Pub Date : 2009-12-04DOI: 10.1109/CCPR.2009.5344050
Xi-Huan Yang, H. Xue, Songcan Chen
Local Ridge Regression Classifier (LRR) is an effective local face recognition method. It suppresses the influence of local changes by setting a voting RR classifier for each image region, thus has partial robustness to local changes caused by lighting, occlusions and poses. LRR uses the concatenated vector of a sub-image as its input feature, such a feature is still not sufficient to represent an image, thus leading to possibly imprecise voting and limited increase in recognition rate. In order to boost its recognition rate, we first develop a novel classifier GLRR which combines LRR classifier and Gabor-LBP features which can improve the feature representation greatly. Experiments on AR database demonstrate that GLRR is superior to LRR and other local methods such as Aw-SpPCA and SpCCA. When just fewer classifiers can be available and some occlusion regions exist, majority-voting recognition rate will still be imprecise. To remedy this, in this paper, we add an occlusion detection step before classification using GLRR for which we call it S-GLRR. In this way, we can purposely shield locally-occluded regions using the detection step, thus get better performance for face recognition. Experiments show that S-GLRR achieves better recognition rate than GLRR, especially when only a few sub-classifiers are provided.
{"title":"Image Region Selection Based GLRR for Face Recognition","authors":"Xi-Huan Yang, H. Xue, Songcan Chen","doi":"10.1109/CCPR.2009.5344050","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344050","url":null,"abstract":"Local Ridge Regression Classifier (LRR) is an effective local face recognition method. It suppresses the influence of local changes by setting a voting RR classifier for each image region, thus has partial robustness to local changes caused by lighting, occlusions and poses. LRR uses the concatenated vector of a sub-image as its input feature, such a feature is still not sufficient to represent an image, thus leading to possibly imprecise voting and limited increase in recognition rate. In order to boost its recognition rate, we first develop a novel classifier GLRR which combines LRR classifier and Gabor-LBP features which can improve the feature representation greatly. Experiments on AR database demonstrate that GLRR is superior to LRR and other local methods such as Aw-SpPCA and SpCCA. When just fewer classifiers can be available and some occlusion regions exist, majority-voting recognition rate will still be imprecise. To remedy this, in this paper, we add an occlusion detection step before classification using GLRR for which we call it S-GLRR. In this way, we can purposely shield locally-occluded regions using the detection step, thus get better performance for face recognition. Experiments show that S-GLRR achieves better recognition rate than GLRR, especially when only a few sub-classifiers are provided.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128807261","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5344086
N.Muru gan, F. Zheng
In the complex stomach epidermis tumor cells, the traditional segmentation algorithms such as the K-means clustering algorithm and the simple threshold segmentation algorithm are unable to get satisfactory results. The relaxation iterative segmentation algorithm can segment the cell clearly, but it wastes a lot of time and the execution efficiency is very low. In this paper the authors propose a new segmentation algorithm based on the maximization of Mutual information in effective information, in which to find the optimal threshold values to segment the stomach epidermis tumor cells.
{"title":"Stomach Epidermis Tumor Cell Segmentation Based on the Maximization of Mutual Information in Effective Information","authors":"N.Muru gan, F. Zheng","doi":"10.1109/CCPR.2009.5344086","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344086","url":null,"abstract":"In the complex stomach epidermis tumor cells, the traditional segmentation algorithms such as the K-means clustering algorithm and the simple threshold segmentation algorithm are unable to get satisfactory results. The relaxation iterative segmentation algorithm can segment the cell clearly, but it wastes a lot of time and the execution efficiency is very low. In this paper the authors propose a new segmentation algorithm based on the maximization of Mutual information in effective information, in which to find the optimal threshold values to segment the stomach epidermis tumor cells.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"202 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123259324","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5343963
Fengxi Song, Yong Xu, David Zhang, Tianwei Liu
This paper presented a novel subspace-based facial discriminant feature extraction method, i.e. Orthogonalized Direct Linear Discriminant Analysis (OD-LDA), whose discriminant vectors could be obtained by performing Gram-Schmidt orthogonal procedure on a set of discriminant vectors of D-LDA. Experimental studies conducted on ORL, FERET, Yale, and AR face image databases showed that OD-LDA could compete with prevailing subspace-based facial discriminant feature extraction methods such as Fisherfaces, N-LDA D-LDA, Uncorrelated LDA, Parameterized D-LDA, K-L expansion based the between-class scatter matrix, and Orthogonal Complimentary Space Method in terms of recognition rate.
