A Novel Video Shot Segmentation Based on Textural Features

Abstract

In this paper, a new method for detecting shot boundaries in video sequences using Box-Counting to extract texture feature and judging the shot boundaries by an improve method to get a dynamic threshold is proposed. Many techniques have been developed to detect the video shot boundaries. But automatic shot boundary detection is difficult. In particular, gradual transitions are generally more difficult to detect. So the paper use the Box-counting method to extract the texture feature. Then the paper use the improve method to get the dynamic threshold to detect the shot boundaries. The dynamic threshold is the average frame difference in a slide window multiplied by a threshold coefficient. Experimental results successfully validate the paper method and show that it can effectively detect both the cuts and gradual transition. In particular, the performance is better than others method when the gradual transition shot boundaries are detected.