Image distance based ship detection using SAR images

A novel method is developed for ship detection in synthetic aperture radar (SRA) images, which is based on image distance computation techniques. Using a second-order hidden Markov mesh model to learn statistical models of images, one can obtain the distance of two images for the purpose of detecting ships. First, the features of an image can be extracted using a method that best matches its statistical model, which is related to dynamic programming. Second, given the state transition matrix and observation distributions within states, statistical distance between images based on the similarity of their statistical models can be estimated. Experimental results demonstrate that this ship detection algorithm can effectively enhance ship target as well as suppress speckle and has better detection precision and lower calculation complexity.