Abstract: The next generation of warships might be virtually invisible to the human eye. These can achieve by painting ships with meta-material. These materials can bend light around an object, making it appear to an observer as though the waves have passed through space. We have developed a method to identify these ships target with the help of image processing. Research on the extraction of ship target in multiple sea-sky backgrounds has significant value to improve the capability of imaging-typed sea navigation and marine traffic control systems. According to the imaging property of complex sea-sky background a reliable ship target extraction method proposed in this study. The general guideline is that getting the sea-sky division line as a priori knowledge and then the potential target area is determined by a discontinuous region of the sea-sky division line. Firstly, a local selective window filter is adapted to filter the image; secondly, eight directions Sobel operator edge detection method and gradient Hough transform are combined to extract sea-sky division line in the picture then a multi-histogram matching technique is adopted to remove the sea and sky background and thus ship target obtained from complex background. The experiments show that our method has the merits of robustness to noise, small computational complexity and stability.