Research On Ship Target Detection Algorithm Of Complex Environment In SAR Imagery

Synthetic Aperture Radar(SAR)ship target detection technology is widely used in marine military target tracking,battlefield environment reconnaissance,maritime strategic early warning and other fields.Research has become a frontier mission in the military field of various countries.Therefore,it is of great significance to carry out research on ship target detection technology based on SAR images.In complex environments,such as complex sea state areas,strong clutter edge environments,and multi-target environments,the detection performance of traditional SAR image ship detection methods is degraded.In this paper,this paper has carried out research on the key technology of SAR image ship detection in complex environment.The main research results are as follows:(1)This article deeply studies the key component of SAR image ship detection preprocessing: speckle filtering,aiming at the contradiction between the speckle noise suppression of traditional speckle noise filtering methods and the preservation of ship target edge texture,the traditional bilateral filter is improved,and an adaptively truncated-statistics-based bilateral filter for speckle reduction in synthetic aperture radar imagery is proposed.The performance of the algorithm is verified and evaluated using Terra SAR-X data.The results show that the algorithm effectively suppresses speckle noise while The edge and texture information of the ship target is maintained,and it has more superior performance than the traditional SAR image speckle noise suppression algorithm.(2)This paper deeply studies the principle and process of the Constant False Alarm Rate(CFAR)ship initial screening algorithm,analyzes various commonly used CFAR initial screening algorithms.In view of the problem of low initial screening rate of traditional ship CFAR methods in complex environments,a truncated clutter statistics based variability index CFAR detector in SAR Imagery(TSVI-CFAR)is proposed.This method can effectively improve the initial screening rate of ship targets in a complex environment,and the false alarm rate of the initial screening is low.The GF-3 SAR data and Envisat-ASAR data were used to prove the effectiveness of the algorithm.(3)This paper studies and analyzes various commonly used deep learning algorithms,and deeply studies the SAR image ship identification method based on deep learning,and applies Faster R-CNN network to ship identification of SARimages.Experimental results show that this method can effectively improve the accuracy of ship identification in complex environments.(4)Finally,this paper combines deep learning with CFAR preliminary screening method to design a set of complex environment SAR image ship detection software system.And for each module of the system: pre-processing,ship target preliminary screening,ship parameter extraction and ship target identification,the corresponding functions were demonstrated.