视频流中运动物体的分割

The Segmentation of Moving Object in Video Flow

为了从视频流中快速、准确地找出运动物体,为后继智能图像处理工作奠定基础,从动态、静态两个方面对运动物体进行分割,采用消除背景的方法来获得视频流中的运动物体,并将获得运动物体作为消除背景图像的结果。在比较帧差算法的基础上,提出了基于视频流的帧间微分算法,并将图像回填技术应用于图像分割,不仅能够快速消除背景,获得运动物体,而且能够减少运算量,极大克服由于光线强弱、背景变化、局部抖动等因素带来的图像处理中的困难。

In order to get moving objects from a video stream quickly and accurately and provide a foundation for intelligent image processing,in this paper,we propose a method of eliminating background to get moving objects in a video stream,in which the moving objects are seen as the result of the elimination of background image.By comparing the frame differences algorithm,a differential frame-based video streaming algorithm is presented,and image backfilling technology is applied to image segmentation.By using these techniques,the background images can be quickly removed.Moving objects can be easily obtained.Computation can be greatly reduced and the difficulties,such as light intensity,background changing,local jitter etc.,can be greatly overcome in image processing.