ROBUST LEARNING-BASED TV COMMERCIAL DETECTION (ThuAmOR2)

Author(s) : Xian-Sheng Hua (Microsoft Research Asia, China) Lie Lu (Microsoft Research Asia, China) Hong-Jiang Zhang (Microsoft Research Asia, China)

Abstract : A robust learning-based TV commercial detection approach is proposed in this paper. In this algorithm, firstly several basic features that can be used to distinguish commercials from general program are analyzed. Next, a set of context-based features, which are more effective to identify commercials, are derived from these basic features. Then, each shot is classified as commercial or general program based on these features by a SVM learning and classification scheme. And last, the detection results are further refined by using scene grouping and some heuristic rules. The scheme is tested on 10-hour TV programs of various genres and obtained relatively high detection accuracy.

Menu