{"title":"A Novel Subspace-Based Facial Discriminant Feature Extraction Method","authors":"Fengxi Song, Yong Xu, David Zhang, Tianwei Liu","doi":"10.1109/CCPR.2009.5343963","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5343963","url":null,"abstract":"This paper presented a novel subspace-based facial discriminant feature extraction method, i.e. Orthogonalized Direct Linear Discriminant Analysis (OD-LDA), whose discriminant vectors could be obtained by performing Gram-Schmidt orthogonal procedure on a set of discriminant vectors of D-LDA. Experimental studies conducted on ORL, FERET, Yale, and AR face image databases showed that OD-LDA could compete with prevailing subspace-based facial discriminant feature extraction methods such as Fisherfaces, N-LDA D-LDA, Uncorrelated LDA, Parameterized D-LDA, K-L expansion based the between-class scatter matrix, and Orthogonal Complimentary Space Method in terms of recognition rate.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116134891","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5344093
Shijin Li, Jiali Zhu, Xiangtao Gao, Jian Tao
Soil erosion is one of the most typical natural disasters in China. However, due to the limitation of current technology, the investigation of soil erosion through remote sensing images is currently by human beings manually which depends on human interpretation and interactive selection. The work burden is so heavy that errors are usually inevitably unavoidable. This paper proposes the technique of content-based image retrieval to tackle this problem. Due to the large amount of computation in co-training retrieval based on multiple classifier systems, and for the purpose of improving efficiency, an improved approach using co-training in two classifier systems is proposed in this paper. Prior to retrieving, we firstly select the optimal color feature and texture feature respectively, and then use the corresponding color classifier and texture classifier for co-training. By this approach, the time of co-training is reduced greatly, meanwhile, the selected optimal features can represent color and texture features better for remote sensing image, resulting in better retrieval accuracy. Experimental results show that the improved approach using co-training in two classifier systems needs less amount of computation and less retrieval time, while it can lead to better retrieval results.
{"title":"Soil Erosion Remote Sensing Image Retrieval Based on Semi-Supervised Learning","authors":"Shijin Li, Jiali Zhu, Xiangtao Gao, Jian Tao","doi":"10.1109/CCPR.2009.5344093","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344093","url":null,"abstract":"Soil erosion is one of the most typical natural disasters in China. However, due to the limitation of current technology, the investigation of soil erosion through remote sensing images is currently by human beings manually which depends on human interpretation and interactive selection. The work burden is so heavy that errors are usually inevitably unavoidable. This paper proposes the technique of content-based image retrieval to tackle this problem. Due to the large amount of computation in co-training retrieval based on multiple classifier systems, and for the purpose of improving efficiency, an improved approach using co-training in two classifier systems is proposed in this paper. Prior to retrieving, we firstly select the optimal color feature and texture feature respectively, and then use the corresponding color classifier and texture classifier for co-training. By this approach, the time of co-training is reduced greatly, meanwhile, the selected optimal features can represent color and texture features better for remote sensing image, resulting in better retrieval accuracy. Experimental results show that the improved approach using co-training in two classifier systems needs less amount of computation and less retrieval time, while it can lead to better retrieval results.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122901312","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5344054
Chuan-Xian Ren, D. Dai
Sparse representation for robust face recognition is a novel concept in the pattern analysis and machine learning community. Through the l1-minimization model, representing a test sample as the sparse combination of the training dictionary can effectively achieve facial images classification. However, when the number of training samples is relatively small, it is insufficient to give the test sample a sparse representation so that the recognition performance degenerates seriously. In this paper, we present a novel approach that employs the Elastic Net regularized regression model. Experimental results on several databases show that the proposed strategy improves the recognition accuracy.
