灰度投影结合Fisher准则的对象分割

视频对象分割是基于内容的视频编码和视频检索中的的对象跟踪方法,提出了一种可以从复杂场景中分割出MPEG-4的视频对象新方法.首先采用灰度投影匹配进行全局运动估计和补偿,用以消除背景变化的影响;然后由二次差分抽取中间帧解决遮挡问题,通过Fisher评价函数结合数学形态学填充得到运动对象分割掩膜,同时消除残余噪声以及平滑边缘.实验结果表明,该方法在一定范围内较好地解决了遮挡问题,并能够高效快速地得到比较精确的视频对象.

Automatic segmentation for video objects based on gray projection and Fisher criterion

Segmentation of video sequence is the key issue in video coding standard MPEG-4 and video retrieval system. A new method for video objects segmentation from different scenes. In this method, the method of the fisher criterion function image thresholding and tracking method based on the motion estimation and compensation are combined. Fistly, the effect of background change is resolved by making use of gray projection for motion estimation and compensation. Secondly, a dual frame difference operator is used to solve the problems of occlusion. Thirdly, using fisher criterion function to threshold image and morphologic method to fill and smooth the gained binary mask. So the accurate contour of moving object could be obtained. Experiment results show that the proposed algorithm is robust, fast and efficient in result of segmentation video object.