Scene Change Detection in a Television Video

Abstract

Scene change detection is very important in video application, including video indexing, semantic features extraction, multimedia information systems, video on demand (VOD) and digital TV.  In this work we propose automatic scene change detection algorithm in broadcasted video. The proposed technique uses audio feature - signal energy, energy difference and visual features- frame difference, histogram difference, edge change ratio, face similarity. Experimental results on videos demonstrate better results for scene change detection.