Scene change detection based on sequence statistics using structural similarity
Jun Shen, Xi Jiang, Jinshan Zhong, Shimeng Yao · 2022
With the development of the short video, video processing technology is becoming more and more important, among which video scene change detection is an important basic technology. However, due to the inconspicuous feature of scene gradual change, the current algorithm cannot detect it very well. Based on this, we proposed a new scene change detection algorithm. The new algorithm is designed based on the structural similarity (SSIM) algorithm, an index to measure the similarity of two images, and the changing law of structural similarity sequence during scene change. Meantime, we designed a new measurement based on sequence statistics to measure the variation law of SSIM sequences. Experiments show that the algorithm has the characteristics of anti-interference and anti-noise, and it can well solve the above difficulty.