Research On Detecting And Tracking Of Small Moving Targets Under Complex Sea Background

Moving targets detection and tracking related to computer image processing, video image processing, pattern recognition, artificial intelligence and other areas, is widely applied in many aspects such as military, industrial, life, and so on.This paper mainly focuses on moving targets under the dynamic background, especially under the complex sea background in the visible light and IR. The study is divided into three parts: image preprocessing, the moving target detection, and tracking and forecast of moving targets.In the image preprocessing aspect, sea-sky background segmentation methods have been summarized and compared. Considering the sea-sky-line is a straight line, a new combined sea-sky-line detection for both visible light and infrared images is presented. First pre-segmentation is accomplished by median filtering and gray extending union maximum variance between classes. Then Hough transformation is applied for line detecting based on the edge detection. The experimental results show that this method can detect the sea-sky-line well in the condition that the background complexly changing and low SNR.In the moving targets detecting aspect, focusing on the grays distributing characteristic of target image and the shortages of Otsu, we advanced a method that optimum threshold based on valley auto searching which improves the detection speed. It combines Median filter and Gaussian smoothing to suppress background, especially the sea-clutters and high gray sky background. It reduces calculation and improves the efficiency. And with morphologic filtering , the target could be detected. With the help of sea-sky-line detection, centroid of area detection is used to orientation. The experiments show that this method has better effect ion both on visible light and infrared images.In the moving targets tracking aspect, meanshift and camshift have been introduced and the experiment results have been compared. An advanced method has been brought forward at last. It could resolve the target losing problem in camshift when the target meets the great shelter or warped with same color object which is not small.