Facial feature extraction is important in many face-related applications, such as face alignment for recognition. Recently, boosting-based methods have led to the state-of-the-art face detection and localization systems. In this paper, we propose a multi-modal boosting algorithm to integrate 3D (range) and 2D (intensity) information provided from a facial scan to detect the face and feature point (nose tip, eyes center). Given a face scan, Gauss and mean curvature are calculated. Face, nose and eyes detectors are trained in color images and curvature maps features space using AdaBoost. As a result, a fully automatic multi-modal face location system is developed. The performance evaluation is conducted for the proposed feature extraction algorithm on a publicly available data-base, containing 4007 facial scans of 466 subjects
{"title":"3D+2D Face Localization Using Boosting in Multi-Modal Feature Space","authors":"Feng Xue, Xiaoqing Ding","doi":"10.1109/ICPR.2006.35","DOIUrl":"https://doi.org/10.1109/ICPR.2006.35","url":null,"abstract":"Facial feature extraction is important in many face-related applications, such as face alignment for recognition. Recently, boosting-based methods have led to the state-of-the-art face detection and localization systems. In this paper, we propose a multi-modal boosting algorithm to integrate 3D (range) and 2D (intensity) information provided from a facial scan to detect the face and feature point (nose tip, eyes center). Given a face scan, Gauss and mean curvature are calculated. Face, nose and eyes detectors are trained in color images and curvature maps features space using AdaBoost. As a result, a fully automatic multi-modal face location system is developed. The performance evaluation is conducted for the proposed feature extraction algorithm on a publicly available data-base, containing 4007 facial scans of 466 subjects","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117008564","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}
A novel watermarking algorithm based on contourlet transform is proposed in this paper. The watermark composed of pseudo-random sequence is embedded in the selected contourlet transform coefficients by means of multiplicative method. The contourlet coefficients are modeled with generalized Gaussian distribution with zero mean, and then watermark detection method is proposed based on maximum likelihood detection. Furthermore the decision rule is optimized via Neyman-Pearson criterion. Experimental results show that the fidelity of the watermarked image is good and robust to signal processing and small geometrical attacks
{"title":"A Novel Blind Watermarking Algorithm in Contourlet Domain","authors":"Haifeng Li, Jianting Wen, Haifeng Gong","doi":"10.1109/ICPR.2006.134","DOIUrl":"https://doi.org/10.1109/ICPR.2006.134","url":null,"abstract":"A novel watermarking algorithm based on contourlet transform is proposed in this paper. The watermark composed of pseudo-random sequence is embedded in the selected contourlet transform coefficients by means of multiplicative method. The contourlet coefficients are modeled with generalized Gaussian distribution with zero mean, and then watermark detection method is proposed based on maximum likelihood detection. Furthermore the decision rule is optimized via Neyman-Pearson criterion. Experimental results show that the fidelity of the watermarked image is good and robust to signal processing and small geometrical attacks","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117087106","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}
Inferring 3D body pose as well as viewpoint from a single silhouette image is a challenging problem. We present a new generative model to represent shape deformations according to view and body configuration changes on a two dimensional manifold. We model the two continuous states by a product space (different configurations times different views) embedded on a conceptual two dimensional torus manifold. We learn a nonlinear mapping between torus manifold embedding and visual input (silhouettes) using empirical kernel mapping. Since every view and body pose has a corresponding embedding point on the torus manifold, inferring view and body pose from a given image becomes estimating the embedding point from a given input. As the shape varies in different people even in the same view and body pose, we extend our model to be adaptive to different people by decomposing person dependent style factors. Experimental results with real data as well as synthetic data show simultaneous estimation of view and body configuration from given silhouettes from unknown people
{"title":"Simultaneous Inference of View and Body Pose using Torus Manifolds","authors":"Chan-Su Lee, A. Elgammal","doi":"10.1109/ICPR.2006.1058","DOIUrl":"https://doi.org/10.1109/ICPR.2006.1058","url":null,"abstract":"Inferring 3D body pose as well as viewpoint from a single silhouette image is a challenging problem. We present a new generative model to represent shape deformations according to view and body configuration changes on a two dimensional manifold. We model the two continuous states by a product space (different configurations times different views) embedded on a conceptual two dimensional torus manifold. We learn a nonlinear mapping between torus manifold embedding and visual input (silhouettes) using empirical kernel mapping. Since every view and body pose has a corresponding embedding point on the torus manifold, inferring view and body pose from a given image becomes estimating the embedding point from a given input. As the shape varies in different people even in the same view and body pose, we extend our model to be adaptive to different people by decomposing person dependent style factors. Experimental results with real data as well as synthetic data show simultaneous estimation of view and body configuration from given silhouettes from unknown people","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129414186","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}
