一种基于单目视觉的人手检测与识别方法

摘要

A human hand location and gesture recognition methods was proposed in this paper .Skin color detection and motion detection technology are combined to achieve hand detection ,followed by threshold segmentation to obtain binary im-ages .Fourier shape descriptors are extracted as binary images'features .After similarity comparison with the samples'fea-tures ,the final recognition result is obtained .Experiments show that this method can effectively implement hand gesture lo-cation and recognition under complex background .%提出了一种单目视觉下的人手检测与识别方法。该方法结合肤色检测与运动前景检测技术，实现了人手的定位，再由阈值分割获取人手的二值图像。提取傅里叶形状描述子作为二值图像的特征，在与样本的特征进行相似性比对之后获得最终识别结果。实验证明，该方法可以有效地实现复杂背景下人手的定位与识别。