Pub Date : 2017-11-01DOI: 10.1109/IPTA.2017.8310131
Kamel Guissous, V. Gouet-Brunet
With the increase of image datasets size and of descriptors complexity in Content-Based Image Retrieval (CBIR) and Computer Vision, it is essential to find a way to limit the amount of manipulated data, while keeping its quality. Instead of treating the entire image, the selection of regions which hold the essence of information is a relevant option to reach this goal. As the visual saliency aims at highlighting the areas of the image which are the most important for a given task, in this paper we propose to exploit visual saliency maps to prune the most salient image features. A novel visual saliency approach based on the local distribution analysis of the edges orientation, particularly dedicated to structured contents, such as street view images of urban environments, is proposed. It is evaluated for CBIR according to three criteria: quality of retrieval, volume of manipulated features and computation time. The proposal can be exploited into various applications involving large sets of local visual features; here it is experimented within two applications: cross-domain image retrieval and image-based vehicle localisation.
{"title":"Image retrieval based on saliency for urban image contents","authors":"Kamel Guissous, V. Gouet-Brunet","doi":"10.1109/IPTA.2017.8310131","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310131","url":null,"abstract":"With the increase of image datasets size and of descriptors complexity in Content-Based Image Retrieval (CBIR) and Computer Vision, it is essential to find a way to limit the amount of manipulated data, while keeping its quality. Instead of treating the entire image, the selection of regions which hold the essence of information is a relevant option to reach this goal. As the visual saliency aims at highlighting the areas of the image which are the most important for a given task, in this paper we propose to exploit visual saliency maps to prune the most salient image features. A novel visual saliency approach based on the local distribution analysis of the edges orientation, particularly dedicated to structured contents, such as street view images of urban environments, is proposed. It is evaluated for CBIR according to three criteria: quality of retrieval, volume of manipulated features and computation time. The proposal can be exploited into various applications involving large sets of local visual features; here it is experimented within two applications: cross-domain image retrieval and image-based vehicle localisation.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"48 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":"127542734","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.2017.8310107
Saba Mumtaz, Naima Mubariz, S. Saleem, M. Fraz
In video-surveillance, person re-identification is described as the task of recognizing distinct individuals over a network of cameras. It is an extremely challenging task since visual appearances of people can change significantly when viewed in different cameras. Many person re-identification methods offer distinct advantages over each other in terms of robustness to lighting, scale and pose variations. Keeping this consideration in mind, this paper proposes an effective new person reidentification model which incorporates several recent state-of-the-art feature extraction methodologies such as GOG, WHOS and LOMO features into a single framework. Effectiveness of each feature type is estimated and optimal weights for the similarity measurements are assigned through a multiple metric learning method. The proposed re-identification approach is then tested on multiple benchmark person re-identification datasets where it outperforms many other state-of-the-art methodologies.
{"title":"Weighted hybrid features for person re-identification","authors":"Saba Mumtaz, Naima Mubariz, S. Saleem, M. Fraz","doi":"10.1109/IPTA.2017.8310107","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310107","url":null,"abstract":"In video-surveillance, person re-identification is described as the task of recognizing distinct individuals over a network of cameras. It is an extremely challenging task since visual appearances of people can change significantly when viewed in different cameras. Many person re-identification methods offer distinct advantages over each other in terms of robustness to lighting, scale and pose variations. Keeping this consideration in mind, this paper proposes an effective new person reidentification model which incorporates several recent state-of-the-art feature extraction methodologies such as GOG, WHOS and LOMO features into a single framework. Effectiveness of each feature type is estimated and optimal weights for the similarity measurements are assigned through a multiple metric learning method. The proposed re-identification approach is then tested on multiple benchmark person re-identification datasets where it outperforms many other state-of-the-art methodologies.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"10 5 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":"129128262","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.2017.8310151
Miguel Bordallo López, Carlos R. del-Blanco, N. García
Fatigue has adverse effects in both physical and cognitive abilities. Hence, automatically detecting exercise-induced fatigue is of importance, especially in order to assist in the planning of effort and resting during exercise sessions. Thermal imaging and facial analysis provide a mean to detect changes in the human body unobtrusively and in variant conditions of pose and illumination. In this context, this paper proposes the automatic detection of exercise-induced fatigue using thermal cameras and facial images, analyzing them using deep convolutional neural networks. Our results indicate that classification of fatigued individuals is possible, obtaining an accuracy that reaches over 80% when utilizing single thermal images.
