Video Similarity Matching Algorithm Based on Spatiotemporal Feature

This paper targeted at the offset of brightness, abrupt intense disturbance and problems caused by video re-editing. Video Spatiotemporal Feature was defined to represent the content of video. The proposed Grads matching algorithm and improved Dynamic Time Warping algorithm(DTW) were introduced to video similarity matching. Experimental database includeed 80 advertising videos and 200 re-editing videos. Results demonstrate the effectiveness of the algorithm.