An algorithm for automatic estimation of 3D transformations between two objects is presented in this paper. Skeletons of the 3D objects are created using a fully parallel thinning technique, feature point pairs (land markers) are automatically extracted from skeletons, and a least squares method is applied to solve an over determined linear system to estimate the 3D transformation matrix. Experiments show that this method is quite accurate when the translations and rotation angles are small, even when there is some noise in the data. The estimation process requires about 2 seconds on an Intel Centrino Laptop with 512 MB memory, for a complex model with about 37,000 object points and 500 object points for its skeletons
{"title":"Automatic Estimation of 3D Transformations using Skeletons for Object Alignment","authors":"Tao Wang, A. Basu","doi":"10.1109/ICPR.2006.298","DOIUrl":"https://doi.org/10.1109/ICPR.2006.298","url":null,"abstract":"An algorithm for automatic estimation of 3D transformations between two objects is presented in this paper. Skeletons of the 3D objects are created using a fully parallel thinning technique, feature point pairs (land markers) are automatically extracted from skeletons, and a least squares method is applied to solve an over determined linear system to estimate the 3D transformation matrix. Experiments show that this method is quite accurate when the translations and rotation angles are small, even when there is some noise in the data. The estimation process requires about 2 seconds on an Intel Centrino Laptop with 512 MB memory, for a complex model with about 37,000 object points and 500 object points for its skeletons","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129507796","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}
A mathematical basis for the digitization of gray value of an image is proposed. This was called Oteru-Koshimizu quantization theorem (OK-QT), on the analogy of the Shannon sampling theorem (Shannon-ST) for the digitization of the shape of the image. Inspired by the fact that the Shannon-ST is the reconstruction theorem of the analog image from the discrete image, OK-QT was modeled as the reconstruction theorem of the shape of the probability density function of gray values of an image. This is a novel and unique mathematical basis for the digitization of the gray scale of an image. This paper outlines this theorem and also shows some experimental results to demonstrate its practical applicability. Through this, the OK-QT gives a clue to the mathematical paradigm for the complete basis for digitization, together with Shannon ST
{"title":"OK-Quantization Theory - A Mathematical Theory of Quantization -","authors":"H. Koshimizu, Yuji Tanaka, T. Fujiwara","doi":"10.1109/ICPR.2006.896","DOIUrl":"https://doi.org/10.1109/ICPR.2006.896","url":null,"abstract":"A mathematical basis for the digitization of gray value of an image is proposed. This was called Oteru-Koshimizu quantization theorem (OK-QT), on the analogy of the Shannon sampling theorem (Shannon-ST) for the digitization of the shape of the image. Inspired by the fact that the Shannon-ST is the reconstruction theorem of the analog image from the discrete image, OK-QT was modeled as the reconstruction theorem of the shape of the probability density function of gray values of an image. This is a novel and unique mathematical basis for the digitization of the gray scale of an image. This paper outlines this theorem and also shows some experimental results to demonstrate its practical applicability. Through this, the OK-QT gives a clue to the mathematical paradigm for the complete basis for digitization, together with Shannon ST","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"89 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129598398","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}
In this paper, the efficiency of the JPEG2000 scheme combined with a complementary denoising process is analyzed on simulated and real denial ortho-pantomographic images, where the simulation images are perturbed by Poisson noise. The case of dental radiography is investigated, because radiographic images are a combination between the relevant signal and a significant amount of acquisition noise, which is per definition not compressible. The noise behaves generally close to Poisson statistics, which generally affects the compression performance. The denoising process is supported by Monte Carlo noise modeling, which is introduced in the JPEG 2000 compression scheme to improve the compression efficiency of the medical images in terms of compression ratio and image quality. Fifty selected images are denoised and the compression ratio, using lossless and lossy JPEG 2000, is reported and evaluated
{"title":"Medical Image Compression: Study of the Influence of Noise on the JPEG 2000 Compression Performance","authors":"A. Belbachir, P. Goebel","doi":"10.1109/ICPR.2006.786","DOIUrl":"https://doi.org/10.1109/ICPR.2006.786","url":null,"abstract":"In this paper, the efficiency of the JPEG2000 scheme combined with a complementary denoising process is analyzed on simulated and real denial ortho-pantomographic images, where the simulation images are perturbed by Poisson noise. The case of dental radiography is investigated, because radiographic images are a combination between the relevant signal and a significant amount of acquisition noise, which is per definition not compressible. The noise behaves generally close to Poisson statistics, which generally affects the compression performance. The denoising process is supported by Monte Carlo noise modeling, which is introduced in the JPEG 2000 compression scheme to improve the compression efficiency of the medical images in terms of compression ratio and image quality. Fifty selected images are denoised and the compression ratio, using lossless and lossy JPEG 2000, is reported and evaluated","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129790093","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}
Numerous techniques exist which can be used for the task of behavioural analysis and recognition. Common amongst these are Bayesian networks and hidden Markov models. Although these techniques are extremely powerful and well developed, both have important limitations. By fusing these techniques together to form Bayes-Markov chains, the advantages of both techniques can be preserved, while reducing their limitations. The Bayes-Markov technique forms the basis of a common, flexible framework for supplementing Markov chains with additional features. This results in improved user output, and aids in the rapid development of flexible and efficient behaviour recognition systems