{"title":"Detecting exercise-induced fatigue using thermal imaging and deep learning","authors":"Miguel Bordallo López, Carlos R. del-Blanco, N. García","doi":"10.1109/IPTA.2017.8310151","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310151","url":null,"abstract":"Fatigue has adverse effects in both physical and cognitive abilities. Hence, automatically detecting exercise-induced fatigue is of importance, especially in order to assist in the planning of effort and resting during exercise sessions. Thermal imaging and facial analysis provide a mean to detect changes in the human body unobtrusively and in variant conditions of pose and illumination. In this context, this paper proposes the automatic detection of exercise-induced fatigue using thermal cameras and facial images, analyzing them using deep convolutional neural networks. Our results indicate that classification of fatigued individuals is possible, obtaining an accuracy that reaches over 80% when utilizing single thermal images.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"19 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":"125728046","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.2017.8310100
Rafael M. O. Cruz, Hiba H. Zakane, R. Sabourin, George D. C. Cavalcanti
Multiple classifier systems focus on the combination of classifiers to obtain better performance than a single robust one. These systems unfold three major phases: pool generation, selection and integration. One of the most promising MCS approaches is Dynamic Selection (DS), which relies on finding the most competent classifier or ensemble of classifiers to predict each test sample. The majority of the DS techniques are based on the K-Nearest Neighbors (K-NN) definition, and the quality of the neighborhood has a huge impact on the performance of DS methods. In this paper, we perform an analysis comparing the classification results of DS techniques and the K-NN classifier under different conditions. Experiments are performed on 18 state-of-the-art DS techniques over 30 classification datasets and results show that DS methods present a significant boost in classification accuracy even though they use the same neighborhood as the K-NN. The reasons behind the outperformance of DS techniques over the K-NN classifier reside in the fact that DS techniques can deal with samples with a high degree of instance hardness (samples that are located close to the decision border) as opposed to the K-NN. In this paper, not only we explain why DS techniques achieve higher classification performance than the K-NN but also when DS should be used.
{"title":"Dynamic ensemble selection VS K-NN: Why and when dynamic selection obtains higher classification performance?","authors":"Rafael M. O. Cruz, Hiba H. Zakane, R. Sabourin, George D. C. Cavalcanti","doi":"10.1109/IPTA.2017.8310100","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310100","url":null,"abstract":"Multiple classifier systems focus on the combination of classifiers to obtain better performance than a single robust one. These systems unfold three major phases: pool generation, selection and integration. One of the most promising MCS approaches is Dynamic Selection (DS), which relies on finding the most competent classifier or ensemble of classifiers to predict each test sample. The majority of the DS techniques are based on the K-Nearest Neighbors (K-NN) definition, and the quality of the neighborhood has a huge impact on the performance of DS methods. In this paper, we perform an analysis comparing the classification results of DS techniques and the K-NN classifier under different conditions. Experiments are performed on 18 state-of-the-art DS techniques over 30 classification datasets and results show that DS methods present a significant boost in classification accuracy even though they use the same neighborhood as the K-NN. The reasons behind the outperformance of DS techniques over the K-NN classifier reside in the fact that DS techniques can deal with samples with a high degree of instance hardness (samples that are located close to the decision border) as opposed to the K-NN. In this paper, not only we explain why DS techniques achieve higher classification performance than the K-NN but also when DS should be used.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"3 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":"134620546","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.2017.8310111
Akram Bayat, M. Pomplun
In this paper, we generate two computational models in order to estimate two dominant global properties (naturalness and openness) for representing a scene based on its global spatial structure. Naturalness and openness are two dominant perceptual properties within a multidimensional space in which semantically similar scenes (e.g., corridor and hallway) are assigned to nearby points. In this model space, the representation of a real-world scene is based on the overall shape of a scene but not on local object information. We introduce the use of a deep convolutional neural network for generating features that are well-suited for estimating the two global properties of a visual scene. The extracted features are integrated in an efficient way and fed into a linear support vector machine (SVM) to classify naturalness versus man-madeness and openness versus closedness. These two global properties (naturalness and openness) of an input image can be predicted from activations in the lowest layer of the convolutional neural network which has been trained for a scene recognition task. The consistent results of computational models in full and restricted spatial frequency ranges suggest that the representation of an image in the lowest layer of the deep scene CNN contains holistic information of the images as it leads to highest accuracy in modelling the global shape of the scene.
