关键词: 图像处理, 手势特征检测, 曲率算法, 特征点分析, 三维人手跟踪

Abstract: To improve the accuracy and speed of gesture features detection, a fast and accurate gesture features detection method under complex background was put forward. Gesture was segmented to obtain the contour by hand segmentation method based on the proposed color luminance. According to different hand types, the contour points were filtered to obtain candidate points by using improved curvature algorithm. Finally, the feature positions were determined by means of analyzing the distribution of the candidate points in the vicinity. And then, a new separation algorithm of feature points was proposed. Compared to the existing algorithms, the main innovations of proposed method lied in the fact that, a new skin model to enhance the robustness of gesture segmentation was proposed; Through introducing scale for different hand types and distribution of hand features, the feature positions were gradually extracted with high accuracy and low time consumption. The experimental results demonstrated that, the proposed method could detect feature under complicated circumstance with higher accuracy and real time performance, and it also reduced the error conclusion of interference and redundancies of feature points in traditional curavture method. It had robustness for different types gestures. The method was proved to be feasible and valid by applying successfully in the studies of 3D human hand tracking.

Key words: image processing, gesture features detection, curvature algorithm, characteristic analysis