Video shot similarity algorithm based on spatio-temporal slices

Video shot similarity has wide foreground in fields of content-based video data clustering, indexing, retrieval and summary, browse and so on. Based on available researches, a spatio-temporal slices-based video shot similarity algorithm is provided in this paper. While traditional approaches to video sequence analysis tend to formulate computational methodologies on two or three adjacent frames, in our approach, the spatio-temporal slices are employed for video content analysis and representation. After constructing video volume from compressing video data, we obtain spatio-temporal slice images of video volume in different directions, and segment the video sequence into shots based on edges and textures of spatio-temporal slice,images. Then we develop the shot similarity algorithm based on spatio-temporal slice images of shots, which calculates weighted average of dominant color percentage of histogram similarities of different direction spatio-temporal slice images as shot similarity. Generally, vertical or horizontal spatio-temporal slice images are adopted to be shot similarity. Video clustering experiments show that the proposed algorithm has satisfactory performance of correctness and high efficiency.