本项目针对船载雷达和自动识别系统对海上弱小目标探测和识别能力差的问题，将视频处理与分析思想引入海上移动目标的探测，利用高分辨率远距离摄像机和热成像仪摄取船舶周边海域的视频图像，通过对船舶姿态的实时测量、海面背景特征的提取以及被探测目标运动特征的分析，对视频图像进行稳像处理，并对可见光视频和热成像视频进行融合，形成稳定的、对被探测目标特征表达更为丰富的视频图像，从而为海上移动目标自动检测、运动跟踪和分类识别等提供基础支撑。通过本项研究，可解决船载视频稳像和融合增强的核心理论问题，形成面向海上目标探测的船载视频全局运动估计算法、基于船舶振动测量技术的船载视频运动补偿算法、基于海面动态背景特征的船载视频全帧补偿算法以及船载多源视频融合处理算法等成套理论和技术。同时，通过研究、提炼和抽象，形成一种通用的视频稳像理论和融合处理算法，丰富目标识别和智能监控理论体系。

Due to the poor performance of ship-borne Radar and AIS (Automatic identification system) on small target detection, the idea of video processing and analysis is introduced by this project. To obtain stable video that expresses more features of detecting target, and support the detection, tracking, classification and identification of moving target at sea, this project uses high-resolution remote cameras and thermal imagers to shoot the waters surrounding the ship, stabilize the video image, and fuse the visible video and thermal video by measuring the attitude of ship real-time, extracting the sea background feature and analyzing the motion characteristics of target which is detected. Through the basic research of this project, the core theoretical issues of ship-borne video image stabilization and fusion would be solved, the theory and technology of global motion estimation algorithm for target detection at sea, motion compensation algorithm based on ship vibration measurement, full frame video compensation algorithm based on the dynamic sea background characteristics, ship-borne multi-source video fusion algorithm would be formed. Meanwhile, through exploring, refining and abstracting these issues, a new and general theory of video image stabilization and fusion processing algorithms would come into being, which would enrich the theories of object recognition and intelligent surveillance.