近距离空间未知目标识别跟踪和精确测量理论与方法

项目名称： 近距离空间未知目标识别跟踪和精确测量理论与方法

项目编号： No.60872079

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 张弘

作者单位： 北京航空航天大学

项目金额： 31万元

中文摘要： 该结题项目主要研究近距离空间未知目标信息的识别跟踪和精确测量的理论和方法。针对近距离未知目标识别及精确跟踪测量技术所涉及的关键问题进行系统研究。从主动飞行器接近、停靠目标的实际出发，根据运动过程中的目标成像特点： 1） 针对特征退化的这一普遍现象，提出特征遗传理论方法，解决变特征失效判断、继承等问题； 2） 针对目标外形等先验信息未知的情况，研究了变分辨率多尺度提取新方法，自动搜索并构建测量点； 3） 研究了具有扩展性的三维建模理论方法，为解决目标姿态测量和超视场问题奠定了基础； 4） 为解决精确测量和定位问题，研究了目标超视场情况下基于模型的对准理论方法。 特征遗传、可扩展三维建模和基于模型的对准，其相关理论方法不仅可以解决"近距离未知目标识别跟踪和精确测量"中的关键问题，还具有普适性，能为其它应用中的姿态测量、定位导航提供理论依据和技术支撑。

中文关键词： 未知信息空间目标;特征退化;特征遗传;可扩展三维建模;基于模型的图像对准技术

英文摘要： This project mainly focuses on theories and methodologies of recognition, tracking, and precise measurement of unknown targets in the near space distance. Critical problems of the above topic is studied systematically, based on the characteristics of the approaching phase of the flying target and its imaging characteristics during the motion. Four problems are resolved: 1) Aiming at the universal phenomenon of feature degeneration, a feature inheritance theory and methodology are proposed to solve the invalidity judgement and inheritance of a feature. 2) A new method on multi-resolution/multi-scale feature extraction is presented to automatically construct measuring points and to solve problems caused by the unknown target shape. 3)Three-dimensional modeling theory is studied and an extensible methodology is presented to measure the attitude of the object and deal with objects out of view. 4) A model-based registration theory is presented for precise measuring and localization of the object, which works when the object is out of field of view. The feature inheritance, extensible 3D modeling and model-based registation methods can not only solve the key problems in the recognition, tracking and precise measurement of near-distance unknown targets, but can also provides theoretical basis and technical support on other application of precise measuring and localization.

英文关键词： Space objects with unknown information; feature degeneration; feature inheritance; extensible 3D modeling; model-based image registration