基于结构参照的手势识别

摘要

This paper proposes a method which introduces human body structure into gesture recognition.Referring to the body structure,the system could rule out background interference quickly,locate the subject's gestures accurately,and track the hand automatical y.This method applies a common monocular webcam instead of expensive depth cameras or complex binocular cameras.lt can simulate the subject's gestures wel .After a modified KNN classification,the corresponding postures and movements are identified.This method avoids global scanning and simplifies the classification process.%在现有的单目视觉识别手势的基础上，引入人体结构参数识别手势。该方法通过判别人体结构，快速地排除背景干扰快速分组近邻搜索，识别实验对象的手势及其动作语义，避免了逐帧全幅扫描图像，大大降低了计算量，很好地解决了目标丢失问题，对复杂动态背景干扰下的手势识别有着很好的稳定性。另外该方法仅使用简单家用摄像头，无需复杂的深度摄像机或者双目摄像机。