Pub Date : 2017-11-01DOI: 10.1109/IPTA.2017.8310128
Hidehiko Shishido, I. Kitahara
Camera calibration that estimates the projective relationship between 3D and 2D image spaces is one of the most crucial processes for such 3D image processing as 3D reconstruction and 3D tracking. A strong calibration method, which needs to place landmarks with known 3D positions, is a common technique. However, as the target space becomes large, landmark placement becomes more complicated. Although a weak-calibration method does not need known landmarks to estimate a projective transformation matrix from the correspondence information among multi-view images, the estimation precision depends on the accuracy of the correspondence. When multiple cameras are arranged sparsely, detecting sufficient corresponding points is difficult. In this research, we propose a calibration method that bridges sparse multiple cameras with mobile camera images. The mobile camera captures video images while moving among sparse multi-view cameras. The captured video resembles dense multi-view images and includes sparse multi-view images so that weak-calibration is effective. We confirmed the appropriate spacing between the images through comparative experiments of camera calibration accuracy by changing the number of bridging images and applied our proposed method to multiple capturing experiments in a large-scale space and verified its robustness.
{"title":"Calibration method for sparse multi-view cameras by bridging with a mobile camera","authors":"Hidehiko Shishido, I. Kitahara","doi":"10.1109/IPTA.2017.8310128","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310128","url":null,"abstract":"Camera calibration that estimates the projective relationship between 3D and 2D image spaces is one of the most crucial processes for such 3D image processing as 3D reconstruction and 3D tracking. A strong calibration method, which needs to place landmarks with known 3D positions, is a common technique. However, as the target space becomes large, landmark placement becomes more complicated. Although a weak-calibration method does not need known landmarks to estimate a projective transformation matrix from the correspondence information among multi-view images, the estimation precision depends on the accuracy of the correspondence. When multiple cameras are arranged sparsely, detecting sufficient corresponding points is difficult. In this research, we propose a calibration method that bridges sparse multiple cameras with mobile camera images. The mobile camera captures video images while moving among sparse multi-view cameras. The captured video resembles dense multi-view images and includes sparse multi-view images so that weak-calibration is effective. We confirmed the appropriate spacing between the images through comparative experiments of camera calibration accuracy by changing the number of bridging images and applied our proposed method to multiple capturing experiments in a large-scale space and verified its robustness.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133908118","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310085
M. Akhloufi, Tom Toulouse, L. Rossi, X. Maldague
This paper proposes a new multimodal stereovision framework for wildland fires detection and analysis. The proposed system uses near infrared and visible images to robustly segment the fires and extract their three-dimensional characteristics during propagation. It uses multiple multimodal stereovision systems to capture complementary views of the fire front. A new registration approach is proposed, it uses multisensory fusion based on GNSS and IMU data to extract the projection matrix that permits the representation of the 3D reconstructed fire in a common reference frame. The fire parameters are extracted in 3D space during fire propagation using the complete reconstructed fire. The obtained results show the efficiency of the proposed system for wildland fires research and firefighting decision support in operational scenarios.
{"title":"Multimodal three-dimensional vision for wildland fires detection and analysis","authors":"M. Akhloufi, Tom Toulouse, L. Rossi, X. Maldague","doi":"10.1109/IPTA.2017.8310085","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310085","url":null,"abstract":"This paper proposes a new multimodal stereovision framework for wildland fires detection and analysis. The proposed system uses near infrared and visible images to robustly segment the fires and extract their three-dimensional characteristics during propagation. It uses multiple multimodal stereovision systems to capture complementary views of the fire front. A new registration approach is proposed, it uses multisensory fusion based on GNSS and IMU data to extract the projection matrix that permits the representation of the 3D reconstructed fire in a common reference frame. The fire parameters are extracted in 3D space during fire propagation using the complete reconstructed fire. The obtained results show the efficiency of the proposed system for wildland fires research and firefighting decision support in operational scenarios.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134147906","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310083
I. Schiopu, M. Gabbouj, Alexandros Iosifidis, B. Zeng, Shuaicheng Liu
The paper proposes an image segmentation method for lossless compression of plenoptic images. Each light-field image captured by the plenoptic camera is processed to obtain a stack of subaperture images. Each subaperture image is encoded by using a gradient-base detector which classifies the image edges and designs refined contexts for an improved prediction and segmentation. The paper's main contribution is a new segmentation method which generates a preliminary segmentation, either by scaling the intensity differences or by using a quantum cut based algorithm, and merges it with an edge ranking-based segmentation. The results show around 2% improved performance compared to the state-of-the-art for a dataset of 118 plenoptic images.