{"title":"Sparse Representation by Adding Noisy Duplicates for Enhanced Face Recognition: An Elastic Net Regularization Approach","authors":"Chuan-Xian Ren, D. Dai","doi":"10.1109/CCPR.2009.5344054","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344054","url":null,"abstract":"Sparse representation for robust face recognition is a novel concept in the pattern analysis and machine learning community. Through the l1-minimization model, representing a test sample as the sparse combination of the training dictionary can effectively achieve facial images classification. However, when the number of training samples is relatively small, it is insufficient to give the test sample a sparse representation so that the recognition performance degenerates seriously. In this paper, we present a novel approach that employs the Elastic Net regularized regression model. Experimental results on several databases show that the proposed strategy improves the recognition accuracy.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131115475","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5344020
Yongming Huang, Guobao Zhang, Xiaoli Xu
In this paper, we present an emotion recognition system using the stacked generalization ensemble neural network for special human affective state in the speech signal. 450 short emotional sentences with different contents from 3 speakers were collected as experiment materials. The features relevant with energy, speech rate, pitch and formant are extracted from speech signals. Stacked Generalization Ensemble Neural Networks are used as the classifier for 5 emotions including anger, calmness, happiness, sadness and boredom. First, compared with the traditional BP network or wavelet neural network, the results of experiments show that the Stacked Generalization Ensemble Neural Network has faster convergence speed and higher recognition rate. Second, after discussing the advantage and disadvantage between different ensemble Neural Networks, suitable decision will be made for Robot Pet.
{"title":"Speech Emotion Recognition Research Based on the Stacked Generalization Ensemble Neural Network for Robot Pet","authors":"Yongming Huang, Guobao Zhang, Xiaoli Xu","doi":"10.1109/CCPR.2009.5344020","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344020","url":null,"abstract":"In this paper, we present an emotion recognition system using the stacked generalization ensemble neural network for special human affective state in the speech signal. 450 short emotional sentences with different contents from 3 speakers were collected as experiment materials. The features relevant with energy, speech rate, pitch and formant are extracted from speech signals. Stacked Generalization Ensemble Neural Networks are used as the classifier for 5 emotions including anger, calmness, happiness, sadness and boredom. First, compared with the traditional BP network or wavelet neural network, the results of experiments show that the Stacked Generalization Ensemble Neural Network has faster convergence speed and higher recognition rate. Second, after discussing the advantage and disadvantage between different ensemble Neural Networks, suitable decision will be made for Robot Pet.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"120 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115167713","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5344146
Qianru Li, H. Wang, J. Yang
Collaborative filtering technology has been successfully used in personalized recommendation systems. With the development of E-commerce, as well as the increase in the number of users and items, the users score data sparsity and the dimension disaster problems have been caused which leads to sharp decline in the quality of their recommend. A calculation of pattern similarity was proposed based on the users pattern similarity to direct at the sparsity and dimension disadvantage of high-dimensional data. Clustering were produced by subspace clustering algorithm based on users pattern similarity, and collaborative filtering algorithm was improved by calculating of model similarity which brings recommendation to users. The experimental result shows that algorithm increase the response speed of the system,at the mean time the recommendation quality has been improved a lot.
{"title":"Collaborative Filtering in Personalized Recommendation Based on Users Pattern Subspace Clustering","authors":"Qianru Li, H. Wang, J. Yang","doi":"10.1109/CCPR.2009.5344146","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344146","url":null,"abstract":"Collaborative filtering technology has been successfully used in personalized recommendation systems. With the development of E-commerce, as well as the increase in the number of users and items, the users score data sparsity and the dimension disaster problems have been caused which leads to sharp decline in the quality of their recommend. A calculation of pattern similarity was proposed based on the users pattern similarity to direct at the sparsity and dimension disadvantage of high-dimensional data. Clustering were produced by subspace clustering algorithm based on users pattern similarity, and collaborative filtering algorithm was improved by calculating of model similarity which brings recommendation to users. The experimental result shows that algorithm increase the response speed of the system,at the mean time the recommendation quality has been improved a lot.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127739085","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}
As one of important information component in multimedia, audio enriches information perception and acquisition. Analyses and extractions of audio features are the base of audio classification. It's important to extract audio features effectively for content-based audio retrieval. In this paper, based on the theory of rough set, audio features are reduced and a lower-dimension feature set can be obtained with more effective. Then the feature set is applied in the general model for audio classification. Experiments show that this method is effective.
