Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469481
Ke Chen
In this paper, I present some new and joint work on local and selective segmentation models and algorithms which have potential applications in medical imaging. First I review a familiar segmentation model of global energy minimization framework in two dimensions (three dimensions may be presented similarly). Then I discuss selective segmentation models and several refined models where pre-defined geometric constraints guide local segmentation. Such 2D models can be generalized to 3D and some brief experiments are given to demonstrate the ideas of the paper. Finally I discuss the use of image registration methods to obtain geometric constraints or equivalent initial contours towards an automatic segmentation framework. As mentioned, the work discussed here represents a small portion of results obtained in the Liverpool's Centre for Mathematical Imaging Techniques (CMIT) and is jointly carried out with collaborators; for this paper, these include Noor Badshah (Peshawar, Pakistan), Jian-ping Zhang and Bo Yu (Dalian, China), Lavdie Rada (Liverpool), Noppadol Chumchob (Silpakorn, Thailand), Carlos Brito (Yucatan, Mexico), and Derek A. Gould (Royal Liverpool University Hospital, Liverpool).
在本文中，我介绍了一些新的和联合研究的局部和选择性分割模型和算法在医学成像中有潜在的应用。首先，我回顾了一个熟悉的二维全球能量最小化框架分割模型(三维也可以类似地呈现)。然后讨论了选择性分割模型和几种精细模型，其中预定义的几何约束指导局部分割。这种二维模型可以推广到三维，并给出了一些简短的实验来证明本文的思想。最后，我讨论了使用图像配准方法来获得自动分割框架的几何约束或等效初始轮廓。如上所述，这里讨论的工作代表了利物浦数学成像技术中心(CMIT)获得的结果的一小部分，并与合作者共同进行;在本文中，这些人包括Noor Badshah(巴基斯坦白沙瓦)，张建平和Bo Yu(中国大连)，Lavdie Rada(利物浦)，Noppadol Chumchob(泰国Silpakorn)， Carlos Brito(墨西哥尤卡坦半岛)和Derek A. Gould(利物浦皇家利物浦大学医院)。
{"title":"Selective variational image segmentation combined with registration: Models and algorithms","authors":"Ke Chen","doi":"10.1109/IPTA.2012.6469481","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469481","url":null,"abstract":"In this paper, I present some new and joint work on local and selective segmentation models and algorithms which have potential applications in medical imaging. First I review a familiar segmentation model of global energy minimization framework in two dimensions (three dimensions may be presented similarly). Then I discuss selective segmentation models and several refined models where pre-defined geometric constraints guide local segmentation. Such 2D models can be generalized to 3D and some brief experiments are given to demonstrate the ideas of the paper. Finally I discuss the use of image registration methods to obtain geometric constraints or equivalent initial contours towards an automatic segmentation framework. As mentioned, the work discussed here represents a small portion of results obtained in the Liverpool's Centre for Mathematical Imaging Techniques (CMIT) and is jointly carried out with collaborators; for this paper, these include Noor Badshah (Peshawar, Pakistan), Jian-ping Zhang and Bo Yu (Dalian, China), Lavdie Rada (Liverpool), Noppadol Chumchob (Silpakorn, Thailand), Carlos Brito (Yucatan, Mexico), and Derek A. Gould (Royal Liverpool University Hospital, Liverpool).","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133756725","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469531
Elena Battini Sonmez, B. Sankur, S. Albayrak
We consider the problem of emotion recognition in faces as well as subject identification in the presence of emotional facial expressions. We propose alternative solutions for this identification and recognition problems using the idea of sparsity, in terms of Sparse Representation based Classifier (SRC) paradigm. In both cases, the problem is formulated as finding the most parsimonious set of representatives from a training set, which will best reconstruct the test image. For emotion classification, we considered the six fundamental states and the SRC performance was compared with that of the Active Appearance Model (AAM) algorithm [1]. For face recognition displaying various emotions, in order to test the robustness of SRC, we considered gallery faces of subjects having one or more expression variety while the probe faces had a different expression. We experimented with both the whole faces or faces observed with multiple blocks. The SRC algorithm, while not demanding any training, performed surprisingly well in both emotion identification across subjects and subject identification across emotions.