{"title":"Subaperture image segmentation for lossless compression","authors":"I. Schiopu, M. Gabbouj, Alexandros Iosifidis, B. Zeng, Shuaicheng Liu","doi":"10.1109/IPTA.2017.8310083","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310083","url":null,"abstract":"The paper proposes an image segmentation method for lossless compression of plenoptic images. Each light-field image captured by the plenoptic camera is processed to obtain a stack of subaperture images. Each subaperture image is encoded by using a gradient-base detector which classifies the image edges and designs refined contexts for an improved prediction and segmentation. The paper's main contribution is a new segmentation method which generates a preliminary segmentation, either by scaling the intensity differences or by using a quantum cut based algorithm, and merges it with an edge ranking-based segmentation. The results show around 2% improved performance compared to the state-of-the-art for a dataset of 118 plenoptic images.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116819904","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310090
Chao Cui, N. Sang, Changxin Gao, Lei Zou
Vehicle re-identification has become a fundamental task because of the growing explosion in the use of surveillance cameras in public security. The most widely used solution is based on license plate verification. But when facing the vehicle without a license, deck cars and other license plate information error or missing situation, vehicle searching is still a challenging problem. This paper proposed a vehicle re-identification method based on deep learning which exploit a two-branch Multi-DNN Fusion Siamese Neural Network (MFSNN) to fuses the classification outputs of color, model and pasted marks on the windshield and map them into a Euclidean space where distance can be directly used to measure the similarity of arbitrary two vehicles. In order to achieve this goal, we present a method of vehicle color identification based on Alex net, a method of vehicle model identification based on VGG net, a method of pasted marks detection and identification based on Faster R-CNN. We evaluate our MFSNN method on VehicleID dataset and in the experiment. Experiment results show that our method can achieve promising results.
{"title":"Vehicle re-identification by fusing multiple deep neural networks","authors":"Chao Cui, N. Sang, Changxin Gao, Lei Zou","doi":"10.1109/IPTA.2017.8310090","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310090","url":null,"abstract":"Vehicle re-identification has become a fundamental task because of the growing explosion in the use of surveillance cameras in public security. The most widely used solution is based on license plate verification. But when facing the vehicle without a license, deck cars and other license plate information error or missing situation, vehicle searching is still a challenging problem. This paper proposed a vehicle re-identification method based on deep learning which exploit a two-branch Multi-DNN Fusion Siamese Neural Network (MFSNN) to fuses the classification outputs of color, model and pasted marks on the windshield and map them into a Euclidean space where distance can be directly used to measure the similarity of arbitrary two vehicles. In order to achieve this goal, we present a method of vehicle color identification based on Alex net, a method of vehicle model identification based on VGG net, a method of pasted marks detection and identification based on Faster R-CNN. We evaluate our MFSNN method on VehicleID dataset and in the experiment. Experiment results show that our method can achieve promising results.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"399 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132217970","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310126
A. Das, K. Ruppin, P. Dave, Sharfudheen Pv
Research in advanced driver assistance system (ADAS) is an important step towards achieving the goal of autonomous intelligent vehicle. Vehicle detection and its distance estimation is an important solution of ADAS for forward collision warning applications. Partial occlusions of passing vehicles makes their detections tedious yet the accuracy of vehicle detection in all its forms in the scene and their corresponding distance estimation is a vital factor to deploy the solution. A small deviation in detection and distance accuracy could end up in a greater mishap in ADAS and AV (Autonomous Vehicle). The proposed framework addresses the aforementioned problems of detection of passing vehicles and perfecting distance measurement by accurate lower bound estimation through Inter and Intra-Frame Flow Correspondence (I2F2C). The proposed generic framework of 12F2C could be employed as a plug-in for the existing machine learning (ML) [1]/ deep learning (DL) [2] based algorithms for improving accuracy of distance estimation of vehicles and also improve accuracy and performance of passing vehicle detection with a detailed mathematical model of motion confidence.
