Ship Target Detection Algorithm Under Sea-sky Background

Due to the far distance of target imaging, so the target area is very small, the shapecharacteristic is really weak, and the detail characteristic is also almost lost. In addition,the air and sea background in general is relatively complex, not only with a few cloudsin the sky background, but also the rolling sea clutter on the surface of sea background.What’s more, the signal is very weak, it is easily covered by the noise andbackground.All of the above reasons has brought a lot of interference factors anddifficulty to the ship target detection. Under the complex background, the infrared smalltarget detection has gradually become a focus research both at home and abroad inrecent years. This paper has done a lot of research, as following:To some extent, there is an important relationship between the difficulty of targetdetection and the background. The detection becomes harder when the background ismore complex. First, the optical images under the background of sea and sky wereanalyzed. And then, the target, background and noise characteristics were discussed indetail, respectively. The focus of the image target detection was the backgroundsuppression. Due to the prior knowledge small ship targets presented in the sea-sky area,so this paper proposes a ship target detection technology based on the sea-sky line. Theproposed technology will extract sea-sky line, then split out the sea-sky region, finallyachieve detecting the target in the sea-sky area. There is important practical significancefor reducing the false alarm probability and improving the detection efficiency of shiptarget detection by extracting sea-sky line.The paper focuses on the analysis and research on the algorithm of sea-sky lineextracting. The paper mainly studied the linear method, and then the linear fittingmethod, Otsu method and average gradient method three extraction algorithms aremainly discussed in the paper, and at the same time the simulation results are compared.Several common image segmentation methods are adopted to detect the ship target inthe sea-sky area at the end of the paper, such as histogram method, the neighborhoodaveraging method, the Otsu method, the gradient threshold method,and the graygradient threshold segmentation method works best. Using the projection method tocomfirm a general ship target location, and combined with threshold segmentationresults determine ship target final location.Finally, the paper summarized the main work and the obtained fruit in the paper, analyzed the matter and expected the extended work.