Target Detection Based On Multilayer CFAR In Very High Resolution SAR Image

Statical modeling is essential to SAR(Synthetic Aperture Radar) image interpretation.Differences between the SAR statistical model(such as the K-distribution, Gaussian distribution, gamma distribution, the lognormal distribution, mixing distribution, etc.)on the differences between the SAR terrain type(such as farmland, forests, grasslands,rivers, etc.) vary. This paper first introduces a different statistical distribution and its characteristics, followed by statistical modeling of the differences between the SAR image terrain type, model the SAR image of different types, and to find what the terrain that is fit.Target detection is very important in SAR images. Then according to the vehicle detection problem in the airborne SAR images combined with CFAR algorithm, the target detection algorithm based on angular reflector scatting characteristics is proposed.We used the real airborne P band and L band SAR image data to prove the effectiveness of the algorithm.VHR SAR image has the characteristics with a large amount of data, the processing time using the traditional CFAR algorithm is complex and the target has form and details. According to these characteristics, we proposed a multilayer CFAR algorithm.The lognormal distribution model is used to fit the SAR image algorithm. The whole SAR image using CFAR algorithm logarithmic based on normal distribution is processed for filtering the strong scattering point to find the background. Then we can detect the ship targets on this processing SAR image. Although the multilayer CFAR algorithm can extract the ship target accurately, but there are still a lot of false targets.According to the prior ship size, we can filter the false targets after the processing by the multilayer CFAR algorithm. After Filtering out false targets, we will find that the targets are not integrity due to VHR SAR image characteristics. We propose algorithm for extract the outline. Then we can fill the outline to get integrity targets. we use two real SAR images to validate the effectiveness of our method. These images are 1m resolution in the Kaohsiung, Taiwan, China, a part of Terra SAR-X images. Our algorithm is compared with the traditional CFAR algorithm and the results prove that our algorithm has good effect. Finally, against the characteristics of VHR SAR mages in car detection and combined the multi-layer CFAR algorithm and subsequent algorithm,we detect the cars on the land.This work was supported in part by the National Natural Science Foundation of China under Grant 61072106 and 61271302; the National Basic Research Program(973Program) of China under Grant 2013CB329402.