In image processing, the detection of multiple infrared ship targets against the background of sea/sky and island is always difficult. The monitoring of multiple ship targets usually uses a wider field of view and a greater field depth. Although this can cover more target information and more imaging pixels against the sea/sky background, the extraction of saliency of multiple ship targets becomes more difficult. At the same time, the increase of field depth makes the ship targets much smaller and their outlines are no longer clear. This may cause great difficulty for the saliency detection of ship targets. By applying the theory of image level diversity and superpixel to the saliency detection of multiple ship targets at sea, a new saliency detection method for multiple ship targets at sea is proposed.