A two-stage shot boundary detection framework in the presence of fast movements: Application to soccer videos
This paper describes a shot segmentation algorithm for online video event detection in broadcast news. The algorithm is based on image detection contour with an adaptive threshold. The performance of this algorithm is evaluated in terms of shot boundaries detection. The relevance of the boundaries detected is also evaluated in order to preview the impact in higher levels of video analyses such has scene detection and semantic events detection (anchor, war, sports). The experimental results, shows this is a good approach for online video event detection.