不同传感器图像的景象匹配区选取准则研究  

Research on Selection Criterion of Scene Matching Field for Different Sensors

作　　者：李俊山[1] 李旭辉[1] 罗蓉[1] 谭园园[1] 

机构地区：[1]第二炮兵工程学院计算机与指挥自动化系,陕西西安710025

出　　处：《微电子学与计算机》2007年第1期210-213,共4页Microelectronics & Computer

摘　　要：景象匹配区的选取工作是影响光学景象匹配系统性能的重要因素,在分析影响景象匹配算法的配准概率和定位精度等因素的基础上,人们已经提出了一些实用的选取方法。文章首先对这些方法进行了研究,然后,从选取准则的角度对影响景象匹配区选取的因素进行了归纳分析,通过实验发现利用这些准则在进行实时图为红外图像、基准图为可见光图像的景象匹配区选取时存在一些问题,文章针对这些问题提出一种基于图像边缘密度与自匹配系数的景象匹配区选取准则。实验证明,该算法既保证了所选匹配区中的边缘信息量,又确保了所选区域具有较好的不相关性,采用粗选加精选的方式提高了选取速度。The selection of the scene matching field is an important factor to affect the optical scene matching system. Some practical selection method has been presented. Firstly this paper researches them and summarizes some selection criterions. Through the selection experiments based on infrared real-time image and optical reference image, some problems are found, then a new selection criterion based on edge density and self-matching coefficient is presented to resolve them. Through the experiments it is proved that this arithmetic is more effective than the others.

关 键 词：景象匹配 选取准则 自匹配系数 边缘密度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]