Infrared ship multi-targets recognizing

The technique of ship multi-targets recognizing is very important to anti-warship missile imaging guidance. In this paper, an algorithm is suggested to distinguish a ship target from the other ones in infrared images. The segmentation images by thresholds - and by a mask of L inverted is used to draw ship targets out of infrared image. The rates of frontier in each region and gray-level classes are also utilized to merge regions. Four features are defined to describe infrared ship targets. A criterion based on BP neural network with these four features is set up to distinguish one ship target from the other ones. A complete algorithm is presented. This algorithm was tested in simulating platform with twenty-five images. The results showed this algorithm is valid when the side of the ship faces the camera. The further studies are needed to do for more situations.