图像识别的静态手势识别与动态跟踪系统设计

Design of Static Gesture Recognition and Dynamic Tracking System for Image Recognition

本文研究了图像手势识别和增强现实技术,设计了可以进行静态手势识别和动态跟踪的系统,通过提前录入不同手势,利用皮肤颜色对图像进行OSTU自适应阈值划分,建立二值化图像,与已知的手势进行匹配,以得到手势结果。实验结果表明,准确率达到96.8%,识别速度达到0.55 s。动态跟踪利用检测每帧图像中手部的位置进行定位和捕捉,图像捕捉帧数达到28帧/s,对手势静态识别和动态跟踪实现了人机之间的良好交互。

In the paper,the gesture recognition technology and augmented reality are researched,the static gesture recognition and dynamic track system are designed,through the early entry of different gestures,by using skin color for OSTU adaptive threshold value division,set up image binarization,matching with known gestures,gesture to get results.The experimental accuracy is 96.8%,and the recognition speed is 0.55 s.Dynamic tracking detects the position of the hand in each frame of the image for positioning and capturing.The number of frames captured in the image is up to 28 frames per second.Good interaction between human and machine is realized for gesture static recognition and dynamic tracking.