Fusion Recognition Method of Target's SAR Images Based on Modified D-S Evidence Theory

The method of geometric hashing technology can effectively recognize the targets distorted partially. But when the known targets in training data set don't satisfy with the condition of 360 azimuths, the effect of recognition degrades. In this paper, we present two aspects improve the correct rate. Firstly, we use a CFAR detector based on power transformation to segment the SAR images. And also discuss the interval of power transformation' parameter when the data is Rayleigh distribution. Secondly, we present a method of modified D-S evidence theory which can solve fusion with a high degree of conflict. Experimental results with MSTAR dataset show that this fusion method is effective and feasible

[1]  Bir Bhanu,et al.  Multiple Look Angle Sar Recognition , 2004, Int. J. Image Graph..

[2]  Keinosuke Fukunaga,et al.  Introduction to Statistical Pattern Recognition , 1972 .

[3]  Reuven Meth.  Target/shadow segmentation and aspect estimation in synthetic aperture radar imagery , 1998, Defense, Security, and Sensing.

[4]  S. Quegan,et al.  Understanding Synthetic Aperture Radar Images , 1998 .

[5]  Sérgio Cavalcante,et al.  An extended approach for Dempster-Shafer theory , 2003, Proceedings Fifth IEEE Workshop on Mobile Computing Systems and Applications.

[6]  J. Fleischman,et al.  Foliage attenuation and backscatter analysis of SAR imagery , 1996, IEEE Transactions on Aerospace and Electronic Systems.

[7]  J. Fleischman,et al.  Analysis of foliage-induced synthetic pattern distortions , 1996, IEEE Transactions on Aerospace and Electronic Systems.