Pub Date : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120608
V. Sowmya, D. Govind, K. Soman
In general, the three main modules of color image classification systems are: color-to-grayscale image conversion, feature extraction and classification. The color-to-grayscale image conversion is the important pre-processing step which must incorporate the significant and discriminative contrast and structure information in the converted grayscale images as in the original color image. All the existing techniques for color-to-grayscale image conversion preserves the significant contrast and structure information in the converted grayscale images in different manners. Hence, the present work is to analyze the significant and discriminative contrast and structure information preserved in the converted grayscale images using two different decolorization techniques called rgb2gray and singular value decomposition based color-to-grayscale image conversion (SVD) applied in the color image classification systems using the three different proposed features. The three different features for color image classification systems are proposed based on the combination of the existing dense SIFT features and the contrast & structure content computed using color-to-gray structure similarity index (C2G-SSIM) metric.
{"title":"Significance of contrast and structure features for an improved color image classification system","authors":"V. Sowmya, D. Govind, K. Soman","doi":"10.1109/ICSIPA.2017.8120608","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120608","url":null,"abstract":"In general, the three main modules of color image classification systems are: color-to-grayscale image conversion, feature extraction and classification. The color-to-grayscale image conversion is the important pre-processing step which must incorporate the significant and discriminative contrast and structure information in the converted grayscale images as in the original color image. All the existing techniques for color-to-grayscale image conversion preserves the significant contrast and structure information in the converted grayscale images in different manners. Hence, the present work is to analyze the significant and discriminative contrast and structure information preserved in the converted grayscale images using two different decolorization techniques called rgb2gray and singular value decomposition based color-to-grayscale image conversion (SVD) applied in the color image classification systems using the three different proposed features. The three different features for color image classification systems are proposed based on the combination of the existing dense SIFT features and the contrast & structure content computed using color-to-gray structure similarity index (C2G-SSIM) metric.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127155850","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120636
Aliyu Nuhu Shuaibu, A. Malik, I. Faye
Learning and recognizing 3-dimension (3D) adaptive features are important for crowd scene understanding in video surveillance research. Deep learning architectures such as Convolutional Neural Networks (CNN) have recently shown much success in various computer vision applications. Existing approaches such as hand-crafted method and 2D-CNN architectures are widely used in adaptive feature representations on image data. However, learning dynamic and temporal features in 3D scale features in videos remains an open problem. In this study, we proposed a novel technique 3D-scale Convolutional Neural Network (3DS-CNN), based on the decomposition of 3D feature maps into 2D spatio and 2D temporal feature representations. Extensive experiments on hundreds of video scene were demonstrated on publicly available crowd datasets. Quantitative and qualitative evaluations indicate that the proposed model display superior performance when compared to baseline approaches. The mean average precision of 95.30% was recorded on WWW crowd dataset.
{"title":"Adaptive feature learning CNN for behavior recognition in crowd scene","authors":"Aliyu Nuhu Shuaibu, A. Malik, I. Faye","doi":"10.1109/ICSIPA.2017.8120636","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120636","url":null,"abstract":"Learning and recognizing 3-dimension (3D) adaptive features are important for crowd scene understanding in video surveillance research. Deep learning architectures such as Convolutional Neural Networks (CNN) have recently shown much success in various computer vision applications. Existing approaches such as hand-crafted method and 2D-CNN architectures are widely used in adaptive feature representations on image data. However, learning dynamic and temporal features in 3D scale features in videos remains an open problem. In this study, we proposed a novel technique 3D-scale Convolutional Neural Network (3DS-CNN), based on the decomposition of 3D feature maps into 2D spatio and 2D temporal feature representations. Extensive experiments on hundreds of video scene were demonstrated on publicly available crowd datasets. Quantitative and qualitative evaluations indicate that the proposed model display superior performance when compared to baseline approaches. The mean average precision of 95.30% was recorded on WWW crowd dataset.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130694361","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120644
Devang S. Pandya, M. Zaveri
Due to rapid growth of digital contents, there has been an increasing demand of summarized videos to save time and other network resources. Automated soccer video analysis is a challenging due to involvement of more actors and rapid movements of players and camera. It is necessary first to detect various events of the video precisely before analyzing and labeling them. This paper proposes frame based approach for the automatic demarcation of events of the soccer video. However this task is very challenging due to variety of soccer leagues, various illumination and ground conditions. To overcome such issues we propose method which is invariant to such conditions and can successfully demarcate the soccer events. We exploit optical flow techniques to measure the motion. We introduce change in optical flow to extract the behavior of an event over the video span. Later, adaptive threshold is computed based on change in optical flow. We conducted number of simulations with variety of videos to validate the method. Proposed method achieves nearly 90% of accuracy and found robust in spite of illumination variation.
