The Ship detection of SAR image using CFAR based on mixture models chosen adaptively

In order to solve the problem of ship detection in SAR image that Constant False Alarm Rate (CFAR) detection algorithm cannot deal with under all scene, and may produce some result not fitting,the paper uses CFAR detector based on finite mixture distributions models chosen adaptively：firstly the input image will be preprocessed to reduce the influence of object pixel to sea clutter; secondly mixture models chosen by studying are used to model each preprocessed image patch and compute the global threshold to get the detection result of the image patch; thirdly the pixels which are not regarded as target at the last step are modeled by mixture distribution to computer detection threshold again until all target pixels in SAR image patch are achieved; finally the paper perform postprocessing for the detection result of each image patch to eliminate false alarm and get true ship target. In order to reduce the false alarm. This method can not only obtain better sea clutter model, but also get more detail of the ship. The experimental result shows that the method of the paper is effective to.