Adaptive Shot Change Detection Technique for Real-time Operation on PMP

Shot change detection is the main technique for automatic temporal segmentation of video. Until now, there were little research reports about real-time shot change detection for applying to hardware terminals with low performance such as PMPs (Portable Multimedia Player) and cellular phones. In this paper, we propose an adaptive real-time shot change detection algorithm using sub-sampling, weighting variance, and an adaptive threshold. We use weighting variance of frame difference for dividing feature extraction at a changed frame. We set an adaptive threshold using the average of variance from the next one of a last shot change frame to the previous one of a current frame. In the experiments, we obtained the detection rate of about 94.4% in overall accuracy. We verified real-time operation of shot change detection by implementing our algorithm on the TVUS HM-900 PMP from the company of Homecast. Therefore, our proposing algorithm will be helpful in searching video data on portable media player such as PMPs and cellular phones.