{"title":"Frame based approach for automatic event boundary detection of soccer video using optical flow","authors":"Devang S. Pandya, M. Zaveri","doi":"10.1109/ICSIPA.2017.8120644","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120644","url":null,"abstract":"Due to rapid growth of digital contents, there has been an increasing demand of summarized videos to save time and other network resources. Automated soccer video analysis is a challenging due to involvement of more actors and rapid movements of players and camera. It is necessary first to detect various events of the video precisely before analyzing and labeling them. This paper proposes frame based approach for the automatic demarcation of events of the soccer video. However this task is very challenging due to variety of soccer leagues, various illumination and ground conditions. To overcome such issues we propose method which is invariant to such conditions and can successfully demarcate the soccer events. We exploit optical flow techniques to measure the motion. We introduce change in optical flow to extract the behavior of an event over the video span. Later, adaptive threshold is computed based on change in optical flow. We conducted number of simulations with variety of videos to validate the method. Proposed method achieves nearly 90% of accuracy and found robust in spite of illumination variation.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126018869","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120662
G. Chan, M. Awais, S. A. A. Shah, T. Tang, Cheng-Kai Lu, F. Mériaudeau
Diabetic Macular Edema (DME) is a common eye disease that causes irreversible vision loss for diabetic patients, if left untreated. Thus, early diagnosis of DME could help in early treatment and prevent blindness. This paper aims to create a framework based on deep learning for DME recognition on Spectral Domain Optical Coherence Tomography (SD-OCT) images through transfer learning. First, images are pre-processed: denoised using Block-Matching and 3-Dimension (BM3D) filtering and cropped through image boundary extraction. Later, features are extracted using CNN of AlexNet and finally images are classified using SVM classifier. The results are evaluated using 8-fold cross-validation. The experiments show that denoised and cropped images lead to better classification performances, exceeding previous other recent published works of 96% accuracy.
{"title":"Transfer learning for Diabetic Macular Edema (DME) detection on Optical Coherence Tomography (OCT) images","authors":"G. Chan, M. Awais, S. A. A. Shah, T. Tang, Cheng-Kai Lu, F. Mériaudeau","doi":"10.1109/ICSIPA.2017.8120662","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120662","url":null,"abstract":"Diabetic Macular Edema (DME) is a common eye disease that causes irreversible vision loss for diabetic patients, if left untreated. Thus, early diagnosis of DME could help in early treatment and prevent blindness. This paper aims to create a framework based on deep learning for DME recognition on Spectral Domain Optical Coherence Tomography (SD-OCT) images through transfer learning. First, images are pre-processed: denoised using Block-Matching and 3-Dimension (BM3D) filtering and cropped through image boundary extraction. Later, features are extracted using CNN of AlexNet and finally images are classified using SVM classifier. The results are evaluated using 8-fold cross-validation. The experiments show that denoised and cropped images lead to better classification performances, exceeding previous other recent published works of 96% accuracy.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"90 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115865626","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120656
A. Eltholth, M. Gaber, M. Fouda, H. Mansour
Filter Bank Multi-Carrier (FBMC), has been proposed to solve the problems of spectral efficiency and adjacent channel leakage power in OFDM. This solution has added implications for the Peak-to-Average Power Ratio (PAPR) problem. This paper aims to reduce PAPR in FBMC through convolutional constellation mapping. The idea is to give a different constellation point to the symbol if successive values of the same symbol occur in the state space, thus reducing the chances of coherent addition of subcarriers without increasing the constellation points or even the need for iterative mapping. Computer-based simulations show the superior performance of the proposed Convolutional Constellation Mapping (CCM) in reducing the PAPR of FBMC.
