基于视频处理的雷达图像目标检测与跟踪的研究

【作者】 徐蔚然；

【导师】 孔祥维；

【作者基本信息】 大连理工大学 ， 信号与信息处理， 2000， 硕士

【摘要】 信息时代下的现代战争是高科技条件下高速度的信息战。迅速发现并识别目标，是决定战争胜负至关重要的因素。因此自动目标检测、跟踪和识别已成为现阶段和未来武器系统的一个重要组成部分。本文主要研究基于我国的非相参雷达图像的舰船运动目标检测、跟踪和识别的具体方法。本课题的内容不仅适用于军事领域，也适用于民用领域。 本文在参考国内外相关文献的基础上针对非相参雷达图像的特性制定了舰船运动目标检测、跟踪和识别的系统框架，并给出了具体方法。本文的工作主要有以下几个方面的内容：1）分析雷达图像的特性；2）提出了模糊高斯方向平滑，还比较了不同图像分割方法的处理效果；3）用光流实现跟踪匹配；4）用相位相关法估计背景位移；5）给出了算法实现中用到的目标数据结构；6）提出用多分类器综合集成实现目标分类的方法。更多还原

【Abstract】 Modem warfare in the information age is information warfare under high technology condition. Detecting and recognizing target in due course is one of the most important factors which determine who will win the war. For this reason, automatic object detection, target tracking and automatic target recognition (ATR) have become an important part of weapon systems of today and tomorrow. In this paper, we mainly discuss the methods about the automatic object detection, target tracking and AIR of ship based on China抯 non- coherent radar. This paper抯 content not only can be applied to military field, but also can be applied to civil field. Aimed at characteristics of non-coherent radar抯 image, this paper has made a system frame of automatic object detection, target tracking and AIR of ship based on related literature inside and outside of China, and presented the specific methods. This paper抯 major content includes the followings: 1) analysis of the characteristics of radar抯 image;2) putting forward fuzzy Gaussian smoothing ,and comparing the results of different image segmentation methods;3)realization of target tracking using optical flow;4)estimate of background displacement with phase-correlation method; 5 )presentation of data structure of targets;6) realization of target classification with multiple classifier integration.更多还原