摘要
利用OpenCV计算机视觉库在vs2008平台上设计了一个基于实时摄像头的集动态手势检测、动态手势跟踪、动态手势轨迹识别的应用。首先,该应用基于静止的背景更新,利用背景差分检测运动手势,再结合颜色直方图的粒子滤波进行动态手势跟踪,最后利用隐马尔可夫模型(HMM)进行运动轨迹识别。在运动检测部分结合了背景差分图与通过颜色直方图获得的反投影图,达到比较满意的实时运动检测效果;在运动手势跟踪部分,改进的颜色直方图的粒子跟踪能够在经过类肤色人脸的干扰后迅速地找回运动手势,基本达到了跟踪的要求,但是同时对于HMM识别轨迹时需要的运动轨迹序列采集造成了影响;在识别轨迹部分,HMM的训练达到了识别的要求,但是识别的效果主要取决于实时运动轨迹序列的采集工作与采集方法的优化。
Using OpenCV computer vision library, the authors designed an application in vs2008 platform, which was based on real-time camera and made up of dynamic gesture detection, d）~namic gesture tracking, dynamic gesture trajectory recognition. Firstly, the application based on static background update, used the background subtraction to detect motion gestures; secondly, it used particle filter combined with the color histogram for dynamic gestures tracking; and then, it used the Hidden Markov Model （HMM） for trajectory recognition. In part of motion detection, combined with the projection map from color histogram and background subtraction, it can get satisfactory effect of real-time motion detection; in part of motion gesture tracking, the particle tracking with improved color histogram can quickly get back the motion gestures when going through interference in color of skin-like face, which basically meet the tracking requirements but influences the collection of HMM trajectory sequence; in part of trajectory recognition, the training of HMM meets the requirements of recognition, but the effect of recognition depends on the collection works of real-time trajectory sequence and the the optimization of collection method.
出处
《计算机应用》
CSCD
北大核心
2012年第A01期128-133,共6页
journal of Computer Applications
关键词
OPENCV
摄像头
粒子跟踪
HMM
轨迹识别
OpenCV
camera
particle tracking
Hidden Markov Model （HMM）
trajectory recognition
