摘要： 目前常用的基于视觉的静态手势特征提取方法只从单一方面进行描述，缺乏全局信息和局部信息的有效结合。为此，提出一种结合CSS形状描述子与傅里叶描述子的手势特征提取方法。将CSS形状描述子与傅里叶描述子相结合，以此作为一种融合手势局部特征和全局特征的新的静态手势特征。实验结果表明，与传统方法相比，该方法的正确率更高，达到98.3%。

关键词: 基于视觉, 静态手势特征, CSS形状描述子, 傅里叶描述子, 局部特征, 全局特征

Abstract: Aiming at the problem that the common static gesture features which are used in vision based recognition often only focus on gestures in a single aspect, and lack of effective gesture description for combination of wholeness and local part, this paper proposes an approach which extracts Fourier descriptor as a kind of global feature, and CSS shape descriptor as a kind of local feature, and then combines them into the final feature. Experimental results verify the effectiveness of this new feature on static gesture recognition by its high correct rate 98.3%, compared with traditional approaches.

Key words: vision based, static gesture feature, CSS shape descriptor, Fourier descriptor, local feature, global feature

中图分类号: 

• TP391.41