛

      ▶ More details

      4 2014 - 3 2016

      Grant type：Competitive

      researchmap

    • Core Information Extraction Using Multilinear Manifold Learning and Hierarchical Algorithm for Image Understanding in Large Scale Dataset

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Young Scientists (B) 

      HAN Xian-Hua, CHEN Yen-Wei

      ▶ More details

      4 2012 - 3 2015

      Grant number：24700179

      Grant type：Competitive

      Grant amount：\3380000 （ Direct Cost: \2600000 、 Indirect Cost：\780000 ）

      In this project, we propose to represent an image as a local descriptor tensor and use a Multilinear Supervised Neighborhood Embedding (MSNE) for discriminant feature extraction, which is able to be used for subject or scene recognition. The contributions of our project include: (1) a image representation framework denoted as local descriptor tensor, which can effectively combine a moderate amount of local features together for image representation and be more efficient than the popular existing Bag-Of-Feature model; (2) a MSNE analysis algorithm, which can directly deal with the local descriptor tensor for extracting discriminant and compact features, and at the same time preserve neighborhood structure in tensor space; (3) a data-driven modeling procedure for raw features instead of the hand-craft local descriptors such as SIFT, SURF. We demonstrate the performance advantages of our proposed approach over existing techniques on understanding different types of benchmark database.

      researchmap

    • Generalized N-Dimensional Sparse Coding and Its Application to Computational Anatomy Models

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) 

      CHEN YAN WEI, TANAKA T, Hiromi, HAN Xian-Hua, SATO Yoshinobu, FURUKAWA Akira, MORIKAWA Shigehiro, TATEYAMA Tomoko

      ▶ More details

      4 2012 - 3 2015

      Grant number：24300076

      Grant type：Competitive

      Grant amount：\16250000 （ Direct Cost: \12500000 、 Indirect Cost：\3750000 ）

      Recently, sparse coding is a hot topic for efficient data representation, and has been widely used in computer vision field. In this project, we proposed a generalized ND sparse coding based on multi-linear algebra, for direct analysis of multi-dimensional data without unfolding process. Experiments results on noise reduction demonstrated that the proposed method can achieve better results compared with the conventional sparse coding. We also proposed a framework for local morphological analysis (local statistical shape models) of 3D organs based on sparse and low rank matrix decomposition and applied our proposed method to computer-aided diagnostics of liver cirrhosis. The local abnormal regions can be detected by estimating the sparse components. The norm of the sparse components can be used as a measure for classification of the normal and abnormal livers. The classification accuracy by our proposed method is improved to 95%.

      researchmap

    • ICA縮退フィルタと多重解像度フィルタを用いたIVR画像の画質改善法の開発

      日本学術振興会  若手研究B 

      韓 先 花

      ▶ More details

      4 2008 - 3 2011

      Grant type：Competitive

      researchmap

    • Statistical Volume Models of Human Anatomy and Their Applications to Computer Aided Diagnostics

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) 

      CHEN Yen-wei, TANAKA Hiromi, TANAKA Satoshi, SATO Yoshinobu, FURUKAWA Akira, MORIKAWA Shigehiro, TATEYAMA Tomoko, HAN Shanfa

      ▶ More details

      2009 - 2011

      Grant number：21300070

      Grant amount：\16510000 （ Direct Cost: \12700000 、 Indirect Cost：\3810000 ）

      We proposed a novel tensor based learning method called generalized N-dimensional principal component analysis(GND-PCA) for multi-dimensional data analysis. We also proposed a framework based on GND-PCA and a 3D shape normalization technique for statistical volume(texture) modeling of the liver. The 3D shape normalization technique is used for normalizing liver shapes in order to remove the liver shape variability and capture pure texture variations. The GND-PCA is used to overcome overfitting problems when the training samples are too much fewer than the dimension of the data. The preliminary results of leave-one-out experiments show that the statistical texture model of the liver built by our method can represent an untrained liver volume well, even though the mode is trained by fewer samples. We also demonstrate its potential application to classification of normal and abnormal(with tumors) livers.

      researchmap

    • IVR image enhancement with ICA domain shrinkage filter and multi-scale filter

      Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research Grant-in-Aid for Young Scientists (B) 

      HAN Xianhua

      ▶ More details

      2008 - 2009

      Grant number：20700170

      Grant amount：\1690000 （ Direct Cost: \1300000 、 Indirect Cost：\390000 ）

      This research project mainly focus on IVR image quality improvement and vessel enhancement. We developed an adaptive ICA shrinkage filter and multi-scale filter for Poisson noise reduction and vessel extraction, and then fusion these two developed filters to achieve high quality IVR images with distinctive vessel and less noise effectiveness. Therefore, we want to apply the developed algorithms for real IVR image enhancement for diagnosis application. The research achievements are as the following :
      (1) Developed multi-scale filter for Vessel extraction from IVR images, and explore the high-speed algorithm;
      (2) Proposed an adaptive ICA shrinkage filter for Poisson noise reduction in IVR images, and then fusion the two developed algorithm for IVR image enhancement;
      (3) Applied our developed algorithm to real IVR images for diagnosis application;
      (4) Adopted DVBV methods for quantitative evaluation of the enhanced IVR images by different algorithms. Evaluation results showed that the enhanced image by our proposed algorithm has the highest DV/BV rate.

      researchmap

    ▼display all

    Industrial property rights

    • 画像解析装置、画像登録装置および画像検索装置

      ▶ More details

      Application no：2008-151090 

      Announcement no：4228031 

      researchmap