{"title":"Convolutional constellation mapping technique for PAPR reduction in filter bank multi-carrier","authors":"A. Eltholth, M. Gaber, M. Fouda, H. Mansour","doi":"10.1109/ICSIPA.2017.8120656","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120656","url":null,"abstract":"Filter Bank Multi-Carrier (FBMC), has been proposed to solve the problems of spectral efficiency and adjacent channel leakage power in OFDM. This solution has added implications for the Peak-to-Average Power Ratio (PAPR) problem. This paper aims to reduce PAPR in FBMC through convolutional constellation mapping. The idea is to give a different constellation point to the symbol if successive values of the same symbol occur in the state space, thus reducing the chances of coherent addition of subcarriers without increasing the constellation points or even the need for iterative mapping. Computer-based simulations show the superior performance of the proposed Convolutional Constellation Mapping (CCM) in reducing the PAPR of FBMC.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133929816","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120581
Anthony Ngozichukwuka Uwaechia, N. Mahyuddin
In this paper, the problem of the deterministic pilot allocation for sparse channel estimation in Orthogonal Frequency Division Multiplexing (OFDM) system is investigated. This method is based on mutual coherence minimization of the measurement matrix associated with the OFDM system pilot subcarriers. It is known that if the set of pilot pattern is a Cyclic Difference Set (CDS), the mutual coherence of the measurement matrix is minimized. However, CDS in most practical OFDM system is not available. Few research efforts have tackled the problem of pilot allocation by proposing methods that lead to suboptimal solutions in order to ignore the computationally complex exhaustive search method. This contribution, however, proposes two pilot allocation design schemes for the construction of deterministic partial Fourier matrices satisfying the Restricted Isometry Property (RIP) namely, the Generic Random Search (GRS) and Progressive Search (PS) based on bounding the mutual coherence between different columns of the measurement matrix. Simulation results show that the two proposed pilot allocation design schemes are effective and offer a better channel estimation performance in terms of MSE when compared to former pilot allocation design methods.
{"title":"New pilot allocation design schemes for sparse channel estimation in OFDM system","authors":"Anthony Ngozichukwuka Uwaechia, N. Mahyuddin","doi":"10.1109/ICSIPA.2017.8120581","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120581","url":null,"abstract":"In this paper, the problem of the deterministic pilot allocation for sparse channel estimation in Orthogonal Frequency Division Multiplexing (OFDM) system is investigated. This method is based on mutual coherence minimization of the measurement matrix associated with the OFDM system pilot subcarriers. It is known that if the set of pilot pattern is a Cyclic Difference Set (CDS), the mutual coherence of the measurement matrix is minimized. However, CDS in most practical OFDM system is not available. Few research efforts have tackled the problem of pilot allocation by proposing methods that lead to suboptimal solutions in order to ignore the computationally complex exhaustive search method. This contribution, however, proposes two pilot allocation design schemes for the construction of deterministic partial Fourier matrices satisfying the Restricted Isometry Property (RIP) namely, the Generic Random Search (GRS) and Progressive Search (PS) based on bounding the mutual coherence between different columns of the measurement matrix. Simulation results show that the two proposed pilot allocation design schemes are effective and offer a better channel estimation performance in terms of MSE when compared to former pilot allocation design methods.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134276449","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120638
R. Lim, Clarence Weihan Cheong, John See, I. Tan, L. Wong, Huai-Qian Khor
Car park video surveillance systems present a huge volume of data that can be beneficial for video analytics and data analysis. We present a vehicle state tracking method for long term video surveillance with the goal of obtaining trajectories and vehicle states of various car park users. However, this is a challenging task in outdoor scenarios due to non-optimal camera viewing angle compounded by ever-changing illumination & weather conditions. To address these challenges, we propose a parking state machine that tracks the vehicle state in a large outdoor car park area. The proposed method was tested on 10 hours of continuous video data with various illumination and environmental conditions. Owing to the imbalanced distribution of parking states, we report the precision, recall and F1 scores to determine the overall performance of the system. Our approach proves to be fairly accurate, fast and robust against severe scene variations.
