Shot Boundary Detection Algorithm Based on HSV Histogram and HOG Feature

Abstract

Spatial information can not be contained in the statistical feature of hue-saturation-value(HSV) color histogram, moreover, which is not robust for the videos with the similarity of background information and affected by illumination. Therefore, a new video shot segment method based on HSV histogram and Histogram of Gradient (HOG) feature is proposed. Firstly,HSV color histogram is used to detect the difference between two adjacent frames which is then compared with the adaptive threshold. After the initial detection of the shot boundary is finished, HOG feature of the video is adopted to make further shot boundary detectionwhich eliminates the wrong shot boundaries and adds missed shot boundaries. Experimental results indicate that the proposed method achieves higher accuracy and recall.