1,721,344 research outputs found
Weiming-Hu/AnalogsEnsemble 3.6.3.3
No full text<p>The C++ and R packages for parallel ensemble forecasts using Analog Ensemble</p>
Spatio-temporal proximity distribution kernels for action recognition
No full textChunfeng Yuan, Weiming Hu, Hanzi Wang, Xi Li, Nianhua Xi
Corner detection of contour images using spectral clustering
Full text linkCorner detection plays an important role in object recognition and motion analysis. In this paper, we propose a hierarchical corner detection framework based on spectral clustering (SC). The framework consists of three stages: contour smoothing, corner cell extraction and corner localization. In the contour smoothing stage, wavelet decomposition is imposed on the raw contour to reduce noise. In the corner cell extraction stage, several atomic corner cells are obtained by SC. In the corner localization stage, the corner points of each corner cell are located by the corner locator based on the kernel-weighted cosine curvature measure. Experimental results demonstrate the superiority of our framework.Xi Li, Weiming Hu, Zhongfei Zhan
On a three-dimensional gait recognition system
No full textThe University of Southampton Multi-Biometric Tunnel is a high performance data-capture and recognition system; designed with airports and other busy public areas in mind. It is able to acquire a variety of non-contact biometrics in a non-intrusive manner, requiring minimal subject cooperation. The system uses twelve cameras to record gait and perform three-dimensional reconstruction; the use of volumetric data avoids the problems caused by viewpoint dependence - a serious problem for many gait analysis approaches. The early prototype by Middleton et al. was used as the basis for creating a new and improved system, designed for the collection of a new large dataset, containing gait, face and ear. Extensive modifications were made, including new software for managing the data collection experiment and processing the dataset. Rigorous procedures were implemented to protect the privacy of participants and ensure consistency between capture sessions. Collection of the new multi-biometric dataset spanned almost one year; resulting in over 200 subjects and 2000 samples.Experiments performed on the newly collected dataset resulted in excellent recognition performance, with all samples correctly classified and a 1.58% equal error rate; the matching of subjects against previous samples was also found to be reasonably accurate. The fusion of gait with a simple facial analysis technique found the addition of gait to be beneficial -- especially at a distance. Further experiments investigated the effect of static and dynamic features, camera misalignment, average silhouette resolution, camera layout, and the matching of outdoor video footage against data from the Biometric Tunnel. The results in this thesis prove significant due to the unprecedented size of the new dataset and the excellent recognition performance achieved; providing a significant body of evidence to support the argument that an individual's gait is unique.L. Middleton, D. K. Wagg, A. I. Bazin, J. N. Carter and M. S. Nixon. A smart environment for biometric capture. Automation Science and Engineering, Proceedings of IEEEInternational Conference on, 57-62, 2006
A coarse-to-fine strategy for vehicle motion trajectory clustering
Full text linkHigh-level semantic understanding of vehicle motion behaviors is often based on vehicle motion trajectory clustering. In this paper, we propose an effective trajectory clustering framework in which a coarse-to-fine strategy is taken. Our framework consists of four stages: trajectory smoothing, feature extraction, trajectory coarse clustering and trajectory fine clustering. Wavelet decomposition is imposed on raw trajectories to reduce noise in the trajectory smoothing stage. Besides the commonly used positional feature, a novel feature called trajectory directional histogram is proposed to describe the statistic directional distribution of a trajectory in the feature extraction stage. Both coarse clustering and fine clustering are based on a novel graph-theoretic clustering algorithm called dominant-set clustering, but they deal with different trajectory features. Experiments in our pre-labeled trajectory database demonstrate that the proposed trajectory clustering framework possesses a very high accuracy.Xi Li, Weiming Hu, Wei H
Video shot segmentation using graph-based dominant-set clustering
No full textVideo shot segmentation is a solid foundation for automatic video content analysis, for most content based video retrieval tasks require accurate segmentation of video boundaries. In recent years, using the tools of data mining and machine learning to detect shot boundaries has become more and more popular. In this paper, we propose an effective video segmentation approach based on a dominant-set clustering algorithm. The algorithm can not only automatically determine the number of video shots, but also obtain accurate shot boundaries with low computation complexity. Experimental results have demonstrated the effectiveness of the proposed shot segmentation approach.Li Li, Xianglin Zeng, Xi Li, Weiming Hu, Pengfei Zh
Image spam filtering using Fourier-Mellin invariant features
No full textImage spam is a new obfuscating method which spammers invented to more effectively bypass conventional text based spam filters. In this paper, a framework for filtering image spams by using the Fourier-Mellin invariant features is described. Fourier-Mellin features are robust for most kinds of image spam variations. A one-class classifier, the support vector data description (SVDD), is exploited to model the boundary of image spam class in the feature space without using information of legitimate emails. Experimental results demonstrate that our framework is effective for fighting image spam.Haiqiang Zuo, Xi Li, Ou Wu, Weiming Hu and Guan Lu
Learning group activity in soccer videos from local motion
No full textYu Kong, Weiming Hu, Xiaoqin Zhang, Hanzi Wang and Yunde Jiahttp://www.cis.pku.edu.cn/accv2009/Technical_Program.htm
Robust Foreground Segmentation Based on Two Effective Background Models
No full textForeground segmentation is a common foundation for many computer vision applications such as tracking and behavior analysis. Most existing algorithms for foreground segmentation learn pixel-based statistical models, which are sensitive to dynamic scenes such as illumination change, shadow movement, and swaying trees. In order to address this problem, we propose two block-based background models using the recently developed incremental rank-(R1, R2, R3) tensor-based subspace learning algorithm (referred to as IRTSA [1]). These two IRTSA-based background models (i.e., IRTSAGBM and IRTSA-CBM respectively for grayscale and color images) incrementally learn low-order tensor-based eigenspace representations to fully capture the intrinsic spatio-temporal characteristics of a scene, leading to robust foreground segmentation results. Theoretic analysis and experimental evaluations demonstrate the promise and effectiveness of the proposed background models.Xi Li, Weiming Hu, Zhongfei Zhang, Xiaoqin Zhanghttp://press.liacs.nl/mir2008/index.htm
Human action recognition using pyramid vocabulary tree
Full text linkThe bag-of-visual-words (BOVW) approaches are widely used in human action recognition. Usually, large vocabulary size of the BOVW is more discriminative for inter-class action classification while small one is more robust to noise and thus tolerant to the intra-class invariance. In this pape, we propose a pyramid vocabulary tree to model local spatio-temporal features, which can characterize the inter-class difference and also allow intra-class variance. Moreover, since BOVW is geometrically unconstrained, we further consider the spatio-temporal information of local features and propose a sparse spatio-temporal pyramid matching kernel (termed as SST-PMK) to compute the similarity measures between video sequences. SST-PMK satisfies the Mercer’s condition and therefore is readily integrated into SVM to perform action recognition. Experimental results on the Weizmann datasets show that both the pyramid vocabulary tree and the SST-PMK lead to a significant improvement in human action recognition.Chunfeng Yuan, Xi Li, Weiming Hu and Hanzi Wan
- …