我们考虑了人脸的情绪识别问题，以及存在情绪面部表情的主体识别问题。我们根据基于稀疏表示的分类器(SRC)范式，为这种识别和识别问题提出了使用稀疏性思想的替代解决方案。在这两种情况下，问题都被表述为从训练集中找到最简洁的代表集，这将最好地重建测试图像。对于情绪分类，我们考虑了六种基本状态，并将SRC算法的性能与Active Appearance Model (AAM)算法的性能进行了比较[1]。对于表现多种情绪的人脸识别，为了检验SRC的鲁棒性，我们考虑了具有一种或多种表情的被试画廊脸，而探测脸具有不同的表情。我们用整张脸或用多个块观察的脸进行了实验。SRC算法虽然不需要任何训练，但在跨主题的情感识别和跨情感的主题识别方面都表现得非常好。
{"title":"Classification with emotional faces via a robust sparse classifier","authors":"Elena Battini Sonmez, B. Sankur, S. Albayrak","doi":"10.1109/IPTA.2012.6469531","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469531","url":null,"abstract":"We consider the problem of emotion recognition in faces as well as subject identification in the presence of emotional facial expressions. We propose alternative solutions for this identification and recognition problems using the idea of sparsity, in terms of Sparse Representation based Classifier (SRC) paradigm. In both cases, the problem is formulated as finding the most parsimonious set of representatives from a training set, which will best reconstruct the test image. For emotion classification, we considered the six fundamental states and the SRC performance was compared with that of the Active Appearance Model (AAM) algorithm [1]. For face recognition displaying various emotions, in order to test the robustness of SRC, we considered gallery faces of subjects having one or more expression variety while the probe faces had a different expression. We experimented with both the whole faces or faces observed with multiple blocks. The SRC algorithm, while not demanding any training, performed surprisingly well in both emotion identification across subjects and subject identification across emotions.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129793679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469562
F. Derraz, L. Peyrodie, A. Taleb-Ahmed, G. Forzy
We present a new unsupervised segmentation based active contours model and local region texture descriptor. The proposed local region texture descriptor intrinsically describes the geometry of textural regions using the shape operator defined in Beltrami framework. The local texture descriptor is incorporated in the active contours using the Cauchy-Schwarz distance. The texture is discriminated by maximizing distance between the probability density functions which leads to distinguish textural objects of interest and background. We propose a fast Bregman split implementation of our segmentation algorithm based on the dual formulation of the Total Variation norm. Finally, we show results on some challenging images to illustrate segmentations that are possible.
{"title":"Texture segmentation using globally active contours model and Cauchy-Schwarz distance","authors":"F. Derraz, L. Peyrodie, A. Taleb-Ahmed, G. Forzy","doi":"10.1109/IPTA.2012.6469562","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469562","url":null,"abstract":"We present a new unsupervised segmentation based active contours model and local region texture descriptor. The proposed local region texture descriptor intrinsically describes the geometry of textural regions using the shape operator defined in Beltrami framework. The local texture descriptor is incorporated in the active contours using the Cauchy-Schwarz distance. The texture is discriminated by maximizing distance between the probability density functions which leads to distinguish textural objects of interest and background. We propose a fast Bregman split implementation of our segmentation algorithm based on the dual formulation of the Total Variation norm. Finally, we show results on some challenging images to illustrate segmentations that are possible.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114542337","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469536
M. M. Daud, Z. Kadim, S. L. Yuen, H. W. Hon, I. Faye, A. Malik
Video or image enhancement is a crucial part in image processing field as it improves the quality of the image before any further processes is applied on the image, which includes feature matching. In this paper, the accuracy of SURF feature descriptors used in feature matching between two input images of extreme illumination levels are evaluated. Based on the evaluation results, a novel pre-processing method to equalize both images intensity with respect to each other while maintaining the image content is proposed. We do so by fusing the cumulative histogram of the input images to compute a new cumulative histogram that will be used to remap both images. From this simple method, the results show that the intensity levels of the images are equalized and accuracy of the feature matching process is improved, in the event of extreme illumination scenario.