{"title":"Vehicle boundary improvement and passing vehicle detection in driver assistance by flow distribution","authors":"A. Das, K. Ruppin, P. Dave, Sharfudheen Pv","doi":"10.1109/IPTA.2017.8310126","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310126","url":null,"abstract":"Research in advanced driver assistance system (ADAS) is an important step towards achieving the goal of autonomous intelligent vehicle. Vehicle detection and its distance estimation is an important solution of ADAS for forward collision warning applications. Partial occlusions of passing vehicles makes their detections tedious yet the accuracy of vehicle detection in all its forms in the scene and their corresponding distance estimation is a vital factor to deploy the solution. A small deviation in detection and distance accuracy could end up in a greater mishap in ADAS and AV (Autonomous Vehicle). The proposed framework addresses the aforementioned problems of detection of passing vehicles and perfecting distance measurement by accurate lower bound estimation through Inter and Intra-Frame Flow Correspondence (I2F2C). The proposed generic framework of 12F2C could be employed as a plug-in for the existing machine learning (ML) [1]/ deep learning (DL) [2] based algorithms for improving accuracy of distance estimation of vehicles and also improve accuracy and performance of passing vehicle detection with a detailed mathematical model of motion confidence.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"108 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124301891","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310102
M. D. Bah, A. Hafiane, R. Canals
Traditional weeds controlling tended to be spraying herbicides in all the fields. Such method not only requires huge quantities of herbicides but impact environment and humans health. In this paper, we propose a new method of crop/weeds discrimination using imagery provided by an unmanned aerial vehicle (UAV). This method is based on the vegetation skeleton, the Hough transform and the spatial relationship of superpixels created by the simple linear iterative clustering (SLIC). The combination of the spatial relationship of superpixels and their positions in the detected crop lines allows to detect intraline weeds. Our method shows its robustness in presence of weed patches close to crop lines as well as for the detection of crop lines as for weed detection.
{"title":"Weeds detection in UAV imagery using SLIC and the hough transform","authors":"M. D. Bah, A. Hafiane, R. Canals","doi":"10.1109/IPTA.2017.8310102","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310102","url":null,"abstract":"Traditional weeds controlling tended to be spraying herbicides in all the fields. Such method not only requires huge quantities of herbicides but impact environment and humans health. In this paper, we propose a new method of crop/weeds discrimination using imagery provided by an unmanned aerial vehicle (UAV). This method is based on the vegetation skeleton, the Hough transform and the spatial relationship of superpixels created by the simple linear iterative clustering (SLIC). The combination of the spatial relationship of superpixels and their positions in the detected crop lines allows to detect intraline weeds. Our method shows its robustness in presence of weed patches close to crop lines as well as for the detection of crop lines as for weed detection.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"102 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115627099","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310080
Pavan Sudheendra, D. Jayagopi
This paper addresses the problem of automatically assessing the aesthetic quality of a photograph and providing actionable feedback to the photographer. Towards this task we have designed novel genre-specific attributes (for e.g. Noise level in a night mode photograph or Depth perception for the landscape mode). Using a collection of these mode relevant attributes we improved the assessment accuracy for three modes and reached state-of-the-art on the other one mode we investigated. These intuitive attributes are also visualized as a visual signature of the photograph. This representation can act as an actionable feedback to an aspiring amateur photographer.
{"title":"Genre linked automated assessment and feedback of photographs based on visual aesthetics","authors":"Pavan Sudheendra, D. Jayagopi","doi":"10.1109/IPTA.2017.8310080","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310080","url":null,"abstract":"This paper addresses the problem of automatically assessing the aesthetic quality of a photograph and providing actionable feedback to the photographer. Towards this task we have designed novel genre-specific attributes (for e.g. Noise level in a night mode photograph or Depth perception for the landscape mode). Using a collection of these mode relevant attributes we improved the assessment accuracy for three modes and reached state-of-the-art on the other one mode we investigated. These intuitive attributes are also visualized as a visual signature of the photograph. This representation can act as an actionable feedback to an aspiring amateur photographer.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128475323","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310137
B. Stojanovic, Oge Marques, A. Neskovic
This paper presents a new image dataset for evaluating approaches for overlapped fingerprint separation. The VLATACOM dataset consists of 120,000 synthetically overlapped test images (and the associated masks), with and without noise, processed with three different rotation angles, and in two variations of overall brightness. Each image in the dataset also contains information about the number of the singular points within its overlapped region, which is a distinctly unique feature of the proposed dataset. The paper also reports early experimental results which demonstrate the suitability of the VLATACOM dataset for overlapped fingerprint separation research. The dataset, along with testing results, is freely and publicly available.
