Automatic cut and camera operation detection for video

Abstract

Video should be segmented into shots to be used for identification and indexing. Various changes may happen within a single shot such as those related to camera operations, object movements, flashes, etc. Various similarities between different shots may also occur. Therefore, techniques based on interframe comparisons may result in detecting false cuts or missing true cuts if the appropriate threshold has not been chosen. In this paper, we propose a detection algorithm that understands and detects various changes within a shot in order to detect the cuts between separate shots.