Identifying Video Segment of the Same Scene Based on Occurrence Pattern of Flashes

摘要

This research studies a method to identify video segments of the same scene which aims at indexing of enormous video data. These segments may be taken from different viewpoints and have different offset in time, so that we use the occurrence pattern of flashes to identify them. This method is exact because the pattern is common despite the difference of viewpoints, and is suitable for the detection of such segments from enormously stored video data because it needs relatively a little computing cost.