中文摘要：

针对目前常用的视频分割方法易受到噪声、亮度突变的影响，很难提取出完整的运动目标这一问题，提出一种基于帧差法和边缘检测法相结合的视频分割算法。该算法首先对连续三帧图像进行差分得到运动区域，然后对当前帧进行Kirsch边缘检测得到边缘图像，综合二者的检测结果得到更为精准的运动对象边缘。采用边缘连接算法完成对断裂边缘的连接，最后通过区域填充得到运动目标掩模图像，从而分割出完整的运动目标。实验表明，该算法可以实时有效地将运动物体从视频序列中自动地分割出来。

英文摘要：

As the commonly-used methods of video segmentation are easily affected by noise and light, it is difficult to extract a complete moving target＇, so a new video object segmentation algorithm combining edge detection and inter-frame difference is proposed. Firstly, the difference of three consecutive frames is used to obtain motion regions, and then the boundary of current frame is defined by using Kirsch edge detection. The two test results are synthesized to get the.more accurate edge of moving objects. Edge linking algorithm is used to connect the breaking edge. Finally,the template of moving object is formed through area filling, and thus a complete moving object is obtained. Experimental results show that this algorithm can effectively segment accurate moving objects in real time.