A detection method for near duplicate video clips based on content similarity

In view of the complexity of video clips data,a method was designed for extracting features to represent video clips and using locality sensitive hashing to compute the frame match sequence between video clips,RANSAC was used to fit the match sequence,obtain the valid match sequence and compute the similarity between clips.Experimental results show that the method can effectively describe the similarity of web video clips as well as detect and eliminate near duplicate video clips.