Research On Parallel Processing Of SAR Image Change Detection

Synthetic Aperture Radar (SAR) has been more widely applied to military andcivilian fields, The major advantage of SAR is its all-weather capability that allowsthe acquisition of time series of imagery with exact acquisition dates under allclimatic conditions. One of the major applications of SAR is change detection. It is aprocess that analyzes a pair of images acquired on the same geographical area atdifferent times in order to identify changes that may have occurred between theconsidered acquisition dates. Because of the computation challenge from a largenumber of SAR image, serial processing based on single computer is difficult tosatisfy rapid processing. So we need to develop a parallel algorithm which fullyintegrates the advantages of software and hardware to satisfy the actual requirement.In this paper, a parallel processing program of change detection based on PCAalgorithm is proposed on the multi-core processor VPF2. The contributions andrelevant work in the paper are as follows:1.Firstly, This paper introduces the concepts of SAR change detection, includingbasic steps of change detection, classical algorithms and some problems existing inthis field, proposes a SAR image change detection algorithm based on PCA. Differentchange detection methods are used to detect the real image and simulative image onMATLAB. Analyzing and comparing the detect results their indicate their advantagesand disadvantages.2.Secondly it introduces the related concepts of parallel processing. Combiningwith the characteristics of change detection algorithms and rich hardware resources onVPF2platforms, a parallel processing program of change detection based on dataparallel model is proposed.3.In Chapter4, it presents design, function and implementation of ChangeDetection Parallel Processing system based on C/S model, which composing by thePC running the client user interface and VPF2acts server processing algorithm. Thespeedup and efficiency of parallel computing are analyzed with the tests of parallelcomputing of change detection by the images of different sizes. Finally, the mainwork of this paper is summarized and the shortages of our work are presented andpersonal opinions about this study field are given.