Sea-sky Background Infrared Small Target Detection

As an important aspect in infrared target recognition, naval ship detection under the sea-sky background has high research value in the military field. Its performance will directly affect the accuracy of the subsequent target identification and target tracking.In this paper, a ship detection process with sea-sky background is proposed and the following three aspects are researched:coastline detection, naval ship with sea-sky background infrared image preprocessing and naval ship segmentation using fractal parameters constraints.As naval ships often appear in the zone near coastline, the detection of coastline is researched. Though Hough transform is effectively used in the detection of coastline, it has two drawbacks which affect the detection efficiency, in this paper; optimized Hough transform is proposed which can improve the detection efficiency while ensuring detection accuracy.This is the first work in which the scaling algorithm Seam Carving algorithm is improved for the detection of coastline. Seam Carving algorithm is an algorithm based on the image content, which can quickly find the unimportant part of the image, the improved algorithm can quickly locate the important part of the image which is the coastline. Experimental results show the effectiveness of algorithm and the low computational load.An optimized preprocessing method is proposed, which compared with the classic preprocessing algorithms performs better in enhancing the contrast of the image, inhibiting the complex background and removing noise. As infrared ship target images of the sea-sky background have characteristics such as low contrast, complicated background and strong noise, a pretreatment process which combines a variety of preprocessing algorithms is proposed in this paper. After contrast stretching, use the band-pass filter to suppress the background image, and finally use the mean shift filter to remove the residual noise.In this work, we proposed a novel process for ship target detection under sea-sky background. First extract corner points in the image by the Harris operator, then constraint the corner points extracted by the fractal dimensions, retaining only the corner points located on the ship target, so that we can get the location information of the ship target, at last use the region growing algorithm in vicinity of the corner points to segment the target from the sea-sky background.A ship target detection process with sea-sky background is proposed in this paper, firstly extract the area near the coastline by detecting the coastline, then preprocess the extracted image to improve the quality of the image, finally segment the ship target from the sea-sky background using the proposed method. The testing results show that the proposed detection process can achieve real-time effective detection of ship target with sea-sky background.