{"title":"A novel synthetic dataset for research in overlapped fingerprint separation","authors":"B. Stojanovic, Oge Marques, A. Neskovic","doi":"10.1109/IPTA.2017.8310137","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310137","url":null,"abstract":"This paper presents a new image dataset for evaluating approaches for overlapped fingerprint separation. The VLATACOM dataset consists of 120,000 synthetically overlapped test images (and the associated masks), with and without noise, processed with three different rotation angles, and in two variations of overall brightness. Each image in the dataset also contains information about the number of the singular points within its overlapped region, which is a distinctly unique feature of the proposed dataset. The paper also reports early experimental results which demonstrate the suitability of the VLATACOM dataset for overlapped fingerprint separation research. The dataset, along with testing results, is freely and publicly available.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116901482","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310113
Le Thanh Nguyen-Meidine, Eric Granger, M. Kiran, Louis-Antoine Blais-Morin
Detecting faces and heads appearing in video feeds are challenging tasks in real-world video surveillance applications due to variations in appearance, occlusions and complex backgrounds. Recently, several CNN architectures have been proposed to increase the accuracy of detectors, although their computational complexity can be an issue, especially for realtime applications, where faces and heads must be detected live using high-resolution cameras. This paper compares the accuracy and complexity of state-of-the-art CNN architectures that are suitable for face and head detection. Single pass and region-based architectures are reviewed and compared empirically to baseline techniques according to accuracy and to time and memory complexity on images from several challenging datasets. The viability of these architectures is analyzed with real-time video surveillance applications in mind. Results suggest that, although CNN architectures can achieve a very high level of accuracy compared to traditional detectors, their computational cost can represent a limitation for many practical real-time applications.
{"title":"A comparison of CNN-based face and head detectors for real-time video surveillance applications","authors":"Le Thanh Nguyen-Meidine, Eric Granger, M. Kiran, Louis-Antoine Blais-Morin","doi":"10.1109/IPTA.2017.8310113","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310113","url":null,"abstract":"Detecting faces and heads appearing in video feeds are challenging tasks in real-world video surveillance applications due to variations in appearance, occlusions and complex backgrounds. Recently, several CNN architectures have been proposed to increase the accuracy of detectors, although their computational complexity can be an issue, especially for realtime applications, where faces and heads must be detected live using high-resolution cameras. This paper compares the accuracy and complexity of state-of-the-art CNN architectures that are suitable for face and head detection. Single pass and region-based architectures are reviewed and compared empirically to baseline techniques according to accuracy and to time and memory complexity on images from several challenging datasets. The viability of these architectures is analyzed with real-time video surveillance applications in mind. Results suggest that, although CNN architectures can achieve a very high level of accuracy compared to traditional detectors, their computational cost can represent a limitation for many practical real-time applications.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114240076","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 : 2017-11-01DOI: 10.1109/IPTA.2017.8310118
O. B. Ahmed, F. Lecellier, M. Paccalin, C. Fernandez-Maloigne
Visual inspection is the first step performed by clinicians during evaluation of medical images in image-based diagnosis. This behavior can be automated using computational saliency models. In this paper, we investigate the potential role of visual saliency for computer-aided diagnosis of Alzheimer's disease (AD). We propose a multi-view saliency-based framework to detect abnormalities from structural Magnitude Resonance Imaging (MRI) and classify subjects in a Multiple Kernel Learning (MKL) framework. The obtained saliency maps are able to detect relevant brain areas for early AD diagnosis. The effectiveness of the proposed approach was evaluated on structural MRI of 509 subjects from the ADNI dataset. We achieved accuracy of 88.98% (specificity of 94.4% and a sensitivity of 83.46%) and 81.31% (specificity of 84.22% and a sensitivity of 74.21%) classification and for respectively AD versus Normal Control(NC) and NC versus Mild Cognitive Impairment (MCI). For the most challenging classification task (AD versus MCI), we reached an accuracy of 79.8%, a specificity of 79.93% and a sensitivity of 64.02%.
{"title":"Multi-view visual saliency-based MRI classification for alzheimer's disease diagnosis","authors":"O. B. Ahmed, F. Lecellier, M. Paccalin, C. Fernandez-Maloigne","doi":"10.1109/IPTA.2017.8310118","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310118","url":null,"abstract":"Visual inspection is the first step performed by clinicians during evaluation of medical images in image-based diagnosis. This behavior can be automated using computational saliency models. In this paper, we investigate the potential role of visual saliency for computer-aided diagnosis of Alzheimer's disease (AD). We propose a multi-view saliency-based framework to detect abnormalities from structural Magnitude Resonance Imaging (MRI) and classify subjects in a Multiple Kernel Learning (MKL) framework. The obtained saliency maps are able to detect relevant brain areas for early AD diagnosis. The effectiveness of the proposed approach was evaluated on structural MRI of 509 subjects from the ADNI dataset. We achieved accuracy of 88.98% (specificity of 94.4% and a sensitivity of 83.46%) and 81.31% (specificity of 84.22% and a sensitivity of 74.21%) classification and for respectively AD versus Normal Control(NC) and NC versus Mild Cognitive Impairment (MCI). For the most challenging classification task (AD versus MCI), we reached an accuracy of 79.8%, a specificity of 79.93% and a sensitivity of 64.02%.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123960979","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：