基于OpenCV动态手势检测识别方法的研究

Study of Hand Gesture Detection and Identification Algorithm Based on OpenVC

摘要: 针对动态手势识别实时性不足和识别效率低的问题，本文提出一种基于OpenCV的动态手势检测和数字识别方法。该方法采用肤色特征提取手势轮廓，以预测和重定位的形式采用HMM来对手势进行跟踪，手势的识别采用先对指尖轮廓进行定位的方式。经过实验，本文方法能够稳定地跟踪手掌区域，识别准确度高，并能够在保证识别率的基础上实现对手指数目的准确识别跟踪。

Abstract: In view of the problem of lack of real-time recognition and low efficiency of dynamic gesture recognition, a dynamic gesture tracking and digital recognition algorithm based on OpenCV is proposed. The method uses the skin color feature to extract the gesture profile, and HMM is used to track gestures in the form of predictions and relocations. The identification of gestures is done by first positioning the fingertip profile. After the experiment, the framework can track the palm area stably, and the recognition accuracy is high, and there is no error and can achieve the accuracy of the target tracking based on the recognition rate.