人体视频运动图像的高精度实时分割仿真研究  

Simulationresearch on the High Precision and Real-Time Segmentation for Video Image of Human Motion

作　　者：樊翠红[1] 

机构地区：[1]长江大学体育学院,湖北荆州434023

出　　处：《计算机仿真》2016年第1期268-271,共4页Computer Simulation

摘　　要：在人体实时运动优化识别中,由于人体运动幅度、背景变化无严格规律,速度与运动背景的变化造成运动过程的RGB空间的色度、饱和度等特征也是一个动态变化的过程,传统的分割方法,针对这种动态变化,无法通过调节阀值对特征变化进行实时区别,导致分割过程不充分。提出采用区域差异的视频运动对象的分割方法。将视频图像的RGB空间转换为HSV空间,利用色度、饱和度和亮度提取到运动对象封闭的轮廓图,根据视频图像中不同区域之间的差异,将视频图像分为运动区域、背景区域和遮挡区域,利用边缘检测算子对运动对象的边缘进行检测,最终分割出视频图像中的运动对象。仿真结果表明,利用改进算法进行视频运动对象分割,能够提高分割精度,满足了实时性的要求。In theoptimal identification process of real-time motion of human body, because the amplitude of human body movement and background changes don＇ t have strict rule, the changes of velocity and motion background cause that the characteristics of chroma, saturability and others of RGB space in the moving process are dynamic processes. The color and saturation of the motion process of the RGB space is also a dynamic change process. For this kind of dynamic change, the traditional segmentation method, which is unable to real-time distinguish the character- istic changes through adjusting threshold value, and is not sufficient for the segmentation process. A segmentation method for video motion objects using region difference is proposed. The RGB space of the video image is converted into HSV space, and the color, saturation and brightness are used to extract the closed contour of the moving object. According to the difference between different regions of video image, the video image is divided into motion region, background region and the occluded region. Edge detection operator is used to detect the edge of moving object and finally to segment the moving object in video image. The simulation results show that the use of improved algorithm for video motion object segmentation, can improve the segmentation accuracy, and meet the real-time requirement.

关 键 词：高精度 运动对象 分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]