Based on visual saliency theory and local probability density function statistical feature, a target detection algorithm for SAR image is proposed. Local probability density function statistical feature reflects the difference between target and clutter on human vision. According to local probability density function statistical feature, saliency map of SAR image could be calculated by using hypothesis testing theory and Bayes theorem. Then target detection result could be acquired from saliency map by binary segmentation. For different kinds of real SAR images, target detections are implemented by the proposed algorithm and CFAR algorithm. The comparison of the detection results shows that the proposed algorithm detects all size-fixed targets with lower false alarm rate than CFAR algorithm.