Global motion video segmentation based on the change of integral brightness in rows (columns)

This paper puts forward a novel method of global motion video segmentation based on the change of integral brightness in rows (columns) with the purpose of segmenting the different movements of some persons. The thought of the new method is that it calculates the brightness´s integration of the images in the video frame by the images´ rows (columns), and then analyses the human body movements through the tendency of the statistical brightness changes in all frames. In the video motion segmentation, we mainly segment the videos by the different frequencies of different movements. It´s obvious that the frequency of different simple movements is also not the same, so in this way, we can find the boundary of different movements and then distinguish the type of simple movements of the persons through the new method. Many experimental results have demonstrated the performance and feasibility of this new method. Compared with other video segmentation methods, the method come up with in this paper has smaller computation and is easy to understand and much easier to be realized. Experiments show that it´s a very practical and feasible algorithm in some tests that don´t have very high precision requirement and require high processing speed if the system configuration is lower.