{"title":"Deriving high-level scene descriptions from deep scene CNN features","authors":"Akram Bayat, M. Pomplun","doi":"10.1109/IPTA.2017.8310111","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310111","url":null,"abstract":"In this paper, we generate two computational models in order to estimate two dominant global properties (naturalness and openness) for representing a scene based on its global spatial structure. Naturalness and openness are two dominant perceptual properties within a multidimensional space in which semantically similar scenes (e.g., corridor and hallway) are assigned to nearby points. In this model space, the representation of a real-world scene is based on the overall shape of a scene but not on local object information. We introduce the use of a deep convolutional neural network for generating features that are well-suited for estimating the two global properties of a visual scene. The extracted features are integrated in an efficient way and fed into a linear support vector machine (SVM) to classify naturalness versus man-madeness and openness versus closedness. These two global properties (naturalness and openness) of an input image can be predicted from activations in the lowest layer of the convolutional neural network which has been trained for a scene recognition task. The consistent results of computational models in full and restricted spatial frequency ranges suggest that the representation of an image in the lowest layer of the deep scene CNN contains holistic information of the images as it leads to highest accuracy in modelling the global shape of the scene.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"40 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":"116969890","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.2017.8310115
Patrick Thiam, F. Schwenker
In this work, an assessment of several fusion architectures is undertaken within the scope of the development of a pain intensity classification system. The assessment is based on the recently recorded SenseEmotion Database [1], which consists of several individuals subjected to three gradually increasing levels of pain intensity, induced through temperature elevation (heat stimulation) under controlled conditions. Several modalities, including audio, video, respiration, electrocardiography, electromyography and electrodermal activity, were synchronously recorded during the experiments. A broad spectrum of descriptors is extracted from each of the involved modalities, followed by an assessment of the combination of the extracted descriptors through several fusion architectures. Experimental validation suggests that the choice of an appropriate fusion architecture, which is able to significantly improve over the performance of the best single modality, mainly depends on the amount of data available for the training of the classification architecture.
