Hand Detection and Gesture Recognition Exploit Motion Times Image in Complicate Scenarios

• Published in: Advances in Visual Computing pp. 628 - 636
• Main Authors: Song, Zhan, Yang, Hanxuan, Zhao, Yanguo, Zheng, Feng
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2010
• Series: Lecture Notes in Computer Science
• Subjects: Gesture Recognition; Hand Gesture; Hand Gesture Recognition; Hand Shape; Radial Basis Function Kernel
• ISBN: 3642172733; 9783642172731
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-642-17274-8_61

Hand gesture recognition in complicate scenario is still a challenging problem in computer vision domain. In this paper, a novel hand gesture recognition system is presented. To detect the exact hand target from complicate scenarios, the color and motion clues are used to obtain potential hand regions. And then a method named Motion Times Image (MTI) is proposed to identify the optimal hand location. The R-transform descriptor is used to describe the hand shape features and an offline trained Support Vector Machine with Radial Basis Function kernels (RBF-SVM) is exploited to perform the hand gesture recognition task. Extensive experiments with different users under dynamic and complicate scenarios are conducted to show its high recognition accuracy and strong robustness.