{"title":"A pre-processing approach for efficient feature matching process in extreme illumination scenario","authors":"M. M. Daud, Z. Kadim, S. L. Yuen, H. W. Hon, I. Faye, A. Malik","doi":"10.1109/IPTA.2012.6469536","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469536","url":null,"abstract":"Video or image enhancement is a crucial part in image processing field as it improves the quality of the image before any further processes is applied on the image, which includes feature matching. In this paper, the accuracy of SURF feature descriptors used in feature matching between two input images of extreme illumination levels are evaluated. Based on the evaluation results, a novel pre-processing method to equalize both images intensity with respect to each other while maintaining the image content is proposed. We do so by fusing the cumulative histogram of the input images to compute a new cumulative histogram that will be used to remap both images. From this simple method, the results show that the intensity levels of the images are equalized and accuracy of the feature matching process is improved, in the event of extreme illumination scenario.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"88 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123957647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469541
F. Vernier, Renaud Fallourd, J. Friedt, Yajing Yan, E. Trouvé, J. Nicolas, L. Moreau
Most of the image processing techniques have been first proposed and developed on small size images and progressively applied to larger and larger data sets resulting from new sensors and application requirements. In geosciences, digital cameras and remote sensing images can be used to monitor glaciers and to measure their surface velocity by different techniques. However, the image size and the number of acquisitions to be processed to analyze time series become a critical issue to derive displacement fields by the conventional correlation technique. In this paper, an efficient correlation software is used to compute from optical images the motion of a serac fall and from Synthetic Aperture Radar (SAR) images the motion of Alpine glaciers. The optical images are acquired by a digital camera installed near the Argentière glacier (Chamonix, France) and the SAR images are acquired by the high resolution TerraSAR-X satellite over the Mont-Blanc area. The results illustrate the potential of this software to monitor the glacier flow with camera images acquired every 2 h and with the size of the TerraSAR-X scenes covering 30 × 50 km2.
{"title":"Glacier flow monitoring by digital camera and space-borne SAR images","authors":"F. Vernier, Renaud Fallourd, J. Friedt, Yajing Yan, E. Trouvé, J. Nicolas, L. Moreau","doi":"10.1109/IPTA.2012.6469541","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469541","url":null,"abstract":"Most of the image processing techniques have been first proposed and developed on small size images and progressively applied to larger and larger data sets resulting from new sensors and application requirements. In geosciences, digital cameras and remote sensing images can be used to monitor glaciers and to measure their surface velocity by different techniques. However, the image size and the number of acquisitions to be processed to analyze time series become a critical issue to derive displacement fields by the conventional correlation technique. In this paper, an efficient correlation software is used to compute from optical images the motion of a serac fall and from Synthetic Aperture Radar (SAR) images the motion of Alpine glaciers. The optical images are acquired by a digital camera installed near the Argentière glacier (Chamonix, France) and the SAR images are acquired by the high resolution TerraSAR-X satellite over the Mont-Blanc area. The results illustrate the potential of this software to monitor the glacier flow with camera images acquired every 2 h and with the size of the TerraSAR-X scenes covering 30 × 50 km2.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"181 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124525511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469519
Amel Boulemnadjel, F. Hachouf
In this paper a new subspaces clustering algorithm is proposed. This method has two levels, the first one is an iterative algorithm based on the minimization of an objective function. The density is introduced in this objective function where the distances between points become relatively uniform in high dimensional spaces. In such cases, the density of cluster may give better results. The idea of the second level is to find the clusters in each subspace individually. We applied the proposed method to medical tomography scan image without Intravenous or IV contrast dye. Then we compare the results with the same image with IV contrast. However in some cases, there are risks associated with this injection, where the mortality risk is low but not null. This method can reduce the use of this injection. Experimental results on synthetic and real datasets show that the proposed method gives good results in medical tomography image.
