Fast Motion Estimation using Evolutionary Strategy Search Algorithm

摘要

Evolutionary strategy search (ESS) algorithm is a novel method for implementing fast block motion estimation (ME) using evolutionary strategy (ES). It uses a combination of ideas based on existing ME algorithms and employs a novel (1+deltalambda) ES implementation. ESS is essentially a succession of random searches, but by controlling the placement and distribution of these searches in a simple way, it proves possible to achieve comparable motion vector (MV) accuracy to the more established ME algorithms, but with enhanced convergence speed especially for large frame size, high pan and high zoom video sequences