{"title":"Multi-modal data fusion for pain intensity assessment and classification","authors":"Patrick Thiam, F. Schwenker","doi":"10.1109/IPTA.2017.8310115","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310115","url":null,"abstract":"In this work, an assessment of several fusion architectures is undertaken within the scope of the development of a pain intensity classification system. The assessment is based on the recently recorded SenseEmotion Database [1], which consists of several individuals subjected to three gradually increasing levels of pain intensity, induced through temperature elevation (heat stimulation) under controlled conditions. Several modalities, including audio, video, respiration, electrocardiography, electromyography and electrodermal activity, were synchronously recorded during the experiments. A broad spectrum of descriptors is extracted from each of the involved modalities, followed by an assessment of the combination of the extracted descriptors through several fusion architectures. Experimental validation suggests that the choice of an appropriate fusion architecture, which is able to significantly improve over the performance of the best single modality, mainly depends on the amount of data available for the training of the classification architecture.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"8 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":"123201315","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.2017.8310152
Reza Moradi Rad, Parvaneh Saeedi, J. Au, J. Havelock
Accurate identification of different components of a developing human embryo play crucial roles in assessing the quality of such embryo. One of the most important components of a day-5 human embryo is Inner Cell Mass (ICM). ICM is a part of an embryo that will eventually develop into a fetus. In this paper, an automatic coarse-to-fine texture based approach presented to identify regions of an embryo corresponding to the ICM. First, blastocyst area corresponding to the textured regions is recognized using Gabor and DCT features. Next, two ICM localization approaches are introduced to identify a rough estimate of the ICM location. Finally, the boundaries of the ICM region is finalized using a region based level-set. Experimental results on a data set of 220 day-5 human embryo images confirm that the proposed method is capable of identifying ICM with average Precision, Recall, and Jaccard Index of 78.7%, 86.8%, and 70.3%, respectively.
{"title":"Coarse-to-fine texture analysis for inner cell mass identification in human blastocyst microscopic images","authors":"Reza Moradi Rad, Parvaneh Saeedi, J. Au, J. Havelock","doi":"10.1109/IPTA.2017.8310152","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310152","url":null,"abstract":"Accurate identification of different components of a developing human embryo play crucial roles in assessing the quality of such embryo. One of the most important components of a day-5 human embryo is Inner Cell Mass (ICM). ICM is a part of an embryo that will eventually develop into a fetus. In this paper, an automatic coarse-to-fine texture based approach presented to identify regions of an embryo corresponding to the ICM. First, blastocyst area corresponding to the textured regions is recognized using Gabor and DCT features. Next, two ICM localization approaches are introduced to identify a rough estimate of the ICM location. Finally, the boundaries of the ICM region is finalized using a region based level-set. Experimental results on a data set of 220 day-5 human embryo images confirm that the proposed method is capable of identifying ICM with average Precision, Recall, and Jaccard Index of 78.7%, 86.8%, and 70.3%, respectively.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"35 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":"121639748","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.2017.8310123
L. Kuncheva, Paria Yousefi, J. Almeida
Evaluation of keyframe video summaries is a notoriously difficult problem. So far, there is no consensus on guidelines, protocols, benchmarks and baseline models. This study contributes in three ways: (1) We propose a new baseline model for creating a keyframe summary, called Closest-to-Centroid, and show that it is a better contestant compared to the two most popular baselines: uniform sampling and choosing the mid-event frame. (2) We also propose a method for matching the visual appearance of keyframes, suitable for comparing summaries of egocentric videos and lifelogging photostreams. (3) We examine 24 image feature spaces (different descriptors) including colour, texture, shape, motion and a feature space extracted by a pre-trained convolutional neural network (CNN). Our results using the four egocentric videos in the UTE database favour low-level shape and colour feature spaces for use with CC.
{"title":"Comparing keyframe summaries of egocentric videos: Closest-to-centroid baseline","authors":"L. Kuncheva, Paria Yousefi, J. Almeida","doi":"10.1109/IPTA.2017.8310123","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310123","url":null,"abstract":"Evaluation of keyframe video summaries is a notoriously difficult problem. So far, there is no consensus on guidelines, protocols, benchmarks and baseline models. This study contributes in three ways: (1) We propose a new baseline model for creating a keyframe summary, called Closest-to-Centroid, and show that it is a better contestant compared to the two most popular baselines: uniform sampling and choosing the mid-event frame. (2) We also propose a method for matching the visual appearance of keyframes, suitable for comparing summaries of egocentric videos and lifelogging photostreams. (3) We examine 24 image feature spaces (different descriptors) including colour, texture, shape, motion and a feature space extracted by a pre-trained convolutional neural network (CNN). Our results using the four egocentric videos in the UTE database favour low-level shape and colour feature spaces for use with CC.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"9 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":"114164676","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.2017.8310129
Torsten Engler, Hans-Joachim Wünsche
In this paper we present the recursive estimation of static scenes with multiple stereo camera pairs. The estimation is based on a point cloud created from the disparities of the cameras. The focus lies on reducing erroneous measurements while obtaining a comparatively dense measurement in real time. While recursive scene estimation via stereo cameras has been presented several times before, the estimation has never been exploited in the measurement algorithm. We propose the usage of the current scene estimation in the disparity measurement to increase robustness, denseness and outlier rejection. A scene prior is created for each measurement using OpenGL taking occlusions, camera positions and existence probability into account. Additionally, multiple stereo pairs with different alignment provide distinct information. Each disparity measurement benefits from the complete scene knowledge the other stereo camera pairs provide. The creation of new points for the point cloud is based on a scaled version of the current scene and allows for simple trade-off between computational effort and point cloud denseness.
