GAOFEN-4 is a high-resolution optical remote sensing satellite on geosynchronous orbit, which has been found the potential to monitor the moving ships on the sea. Some classic moving ships target detection algorithms are studied. A new algorithm is proposed for the sequence images of GAOFEN-4 satellite, which can effectively suppress the interference of noise on moving ships target detection by performing the defogging of dark channel prior and a moving detection with visual background extractor. Some experiments are executed with real sequence images. It shows that the new algorithm is better than classic target detection algorithms for the sequence images.
With high time resolution and wide coverage, GF-4 camera can achieve high-frequency repeated monitoring over a wide range of areas within a certain period of time. However, there are some problems of its sequence images, such as shaking, non-prominent ship targets and insufficient brightness in sea area, which are not conducive to the detection of moving ships. Thus, a hierarchical registration algorithm combining SURF and SIFT, and an image enhancement algorithm combining wavelet threshold denoising and Top-hat filtering was proposed to solve these problems. Matlab’s experiment proves that the processed image can almost only contain the highlighted target and the black background, which improves the accuracy of subsequent ship monitoring.