The Video Object Extraction Method Based on the Intersection of Three Times Frame Change Edge

摘要

In order to eliminate the negative effects caused by noises, complex motion and background exposure in the process of detecting and segmenting video object using change, a new method based on the intersection of three times frame change edge is proposed. In this method, we use the difference of two k intervals frames instead of two successive frames, thus the articulation motion and slow motion can be processed properly. Besides, the background culling and exposure caused by the motions of video objects can be effectively eliminated by using the edge intersection of three times frame change, and the negative effects due to background noise can be eliminated at the same time. Then, the contour can be obtained by scanning all the rows and columns of the intersection, and the breakpoints of the contour is need to be repaired, the binary image is formed by filling the contour. Finally, mathematical morphology processing is performed on the binary image obtained from the previous steps in order to get the object segmentation result. The Experiment results show that video objects can be segmented with the new algorithm accurately and automatically.