{"title":"A new method for finding clusters embedded in subspaces applied to medical tomography scan image","authors":"Amel Boulemnadjel, F. Hachouf","doi":"10.1109/IPTA.2012.6469519","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469519","url":null,"abstract":"In this paper a new subspaces clustering algorithm is proposed. This method has two levels, the first one is an iterative algorithm based on the minimization of an objective function. The density is introduced in this objective function where the distances between points become relatively uniform in high dimensional spaces. In such cases, the density of cluster may give better results. The idea of the second level is to find the clusters in each subspace individually. We applied the proposed method to medical tomography scan image without Intravenous or IV contrast dye. Then we compare the results with the same image with IV contrast. However in some cases, there are risks associated with this injection, where the mortality risk is low but not null. This method can reduce the use of this injection. Experimental results on synthetic and real datasets show that the proposed method gives good results in medical tomography image.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123434572","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469565
A. Karantza, Sonia Lopez-Alarcon, N. Cahill
B-spline signal processing operations are widely used in the analysis of two and three-dimensional images. In this paper, we investigate and compare some of these basic operations (direct transformations, indirect transformations, and computation of partial derivatives) by (1) recursive filter based implementations in MATLAB and C++, and (2) GPU-accelerated implementations in CUDA. All operations are compared at a variety of resolution levels on a 2-D panoramic image as well as a 3-D magnetic resonance (MR) image. Results indicate significant improvements in efficiency for the CUDA implementations. A MATLAB toolkit implementing the various B-spline signal processing tasks as well as the C++ and CUDA implementation described here is currently publicly available.
{"title":"A comparison of sequential and GPU-accelerated implementations of B-spline signal processing operations for 2-D and 3-D images","authors":"A. Karantza, Sonia Lopez-Alarcon, N. Cahill","doi":"10.1109/IPTA.2012.6469565","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469565","url":null,"abstract":"B-spline signal processing operations are widely used in the analysis of two and three-dimensional images. In this paper, we investigate and compare some of these basic operations (direct transformations, indirect transformations, and computation of partial derivatives) by (1) recursive filter based implementations in MATLAB and C++, and (2) GPU-accelerated implementations in CUDA. All operations are compared at a variety of resolution levels on a 2-D panoramic image as well as a 3-D magnetic resonance (MR) image. Results indicate significant improvements in efficiency for the CUDA implementations. A MATLAB toolkit implementing the various B-spline signal processing tasks as well as the C++ and CUDA implementation described here is currently publicly available.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123458985","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469555
R. Samet, S. E. Amrahov, Ali Hikmet Ziroglu
Image segmentation is a process of partitioning the images into meaningful regions that are ready to analyze. Segmentation of rock thin section images is not trivial task due to the unpredictable structures and features of minerals. In this paper, we propose Fuzzy Rule-Based Image Segmentation technique to segment rock thin section images. Proposed technique uses RGB images of rock thin sections as input and gives segmented into minerals images as output. In order to show an advantage of proposed technique the rock thin section images were also segmented by known Fuzzy C-Means technique. Both techniques were applied to many different rock thin section images. The obtained results of proposed Fuzzy Rule-Based Image Segmentation and Fuzzy C-Means techniques were compared. Implementation results showed that proposed image segmentation technique has better accuracy than known ones.
