The characteristics of the ocean background and the ship target in high-resolution SAR images were analyzed. A two-stage fast detection algorithm for ship target detection in high-resolution SAR images was proposed. At the first stage, an improved spectral residual visual salience calculation model was used to quickly obtain the visual region of interest. At the second stage of the detection, we designed a local maximum posteriori probability classifier for pixel classification based on Bayesian theory in binary hypothesis testing. After the parameter estimation, the criterion was completed and the pixels in the significant region were divided into two categories to achieve the target detection. Experiments were carried out using Terra-SAR-X and a large amount of military satellite data. The results showed that the proposed algorithm has good detection performance and is more in line with the application requirements of the actual high-resolution image ship target detection. By comparison with the conventional detection algorithm, it showed that: the algorithm proposed in this paper can not only reduce the false alarm caused by speckle noise, but also improve the detection speed by 25 percent to even 50 percent.