Segment And Handle The Moving Athlete Objects In Diving Video Sequences

Currently, video-based human motion analysis is a very active research field, it human-computer interaction, intelligent control, sports performance analysis, content retrieval and other fields have a wide range of applications. Video of human subjects to separate out from a dynamic scene, all kinds of follow-up high-level video processing and application base, including object classification and activity recognition, moving object tracking, event detection, etc., are very important.Papers from the global motion estimation, spatial frame segmentation, joint space-time segmentation discussed several aspects of the development of domestic and international status of the video segmentation. Diving video for moving object segmentation in the human body, this paper proposes to use color information for moving target detection and segmentation algorithms. First use of hue and brightness of the two color components to enhance the image of the color difference, OTSU algorithm using image segmentation, but also with athletes dress is simple, relatively small and the characteristics of skin occlusion RGB color space, color heuristic clustering determine the connected region where moving targets, thus completing the first frame automatically detects moving targets. In the subsequent frame processing, mathematical morphology for adaptive sports regional forecast, the use of improved OTSU algorithm to improve the segmentation speed.Experiments show that this method can effectively overcome the impact of changes in complex background, to quickly achieve target segmentation diving, and the rapid movement of the moving objects are more robust.