{"title":"Fuzzy Rule-Based Image Segmentation technique for rock thin section images","authors":"R. Samet, S. E. Amrahov, Ali Hikmet Ziroglu","doi":"10.1109/IPTA.2012.6469555","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469555","url":null,"abstract":"Image segmentation is a process of partitioning the images into meaningful regions that are ready to analyze. Segmentation of rock thin section images is not trivial task due to the unpredictable structures and features of minerals. In this paper, we propose Fuzzy Rule-Based Image Segmentation technique to segment rock thin section images. Proposed technique uses RGB images of rock thin sections as input and gives segmented into minerals images as output. In order to show an advantage of proposed technique the rock thin section images were also segmented by known Fuzzy C-Means technique. Both techniques were applied to many different rock thin section images. The obtained results of proposed Fuzzy Rule-Based Image Segmentation and Fuzzy C-Means techniques were compared. Implementation results showed that proposed image segmentation technique has better accuracy than known ones.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"292 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114720497","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469571
C. Revol-Muller, J. Rose, A. Pacureanu, F. Peyrin, C. Odet
In this paper, we propose two solutions to integrate shape prior in a segmentation process based on region growing. Our special region growing algorithm relies upon a variational framework which allows to easily take into account shape prior in the segmentation process. Region growing is described as an optimization process that aims to minimize some special energy combining intensity function and shape information. Two kinds of energy are proposed depending on the existence of a reference model or the possibility to assess some shape features at voxel level. We applied positively these two approaches in the context of life imaging in order to segment mice kidneys from small animal CT-images and lacuno-canicular network from experimental high resolution Synchrotron Radiation X-Ray Computed Tomography (SRμCT) images.
{"title":"Shape prior in Variational Region Growing","authors":"C. Revol-Muller, J. Rose, A. Pacureanu, F. Peyrin, C. Odet","doi":"10.1109/IPTA.2012.6469571","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469571","url":null,"abstract":"In this paper, we propose two solutions to integrate shape prior in a segmentation process based on region growing. Our special region growing algorithm relies upon a variational framework which allows to easily take into account shape prior in the segmentation process. Region growing is described as an optimization process that aims to minimize some special energy combining intensity function and shape information. Two kinds of energy are proposed depending on the existence of a reference model or the possibility to assess some shape features at voxel level. We applied positively these two approaches in the context of life imaging in order to segment mice kidneys from small animal CT-images and lacuno-canicular network from experimental high resolution Synchrotron Radiation X-Ray Computed Tomography (SRμCT) images.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"138 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116372246","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2012-10-01DOI: 10.1109/IPTA.2012.6469569
S. Mahmoudi, P. Manneback
Image processing algorithms present a necessary tool for various domains related to computer vision, such as video surveillance, medical imaging, pattern recognition, etc. However, these algorithms are hampered by their high consumption of both computing power and memory, which increase significantly when processing large sets of images. In this work, we propose a development scheme enabling an efficient exploitation of parallel (GPU) and heterogeneous platforms (Multi-CPU/Multi-GPU), for improving performance of single and multiple image processing algorithms. The proposed scheme allows a full exploitation of hybrid platforms based on efficient scheduling strategies. It enables also overlapping data transfers by kernels executions using CUDA streaming technique within multiple GPUs. We present also parallel and heterogeneous implementations of several features extraction algorithms such as edge and corner detection. Experimentations have been conducted using a set of high resolution images, showing a global speedup ranging from 5 to 30, by comparison with CPU implementations.
{"title":"Efficient exploitation of heterogeneous platforms for images features extraction","authors":"S. Mahmoudi, P. Manneback","doi":"10.1109/IPTA.2012.6469569","DOIUrl":"https://doi.org/10.1109/IPTA.2012.6469569","url":null,"abstract":"Image processing algorithms present a necessary tool for various domains related to computer vision, such as video surveillance, medical imaging, pattern recognition, etc. However, these algorithms are hampered by their high consumption of both computing power and memory, which increase significantly when processing large sets of images. In this work, we propose a development scheme enabling an efficient exploitation of parallel (GPU) and heterogeneous platforms (Multi-CPU/Multi-GPU), for improving performance of single and multiple image processing algorithms. The proposed scheme allows a full exploitation of hybrid platforms based on efficient scheduling strategies. It enables also overlapping data transfers by kernels executions using CUDA streaming technique within multiple GPUs. We present also parallel and heterogeneous implementations of several features extraction algorithms such as edge and corner detection. Experimentations have been conducted using a set of high resolution images, showing a global speedup ranging from 5 to 30, by comparison with CPU implementations.","PeriodicalId":267290,"journal":{"name":"2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121268145","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：