{"title":"On vehicle state tracking for long-term carpark video surveillance","authors":"R. Lim, Clarence Weihan Cheong, John See, I. Tan, L. Wong, Huai-Qian Khor","doi":"10.1109/ICSIPA.2017.8120638","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120638","url":null,"abstract":"Car park video surveillance systems present a huge volume of data that can be beneficial for video analytics and data analysis. We present a vehicle state tracking method for long term video surveillance with the goal of obtaining trajectories and vehicle states of various car park users. However, this is a challenging task in outdoor scenarios due to non-optimal camera viewing angle compounded by ever-changing illumination & weather conditions. To address these challenges, we propose a parking state machine that tracks the vehicle state in a large outdoor car park area. The proposed method was tested on 10 hours of continuous video data with various illumination and environmental conditions. Owing to the imbalanced distribution of parking states, we report the precision, recall and F1 scores to determine the overall performance of the system. Our approach proves to be fairly accurate, fast and robust against severe scene variations.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115832987","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120663
Rahul Hooda, S. Sofat, Simranpreet Kaur, Ajay Mittal, F. Mériaudeau
Tuberculosis (TB) is a major health threat in the developing countries. Many patients die every year due to lack of treatment and error in diagnosis. Developing a computer-aided diagnosis (CAD) system for TB detection can help in early diagnosis and containing the disease. Most of the current CAD systems use handcrafted features, however, lately there is a shift towards deep-learning-based automatic feature extractors. In this paper, we present a potential method for tuberculosis detection using deep-learning which classifies CXR images into two categories, that is, normal and abnormal. We have used CNN architecture with 7 convolutional layers and 3 fully connected layers. The performance of three different optimizers has been compared. Out of these, Adam optimizer with an overall accuracy of 94.73% and validation accuracy of 82.09% performed best amongst them. All the results are obtained using Montgomery and Shenzhen datasets which are available in public domain.
{"title":"Deep-learning: A potential method for tuberculosis detection using chest radiography","authors":"Rahul Hooda, S. Sofat, Simranpreet Kaur, Ajay Mittal, F. Mériaudeau","doi":"10.1109/ICSIPA.2017.8120663","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120663","url":null,"abstract":"Tuberculosis (TB) is a major health threat in the developing countries. Many patients die every year due to lack of treatment and error in diagnosis. Developing a computer-aided diagnosis (CAD) system for TB detection can help in early diagnosis and containing the disease. Most of the current CAD systems use handcrafted features, however, lately there is a shift towards deep-learning-based automatic feature extractors. In this paper, we present a potential method for tuberculosis detection using deep-learning which classifies CXR images into two categories, that is, normal and abnormal. We have used CNN architecture with 7 convolutional layers and 3 fully connected layers. The performance of three different optimizers has been compared. Out of these, Adam optimizer with an overall accuracy of 94.73% and validation accuracy of 82.09% performed best amongst them. All the results are obtained using Montgomery and Shenzhen datasets which are available in public domain.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134080171","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120652
A. Kale, S. Sonavane
The crucial objective of this paper is to design a hybrid model of the genetic algorithm for fuzzy extreme learning machine classifier (GA-FELM), which selects an optimal feature subset by using the multilevel parameter optimization technique. Feature subset selection is an important task in pattern classification and knowledge discovery problems. The generalization performance of the system is not only depending on optimal features but also dependent upon the classifier (learning algorithm). Therefore, it is an important task to select a fast and efficient classifier. Research efforts have affirmed that extreme learning machine (ELM) has superior and accurate classification ability. However, ELM is failed to handle the uncertain data. One of the alternative solutions is fuzzy-ELM, which combines the advantages of fuzzy logic and ELM. GA-FELM is able to handle curse of dimensionality problem, optimization problem and weighted classification problem with maximizing classification accuracy by minimizing the number of features. In order to validate the efficiency of GA-FELM, the comparative performance is evaluated by using three different approaches viz. 1. ELM and GA-ELM 2. GA-ELM and GA-FELM 3. GA-FELM and GA-existing classifier. The result analysis shows that classification accuracy is improved with 9% while reducing 62% features.
