视频对象分割算法的性能好坏将直接影响MPEG 4编码产品的质量。连续两次差分后自适应处理,对差分图像取交集获得运动对象的边界,形态学处理后最终获取运动目标。基于改进的Hausdorff距离度量法对后续帧中视频对象进行跟踪。实验结果证明,该方法能够从背景不变的图像序列中较好的提取出运动对象,具有较强的鲁棒性。 The performance of video object segmentation algorithm will directly affect the quality of MPEG 4 coding products. Two consecutive differential adaptive processing, the intersection of the differential image to get the boundary of the moving object, morphological processing and finally get the moving target. Tracking video objects in subsequent frames based on improved Hausdorff distance measure. Experimental results show that the proposed method can extract moving objects from the invariant image sequences and has good robustness.