A principled approach to fast partitioning of uncompressed video

Abstract

One basic technique often used to reduce the huge computation involved in video partitioning is by selective processing of a subpart of the video frames. However, the choice of this proportion has so far been made randomly, without any formal basis. An ad hoc selection of this subpart cannot always guarantee a reduction in computation while ensuring effective partitioning. The paper argues for a principled basis for the selection of such parameters to ensure effective and efficient partitioning of video sequences. It presents a formal method for determining the optimal window size and the minimum thresholds which ensure that decisions on scene similarity are made on a reliable, effective and principled basis. To account for illumination variations and possible motion in the video, the use of neighbourhood-based colour ratio features is proposed, which avoids the huge computational requirements of explicit motion compensation operations.