{"title":"Optimal feature subset selection for fuzzy extreme learning machine using genetic algorithm with multilevel parameter optimization","authors":"A. Kale, S. Sonavane","doi":"10.1109/ICSIPA.2017.8120652","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120652","url":null,"abstract":"The crucial objective of this paper is to design a hybrid model of the genetic algorithm for fuzzy extreme learning machine classifier (GA-FELM), which selects an optimal feature subset by using the multilevel parameter optimization technique. Feature subset selection is an important task in pattern classification and knowledge discovery problems. The generalization performance of the system is not only depending on optimal features but also dependent upon the classifier (learning algorithm). Therefore, it is an important task to select a fast and efficient classifier. Research efforts have affirmed that extreme learning machine (ELM) has superior and accurate classification ability. However, ELM is failed to handle the uncertain data. One of the alternative solutions is fuzzy-ELM, which combines the advantages of fuzzy logic and ELM. GA-FELM is able to handle curse of dimensionality problem, optimization problem and weighted classification problem with maximizing classification accuracy by minimizing the number of features. In order to validate the efficiency of GA-FELM, the comparative performance is evaluated by using three different approaches viz. 1. ELM and GA-ELM 2. GA-ELM and GA-FELM 3. GA-FELM and GA-existing classifier. The result analysis shows that classification accuracy is improved with 9% while reducing 62% features.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132384930","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 : 2017-09-01DOI: 10.1109/ICSIPA.2017.8120643
F. Goodarzi, F. Rokhani, M. Saripan, M. Marhaban
The problem of recognizing and discriminating mixed emotions in multi view faces using a web camera is discussed in this paper. Based on the literature, there are mainly seven basic emotions that humans can express and understand. However, in some faces in databases, there are characteristics of two or more of this basic emotions. The two databases of BU3DFE and UPM3DFE were tested for mixed emotion accuracy using the proposed multi view face emotion recognition method. The results show an improvement over existing works in mixed emotions recognition.
{"title":"Mixed emotions in multi view face emotion recognition","authors":"F. Goodarzi, F. Rokhani, M. Saripan, M. Marhaban","doi":"10.1109/ICSIPA.2017.8120643","DOIUrl":"https://doi.org/10.1109/ICSIPA.2017.8120643","url":null,"abstract":"The problem of recognizing and discriminating mixed emotions in multi view faces using a web camera is discussed in this paper. Based on the literature, there are mainly seven basic emotions that humans can express and understand. However, in some faces in databases, there are characteristics of two or more of this basic emotions. The two databases of BU3DFE and UPM3DFE were tested for mixed emotion accuracy using the proposed multi view face emotion recognition method. The results show an improvement over existing works in mixed emotions recognition.","PeriodicalId":268112,"journal":{"name":"2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133710779","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：