Based On The DSP Array Image Fast Matching Technology Research

The traditional els attack uses recent detection technology to acquisite targetinformation, and most of the target information are one dimensional,whose informationare very small. Although in recent years, scientists successfully collect3d target imageinformation by using the onboard laser imaging system, and the goal information arevery abundant, but a lot of information extend the time of image processing, and thereal-time performance is bad. Basd on this reason,the paper puts forward an array ofimage matching processing DSP technology.By using this method, the bottleneckproblem of processing time can be solved effectively and els attack attitudeparameters can be calculated quickly.At last, transient posture parameters of targetimage will be used as the basis for directional detonation determine detonation directionand achieve more effective attack against target.The main study content of this paper is as follows:(1) Here this paper analyzes the image matching algorithms which are usedcommonly,such as Haussdorff matching,Euclidean matching horse matching,normalization of martensite matching,SIFT matching and so on. According to imagedeformation problems such as scaling,rotation, scale, the paper simulates and verifiesthe above matching algorithm. By comparing the performance of this several matchingalgorithms,we can draw a conclusion that SIFT matching algorithm is the bestperformance.(2) Design and implementation SIFT matching based on single DSP and parallelSIFT matching algorithm based on DSP array. On the basis of in-depth study of SIFTalgorithm principle and in the light of the characteristics of DSP array architecture,thepaper analyzes the implementation process of SIFT matching algorithm and putsforward DSP array data parallel processing algorithm. Through the reasonablesegmentation image, four DSP cores can parallelly extract image features; When theimage matching feature points, each DSP core is a master node in turn and looks forreal-time image matching points. Through DSP array simulation,we can make aconclusion that image matching speed is improved.(3) Focus on image parallel fast matching algorithm based on DSP array of applications,and provides technical support for the target attitude quickly calculationand warhead real-time adjustment.