Fast matching pursuit for traffic images using differential evolution

FENG Xiao-qiang, HE Tie-jun

Intelligent Transportation System(ITS) Research Center,South East University,Nanjing 210096,China



Abstract:

To obtain the sparse decomposition and flexible representation of traffic images,this paper proposes a fast matching pursuit for traffic images using differential evolution. According to the structural features of traffic images,the introduced algorithm selects the image atoms in a fast and flexible way from an over-complete image dictionary to adaptively match the local structures of traffic images and therefore to implement the sparse decomposition. As compared with the traditional method and a genetic algorithm of matching pursuit by using extensive experiments,the differential evolution achieves much higher quality of traffic images with much less computational time,which indicates the effectiveness of the proposed algorithm.

Key words:  intelligent transportation system  digital image processing  matching pursuit  differential evolution

DOI£º10.11916/j.issn.1005-9113.2010.02.009

Clc Number:TP391.41

Fund: