基于BOF-Gist特征的手势识别算法研究

摘要

针对静态手势识别算法存在特征计算复杂度高,实时性差的问题,提出了一种新的BOF-Gist特征对手势图像进行表示.该特征的优势是在保持Gist特征原有优势的基础上,有效地表征手势图像的局部特征和全局特征,并且特征维数明显降低,实时性好.在标准数据库上的测试表明,该算法对于简单、复杂背景下十种手语手势分别得到了90.42%与79.05%的正确识别率,同时验证了算法的实时性.%Aiming at the problem of high computational complexity and poor real-time performance of static gesture rec-ognition algorithm, a novel BOF-Gist feature is proposed. This feature can effectively represent the local features and global features of hand posture image with the original advantages of Gist features, the feature dimension is relatively lower, and the real-time performance is better. The testing results on the standard database show that correct recognition rate of the algorithm is 90.42% and 79.05% respectively for ten sign language postures under simple and complex background, and at the same time the real-time performance of the algorithm is verified.