面向探月卫星视觉融合定轨的月面适配性研究

项目名称： 面向探月卫星视觉融合定轨的月面适配性研究

项目编号： No.41204026

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 地球物理学和空间物理学

项目作者： 卜彦龙

作者单位： 北京航天飞行控制中心

项目金额： 25万元

中文摘要： 以探月卫星视觉/无线电融合定轨为背景，基于嫦娥任务真实图象和轨道数据，并结合卫星轨道和地面测控约束，采用将模式分类与多属性决策相融合的基本策略进行月面景象区域适配性问题综合分析，拟构建一套高可靠的月面景象适配区辨识方法。首先结合卫星视觉辅助定轨原理、月面光照与成像特征、多时相月面匹配特点，以及轨道和测控约束条件，对月面景象区域适配性问题进行综合描述，建立理论框架，将月面景象区域适配性问题分解成基于像面信息的综合适配特征构建和考虑轨道及测控约束后的综合决策与评价准则构建两个基本问题；通过若干基本特征的进化学习合成像面综合适配特征并进行可靠性测试；将训练得到的月面像面适配特性，与卫星轨道与地面测控条件等约束联系在一起构建月面景象区域适配性多属性综合决策模型，并进一步得出景象适配区综合辨识方法。本课题力求拓展我国嫦娥卫星实测数据的应用，为深空精确定轨和精确定点软着陆提供适配区辨识方面的理论指导。

中文关键词： 深空探测；视觉导航；适配性；信息融合；图像处理

英文摘要： On the background of vision/radiowave fusion based orbits determination for lunar probes, a set of suitable matching-area identification method for lunar surface with high reliability will be studied based on Chang E actual measured data and constrains from statllite orbits and ground tracking, in which the problem of lunar surface-matching suitability will be analyzed synthetically under fusion of pattern classification and multi-attribute decision-making. Firstly, a theoretical framework of lunar surface-matching suitability will be constructed based on synthetical description combining vision aided orbit determination, lunar surface lighting and imaging, multi-temporal image matching, and constrains from orbit and tracking. Through the framework, the problem is divided into two sub-problems, which respectively are construction of image-information-based synthetic features and build of constrains-based decision-making and evaluation principles.After design of several primary suitability features according to characteristics of image textures, imaging and matching algorithm, a genetic-based optimization searching algorithm will be established to find synthetic suitability feature with the highest fitness. The selected synthetic feature, combining with constrains from orbit and tracking, will be used to construc

英文关键词： deep space exploration；vision-based navigation；matching suitability；information fusion；image processing