1.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;2.College of Electronic and Information Engineering, Jiangsu University, Zhenjiang 210023, China; 3.School of Civil Engineering and Architecture, Zhejiang University of Science and Technology, Hangzhou 310023, China
A head-reference human contour model (HHCM) was proposed in order to improve the human motion recognition performance with human contour as the main feature. The human contour was extracted from the video frame sequence by employing the level set method without re-initialization, and the frame with the contour of the smallest aspect ratio was defined as the key frame. The contour endpoints in the key frame, i.e., the head, hands, and feet, etc were determined by the local maximum of Euclidean distances. The head location was obtained by the skin color model and was further considered as the base point of the contour. Then the head-reference contour model was established. Features were extracted based on the model, and support vector machine (SVM) was applied to identify human behaviors. The effectiveness of the model and the behavior recognition approach were verified by the experiments on the WEIZMANN database.
[1] WANG L, HU W M, TAN T N. Recent developments in human motion analysis [J]. Pattern Recognition, 2003, 36(3): 585-601. [2] MOESLUND T, HILTON A, KRUGER V. A survey of advances in vision based human motion capture and analysis [J]. Computer Vision and Image Understanding, 2006, 104(2/3): 90-126. [3] FUJIYOSHI H, LIPTON A J. Realtime human motion analysis by image skeletonization [C]∥Proceedings of the 4th IEEE Workshop on Applications of Computer Vision. Princeton: IEEE, 1998: 15-21. [4] CHEN H S, CHEN H T, CHEN Y W, et al. Human action recognition using Star skeleton [C]∥ Proceeding of 4th ACM International Workshop on Video Surveillance and Sensor Networks. New York: ACM, 2006: 171-178. [5] CHUN S, HONG K, JUNG K. 3D Star skeleton for fast human posture representation [C]∥ Proceeding of World Academy of Science, Engineering and Technology. Auckland: WASET, 2008: 273-282. [6] YU E, AGGARWAL J K. Detection of fence climbing from monocular video [C]∥ 18th International Conference on Pattern Recognition. HongKong: IEEE, 2006: 375-378. [7] SCHINDLER K, VAN GOOL L. Action snippets: how many frames does human action recognition requires? [C]∥Computer Vision and Pattern Recognition(CVPR). Anchorage, AK: IEEE, 2008. [8] STAUFFER C, GRIMSON W E L. Adaptive background mixture models for realtime tracking [C]∥ Computer Vision and Pattern Recognition. Fort Collins, CO: IEEE, 1999: 246-252. [9] LI C M, XU C Y, GUI C F, et al. Level set evolution without reinitialization: a new variational formulation [C]∥ Computer Vision and Pattern Recognition. Washington: IEEE, 2005: 430-436. [10] MALLADI R, SETHIAN J A, VEMURI B C. Shape modeling with front propagation: a level set approach [J]. Pattern Analysis and Machine Intelligence, 1995, 17(2): 158-175. [11] CHEN M L, KEE S. Head tracking with shape modeling and detection [C]∥ Computer and Robot Vision. Canada: IEEE, 2005. [12] BIRCHFIELD S. Elliptical head tracking using intensity gradients and color histograms [C]∥ Computer Vision and Pattern Recognition. Santa Barbara: IEEE, 1998: 232-237. [13] ALEXANDER B, RAINER H. Robust head detection and tracking in cluttered workshop environments using GMM [J]. Lecture Notes in Computer Science, 2005,3663: 442-450. [14] KEOGH E J, PAZZANI M J. Derivative dynamic time warping [C]∥ 1st SIAM International Conference on Data Mining. USA: SIAM, 2001. [15] BLANK M, GORELICK L. Actions as spacetime shapes [C]∥ Proceedings of the 10th IEEE International Conference on Computer Vision (ICCV). Beijing: IEEE, 2005: 1395-1402. [16]NIEBLES J C, FEIFEI L. A hierarchical model of shape and appearance for human action classification [C]∥ Computer Vision and Pattern Recognition. Minneapolis: IEEE, 2007. [17] ALE S, BASHARAT A, SHAH M. Chaotic invariants for human action recognition [C]∥ International Conference on Computer Vision. Rio de Janeiro, Brazil: IEEE, 2007. [18] LI Ning, XU De. Action recognition using weighted threestate hidden Markov model [C]∥ IEEE ICSP2008 Proceedings. Beijing: IEEE, 2008.