基于有理多项式系数模型的物方面元最小二乘匹配

摘要

Since patch-based least squares image matching only applies to the production of regional DSM that based on single-center projection with frame imaging, patch-based least squares image matching based on rational polynomial coefficients（RPC）model is proposed, with multi-center parallel projection in the matching region and RRC-based projection equation constructed, which extends patch-based least squares image matching from singlecenter projection with frame imaging to stereo-matching of multi-center projection with push-broom imaging, and the algorithm is proved by the experiment withSPOTS-HRG , GeoEye and tKONOS stereo images. The experimental results show that RPC model can be used in patch-based least squares image matching with high matching accuracy maintained to improve its applications and value.%针对物方面元最小二乘匹配仅适用于单中心投影框幅式成像的匹配制作区域DSM问题，提出基于有理多项式系数（RPC）模型的物方面元最小二乘匹配算法，结合匹配窗口区域内多中心平行投影方式，构建RPC模型下的投影方程，将物方面元最小二乘匹配算法从适用于单中心投影框幅式成像扩展到适用于多中心投影推扫式成像的立体匹配，并用SPOT5-HRG、GeoEye、IKONOS立体影像进行试验验证。试验表明，RPC模型能用于物方面元最小二乘匹配且不损失匹配精度，增加了物方面元最小二乘匹配的应用范围和价值。