{"title":"A Combined Bayesian Markovian Approach for Behaviour Recognition","authors":"N. Carter, D. P. Young, J. Ferryman","doi":"10.1109/ICPR.2006.47","DOIUrl":"https://doi.org/10.1109/ICPR.2006.47","url":null,"abstract":"Numerous techniques exist which can be used for the task of behavioural analysis and recognition. Common amongst these are Bayesian networks and hidden Markov models. Although these techniques are extremely powerful and well developed, both have important limitations. By fusing these techniques together to form Bayes-Markov chains, the advantages of both techniques can be preserved, while reducing their limitations. The Bayes-Markov technique forms the basis of a common, flexible framework for supplementing Markov chains with additional features. This results in improved user output, and aids in the rapid development of flexible and efficient behaviour recognition systems","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"87 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128215689","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}
In this paper the issue of image segmentation within the framework of nonlinear multiscale watersheds in combination with graph theory based techniques is addressed. First, a graph is created which decomposes the image in scale and space using the concept of multiscale watersheds. In the subsequent step the obtained graph is partitioned using recursive graph cuts in a coarse to fine manner. In this way, we are able to combine scale and feature measures in a flexible way: the feature-set that is used to measure the dissimilarities may change as we progress in scale. We employ the earth mover's distance on a featureset that combines color, scale and contrast features to measure the dissimilarity between the nodes in the graph. Experimental results demonstrate the efficiency of the proposed method for natural scene images
{"title":"Nonlinear Multiscale Graph Theory based Segmentation of Color Images","authors":"I. Vanhamel, H. Sahli, I. Pratikakis","doi":"10.1109/ICPR.2006.866","DOIUrl":"https://doi.org/10.1109/ICPR.2006.866","url":null,"abstract":"In this paper the issue of image segmentation within the framework of nonlinear multiscale watersheds in combination with graph theory based techniques is addressed. First, a graph is created which decomposes the image in scale and space using the concept of multiscale watersheds. In the subsequent step the obtained graph is partitioned using recursive graph cuts in a coarse to fine manner. In this way, we are able to combine scale and feature measures in a flexible way: the feature-set that is used to measure the dissimilarities may change as we progress in scale. We employ the earth mover's distance on a featureset that combines color, scale and contrast features to measure the dissimilarity between the nodes in the graph. Experimental results demonstrate the efficiency of the proposed method for natural scene images","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128406989","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}
Shape warping is a key problem in statistical shape analysis. This paper proposes a framework for geometric shape warping based on both shape distances and landmarks. Our method is compatible with implicit representations and a matching between shape surfaces is provided at no additional cost. It is, to our knowledge, the first time that landmarks and shape distances are reconciled in a pure geometric level set framework. The feasibility of the method is demonstrated with two- and three-dimensional examples. Combining shape distance and landmarks, our approach reveals to need only a small number of landmarks to obtain improvements on both warping and matching
{"title":"Reconciling Landmarks and Level Sets","authors":"Pierre Maurel, R. Keriven, O. Faugeras","doi":"10.1109/ICPR.2006.979","DOIUrl":"https://doi.org/10.1109/ICPR.2006.979","url":null,"abstract":"Shape warping is a key problem in statistical shape analysis. This paper proposes a framework for geometric shape warping based on both shape distances and landmarks. Our method is compatible with implicit representations and a matching between shape surfaces is provided at no additional cost. It is, to our knowledge, the first time that landmarks and shape distances are reconciled in a pure geometric level set framework. The feasibility of the method is demonstrated with two- and three-dimensional examples. Combining shape distance and landmarks, our approach reveals to need only a small number of landmarks to obtain improvements on both warping and matching","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128373351","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}
We present an approach for differentiating between large numbers of similar feature points. The approach employs a learning strategy which utilizes mutual information to yield relational information or structure between feature points. It learns an ordered list of jumps in space and scale which is used for differentiation. To test the viability and potential of the approach, two datasets containing faces and objects were used
{"title":"Differentiating Between Many Similar Features using Relational Information in Space and Scale","authors":"Timothy S. Y. Gan, T. Drummond","doi":"10.1109/ICPR.2006.449","DOIUrl":"https://doi.org/10.1109/ICPR.2006.449","url":null,"abstract":"We present an approach for differentiating between large numbers of similar feature points. The approach employs a learning strategy which utilizes mutual information to yield relational information or structure between feature points. It learns an ordered list of jumps in space and scale which is used for differentiation. To test the viability and potential of the approach, two datasets containing faces and objects were used","PeriodicalId":236033,"journal":{"name":"18th International Conference on Pattern Recognition (ICPR'06)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129023346","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：