Commercial Detection in Program Videos

Detection of the commercials in TV videos is hard because the diversity of them puts up a rather high barrier to find an appropriate model. After some studies of existing commercial detection works, we try to deal with this problem through a robust TV commercial detection approach. Firstly a sets of basic features that facilitate distinguishing commercial from general program are analyzed. Then, the commercial detection scheme, which is more effective for identifying commercials, is derived from these basic features. Next, each shot is classified as commercial or general program based on these features by a pre-trained SVM classifier. And last, the detection results are further refined by some rules. Experiments show good results of proposed scheme on detection commercial in general program videos.