高分辨率卫星影像物理模型与有理函数模型转换  

Rational polynomial coefficients generation and physical sensor model recovery for high resolution satellite stereo imagery

作　　者：童小华[1] 刘世杰[1] 

机构地区：[1]同济大学测量与国土信息工程系,上海200092

出　　处：《中国科技论文在线》2008年第11期793-800,共8页

基　　金：国家自然科学基金(40771174);教育部新世纪优秀人才支持计划(NCET-06-0381);上海市"曙光学者"计划(07SG24);高等学校博士学科点专项科研基金(20070247046)

摘　　要：高分辨率卫星遥感影像的物理模型描述影像与地面之间的严格几何成像关系,其参数具有可理解的物理意义,但其形式复杂且与传感器类型相关,出于技术保密其物理参数通常不被提供。有理函数模型作为通用传感器模型,解决了物理模型的缺陷,但缺点是其参数不具有可理解性。因此,拟以Quickbird立体影像为例,研究物理模型和有理函数模型的转换,以综合利用两者的优点。研究结果表明,从物理模型得到的有理函数模型,具有与物理模型相近的精度,这与已有研究结果一致;从有理函数模型可以在一定假设条件上恢复重建严格物理模型,恢复的严格物理模型的定位精度不受假设条件影响。A physical sensor model of the remote sensing image describes the geometric imaging relation between the image and the ground with interpretable sensor parameters. These parameters are complex and may change for various types of imaging sensor. Sometimes, physical sensor model parameters may not be available because of its technology security. As one of the general sensor models, Rational Function Model （RFM） sloves the problems caused by physical sensor model. But it has also some disadvantages such as difficulty in interpreting the parameters. This paper discusses the transformation between the physical sensor model and RFM by using a Quickbird stereo image pair. The research results indicate that the RFM calculated from a physical sensor model has almost the same accuracy with the physical sensor model, which is in agreement with other published research results. On the other hand, the physical sensor model can be retrieved from the RFM under a certain assumption and the geo-positioning accuracy using such a recovered physical sensor model keeps steadily high under different scenarios.

关 键 词：摄影测量与遥感 模型恢复 高分辨率卫星影像 物理模型 有理函数模型 立体定位 

分 类 号：P237[天文地球—摄影测量与遥感]