In order to solve the problem that majority of ship detection algorithms show the low accuracy and slow rate, a fine rapid ship detection algorithm based on salient feature guidance is proposed. First, the candidate target area is located through visual saliency model based on the local and global integration, and candidate target slices are gained through the region extraction. Then, the improved means clustering method is used to divide the target slice into super pixels. Finally, ship target regions are gained through filtering super pixels belonging to the target to achieve a fine segmentation after the integration of a significant figure and super-pixel segmentation. Experimental results show that the algorithm proposed in this paper can quickly and accurately locate the target ship, accurately depict the object contour, and is more conducive to subsequent follow-up work.