Abstract: Using sonar for underwater target recognition,distance measurement and direction finding is one of the important methods for current underwater target recognition and tracking.Sonar image is seriously influenced by noise and the resolution is low,background modeling for sonar image is helpful to the target segmentation and recognition. Therefore,the gray statistical characteristics of sonar image background region are analyzed,and then combined with the features of underwater sonar image,the Gaussian distribution,gamma distribution,Weibull distribution,and Rayleigh distribution models are used to fit the statistical characteristics of sonar images with six kinds of different background regions.The experimental results are evaluated by the χ² criterion and Kolmogorov distance error evaluation criterion.The results of comparison show that the Gauss distribution,Gamma distribution and Weibull distribution could better approximate the gray statistical distribution of sonar image background region.In order to meet the requirements of real-time applications,the use of the Gauss distribution model as the description of background area gray statistical distribution is feasible and reasonable,this provides a theoretical basis for sonar image preprocessing and target segmentation.