A Low Complexity Adaptive Motion Estimation Algorithm Based on Evolution Strategies

WANG Hui,MAO Zhi-gang
DOI: https://doi.org/10.3969/j.issn.1006-8961.2005.07.012
2005-01-01
Journal of Image and Graphics
Abstract:Based on evolution strategies(ESs) a novel adaptive motion estimation search algorithm is presented in order to improve the encoder quality.The mutation operator in ESs is used with frequency corresponding to the normal distribution law.This algorithm applies the ES algorithm to block motion estimation,the motion direction factor participates in motion vector computing as a variable for the first time,and affects the whole search process,neither just being an implicit factor nor a predictive measure.The adaptive schemes are advanced in the step length control and population sizing.Experimental results demonstrate that this algorithm has similar performance to that of the full search algorithm,and owing to its inherent parallelism and low complexity it is suitable for VLSI implementation.
What problem does this paper attempt to address?