Automatic Detection Method of Ships Based on Shortwave Infrared Remote Sensing Images

Aiming at the problem of the ship detection with a low accuracy in the offshore and inland river scenes, a method based on shortwave infrared multispectral remote sensing images is proposed to realize water segmentation and automatic detection of ship. Based on the low reflectance characteristic of water area in the shortwave infrared frequency range, the water area is rapidly and accurately extracted from the images by using the threshold segmentation and morphological processing. Then, the image chips of candidate targets arc extracted by using the visual saliency model for searching the targets in the water areas. As for the possible existence of phony targets, the gray-scale distribution histogram is proposed to describe the characteristics of gray-scale distribution of the target chips, which arc combined with the gradient direction information to eliminate phony targets by the method of threshold constraint. The results show that the proposed method can efficiently detect the ship targets with different sizes in offshore and inland rivers. 279 candidate targets arc obtained after the saliency detection and 138 of 112 true targets arc detected after the target discrimination step. The false discovery rate is less than 6% and the recall rate is higher than 97%.