{"title":"Recursive 3D scene estimation with multiple camera pairs","authors":"Torsten Engler, Hans-Joachim Wünsche","doi":"10.1109/IPTA.2017.8310129","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310129","url":null,"abstract":"In this paper we present the recursive estimation of static scenes with multiple stereo camera pairs. The estimation is based on a point cloud created from the disparities of the cameras. The focus lies on reducing erroneous measurements while obtaining a comparatively dense measurement in real time. While recursive scene estimation via stereo cameras has been presented several times before, the estimation has never been exploited in the measurement algorithm. We propose the usage of the current scene estimation in the disparity measurement to increase robustness, denseness and outlier rejection. A scene prior is created for each measurement using OpenGL taking occlusions, camera positions and existence probability into account. Additionally, multiple stereo pairs with different alignment provide distinct information. Each disparity measurement benefits from the complete scene knowledge the other stereo camera pairs provide. The creation of new points for the point cloud is based on a scaled version of the current scene and allows for simple trade-off between computational effort and point cloud denseness.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"32 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":"114717798","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.2017.8310089
Jing Qin, Feng Liu, Shouyi Wang, J. Rosenberger
EEG serves as an essential tool for brain source localization due to its high temporal resolution. However, the inference of brain activities from the EEG data is, in general, a challenging ill-posed inverse problem. To better retrieve task related discriminative source patches from strong spontaneous background signals, we propose a novel EEG source imaging model based on spatial and temporal graph structures. In particular, graph fractional-order total variation (gFOTV) is used to enhance spatial smoothness, and the label information of brain state is enclosed in a temporal graph regularization term to guarantee intra-class consistency of estimated sources. The proposed model is efficiently solved by the alternating direction method of multipliers (ADMM). A two-stage algorithm is proposed as well to further improve the result. Numerical experiments have shown that our method localizes source extents more effectively than the benchmark methods.
{"title":"EEG source imaging based on spatial and temporal graph structures","authors":"Jing Qin, Feng Liu, Shouyi Wang, J. Rosenberger","doi":"10.1109/IPTA.2017.8310089","DOIUrl":"https://doi.org/10.1109/IPTA.2017.8310089","url":null,"abstract":"EEG serves as an essential tool for brain source localization due to its high temporal resolution. However, the inference of brain activities from the EEG data is, in general, a challenging ill-posed inverse problem. To better retrieve task related discriminative source patches from strong spontaneous background signals, we propose a novel EEG source imaging model based on spatial and temporal graph structures. In particular, graph fractional-order total variation (gFOTV) is used to enhance spatial smoothness, and the label information of brain state is enclosed in a temporal graph regularization term to guarantee intra-class consistency of estimated sources. The proposed model is efficiently solved by the alternating direction method of multipliers (ADMM). A two-stage algorithm is proposed as well to further improve the result. Numerical experiments have shown that our method localizes source extents more effectively than the benchmark methods.","PeriodicalId":316356,"journal":{"name":"2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA)","volume":"97 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":"124828747","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：