针对海上移动目标的监视需求,分析实时检测系统的技术要求,提出了一种利用视频卫星图像实时检测海上目标的方法。采用机器学习和图像滤波技术,研究了基于支持向量机的海陆分割、基于自适应滤波的目标粗检测和基于AdaBoost的目标精检测等算法,使用某视频卫星图像对方法进行验证,结果表明该方法能够实现海上目标的检测功能和实时检测系统的技术要求。

According to the monitoring needs of moving targets on sea surface, the technical requirements of real-time detection system are analyzed. A method of real-time detecting sea surface targets is presented by use of video satellite image. The technology about machine learning and image filtering is putforward. The algorithms of land sea segmentation based on support vector machine, target rough detection based on adaptive filtering and target precision detection based on AdaBoost are studied. Finally, the algorithms are tested and verified by use of video satellite image. The results show that the detection functions of sea surface targets and the technical requirements of real-time detection system can be achieved by the algorithms.