基于视觉的实时手势识别及其在演示控制中的应用

研究基于计算机视觉的实时动态手势识别技术，并利用OpenCV计算机视觉库在VS2010平台上设计一个基于该技术在多媒体教学中PPT演示控制方面的应用。首先，利用背景差分法进行手势检测，在背景更新的基础上，通过背景差分图和颜色直方图的反投影图来检测运动手势区域，可以达到较为满意的实时运动手势检测效果；其次，采用基于颜色直方图的粒子跟踪算法进行手势跟踪，基本能满足跟踪的实时性；最后，在手势识别阶段，采用基于Hu不变矩的轮廓匹配算法，得到较好的手势识别效果；使用六种手势，来实现演示文稿中的控制应用。

The Real-time Dynamic Gesture Recognition Based on Computer Vision Technology and its Application for Powerpoint Control

The study is based on the real-time dynamic gesture recognition in computer vision technology. And it designs a applications about PowerPoint control based on the technology in multimedia teaching using OpenCV vision library in theVS2010 platform .First, using the background subtraction method, the hand gesture detecting motion gesture area through the background difference diagrams and color histogram back projection figure detected on the basis of the background update, can achieve more satisfactory detection results of real-time motion. Secondly, gesture tracking using particle tracking algorithm based on color histogram, basically meet the real-time tracking. Finally, using the contours-matching method that based on Hu moments, this method can bring to a satisfactory conclusion. And the application of controlling PowerPoint uses six gestures, can meet the requirements of the application.