Study On Change Detection Method Of Concealed Target Detection Using UWB SAR Based On The Ratio Operator

At present,with the increasing requirement of military on detecting concealed targets,as a new radar system,Ultra-Wide Band(UWB)Synthetic Aperture Radar(SAR)has a great potential on detection of concealed targets.Using dual-temporal UWB SAR images to change detection has been proved to be comparatively precise on ground vehicles indication.Comparing with other detection methods,it has advantages of low false alarm rate and high detection rate.The main approaches to detect concealed targets using UWB SAR images are based on registered dual-temporal image ratio operator or Log-Ratio(LR)operator to conduct Constant False Alarm Rate(CFAR)detection.The key point of the CFAR detection is the establishment of background statistical model.As a matter of fact,existing methods often adopt the Gaussian distribution.However,the Gaussian distribution is difficult to accurately describe the statistical characters of LR images due to the interference caused by noise and imaging properties.Some investigations consider the Generalized Gaussian(GG)has stronger ability to model the LR images in the researches on change detection of SAR images.Nevertheless,the deduction to CFAR detection algorithm is still difficult.Meanwhile,there is no further investigation on the issue whether the GG distribution could accurately describe the statistical characteristics of UWB SAR LR images.Additionally,the critical step to change detection on multi-temporal UWB SAR images is how to determine the exact model for LR operator.However,there is no appropriate theory on Probability Density Function(PDF)and parameter estimation about LR statistics till now.Based on the above analysis,firstly,the paper applies the GG model to the concealed targets detection of UWB SAR,and derives the CFAR threshold corresponding to GG model,which verifies effectiveness of the proposed method and obtains precise detection of concealed targets comparing with common CFAR detection on Gaussian distribution.Then,departing from the imaging characteristics of UWB SAR images,the PDF of the LR operator is theoretically derived and we utilizes maximum likelihood(ML)estimation to estimate the PDF parameters.Incorporating with threshold given by PDF,the paper accomplishes the concealed targets detection in UWB SAR images.The proposed model has higher fitting precision compared with CFAR detection based on GG model,thus it improve the performance of detecting concealed targets.