A multistage fast motion estimation scheme for video compression

Abstract

This paper presents a novel multistage motion estimation (ME) scheme called content adaptive search technique (CAST). The proposed scheme consists of four stages: motion vector field (MVF) prediction, block-based segmentation, motion parameter extraction, and adaptive search strategy. Through pre-processing the MVF of the previous reference frame in the first three stages, CAST extracts the motion parameters for each region. The 4th stage is a combination of various techniques including MV prediction, search area decision and an adaptive fast search algorithm that is adjusted by a mathematical model for the block distortion surface (BDS). Experiment shows that the proposed scheme improves the visual quality, while yielding a faster speed, comparing with the other predictive ME algorithms.