基于尺度空间特征实时手势识别

The Real-time Gesture Recognition Based on Scale Space Features

为了实现快速准确的对图像中的手势进行检测与识别,同时为相关后续研究打下基础,在对现有的方法进行改进的基础上,提出了使用基于尺度空间的特征检测方法在图像中分别检测手指和手掌的结构,进而完成手势识别实验.实验数据表明,这种方法在手势识别的正确率上较其他方法有了一定的提高.

The purpose of this paper is to detect and recognize human gesture quickly and accurately, as well as to lay the foundation for future studies. The detection method of scale - space was used to detect the structure of the fingers and palms, thus completing the gesture recognition. Judging from the experimental data, the method obtained high accuracy rate in gesture recognition.