{"title":"A Method Based on General Model and Rough Set for Audio Classification","authors":"Xin He, Ying-Chun Shi, Fuming Peng, Xianzhong Zhou","doi":"10.1109/CCPR.2009.5344044","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344044","url":null,"abstract":"As one of important information component in multimedia, audio enriches information perception and acquisition. Analyses and extractions of audio features are the base of audio classification. It's important to extract audio features effectively for content-based audio retrieval. In this paper, based on the theory of rough set, audio features are reduced and a lower-dimension feature set can be obtained with more effective. Then the feature set is applied in the general model for audio classification. Experiments show that this method is effective.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115100454","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5344142
Haiyang Hua, Huaici Zhao
Many machine learning algorithms can be applied only to data described by categorical attributes. So discretizatioti of continuous attributes is one of the important steps in preprocessing of extracting knowledge. Traditional discretization algorithms based on clustering need a pre-determined clustering number k, also typically are applied in an unsupervised learning framework. This paper describes such an algorithm, called SX-means (Supervised X-means), which is a new algorithm of supervised discretization of continuous attributes on clustering. The algorithm modifies clusters with knowledge of the class distribution dynamically. And this procedure can not stop until the proper k is found. For the number of clusters k is not pre-determined by the user and class distribution is applied, the random of result is decreased greatly. Experimental evaluation of several discretization algorithms on six artificial data sets show that the proposed algorithm is more efficient and can generate a better discretization schema. Comparing the output of C4.5, resulting tree is smaller, less classification rules, and high accuracy of classification.
{"title":"A Discretization Algorithm of Continuous Attributes Based on Supervised Clustering","authors":"Haiyang Hua, Huaici Zhao","doi":"10.1109/CCPR.2009.5344142","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5344142","url":null,"abstract":"Many machine learning algorithms can be applied only to data described by categorical attributes. So discretizatioti of continuous attributes is one of the important steps in preprocessing of extracting knowledge. Traditional discretization algorithms based on clustering need a pre-determined clustering number k, also typically are applied in an unsupervised learning framework. This paper describes such an algorithm, called SX-means (Supervised X-means), which is a new algorithm of supervised discretization of continuous attributes on clustering. The algorithm modifies clusters with knowledge of the class distribution dynamically. And this procedure can not stop until the proper k is found. For the number of clusters k is not pre-determined by the user and class distribution is applied, the random of result is decreased greatly. Experimental evaluation of several discretization algorithms on six artificial data sets show that the proposed algorithm is more efficient and can generate a better discretization schema. Comparing the output of C4.5, resulting tree is smaller, less classification rules, and high accuracy of classification.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114363223","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 : 2009-12-04DOI: 10.1109/CCPR.2009.5343973
Jun Ye, Lin-Lin Huang, X. Hao
The detection of texts in video images is an important task towards automatic content-based information indexing and retrieval system. In this paper, we propose a texture-based method for text detection in complex video images. Taking advantage of the desirable characteristic of gray-scale invariance of local binary patterns (LBP), we apply a modified LBP operator to extract feature of texts. A polynomial neural network (PNN) is employed to make classification. The PNN is trained with large quantities of samples collected using a bootstrap strategy. In addition, post-processing procedure including verification and integration is performed to refine the detected results. The effectiveness of the proposed method is demonstrated by experimental results.
{"title":"Neural Network Based Text Detection in Videos Using Local Binary Patterns","authors":"Jun Ye, Lin-Lin Huang, X. Hao","doi":"10.1109/CCPR.2009.5343973","DOIUrl":"https://doi.org/10.1109/CCPR.2009.5343973","url":null,"abstract":"The detection of texts in video images is an important task towards automatic content-based information indexing and retrieval system. In this paper, we propose a texture-based method for text detection in complex video images. Taking advantage of the desirable characteristic of gray-scale invariance of local binary patterns (LBP), we apply a modified LBP operator to extract feature of texts. A polynomial neural network (PNN) is employed to make classification. The PNN is trained with large quantities of samples collected using a bootstrap strategy. In addition, post-processing procedure including verification and integration is performed to refine the detected results. The effectiveness of the proposed method is demonstrated by experimental results.","PeriodicalId":354468,"journal":{"name":"2009 Chinese Conference on Pattern Recognition","volume":"170 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117276801","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：