遥感图像几何校正的非线性有理逼近模型  

The model of nonlinear rational approximation for geometric correction of remote-sensing image

作　　者：张连蓬[1] 陶华学[1] 刘国林[1] 江涛[1] 

机构地区：[1]山东科技大学地球信息科学与工程学院,山东泰安271019

出　　处：《煤炭学报》2003年第2期140-144,共5页Journal of China Coal Society

基　　金：国家自然科学基金资助项目(40174003);山东省优秀中青年科学家科研奖励基金资助项目(20507)

摘　　要：遥感图像几何校正是遥感图像处理与分析中的一个重要环节,常用的校正模型有仿射变换和代数多项式.由于引起图像几何变形的原因非常复杂,从而使变形呈现非常复杂的非线性现象,传统的校正方法常常无法得到很好的校正结果.有理多项式有更强的非线性特性,能更准确地描述图像的非线性变形.提出了一个有理多项式校正模型并给出了一套快速的参数估计方法,实例验证表明,有理多项式模型更精确地逼近了非线性变形,提高了几何校正的精度.Geometric correction of remotesensing image is one of important parts during processing and analyzing remotesensing image. The main traditional algorithm is based on the affine transformation and polynomial model. Because the reasons of geometric distortion are very complicated, the geometric distortion is usually a kind of nonlinear phenomena, the traditional model can not correct the distortion very well. Rational polynomial has more nonlinear characters, and can describe nonlinear distortion more accurately. This paper proposes a rational polynomial model and derives a direct solution for the parameters with respect to the usual iterative solution. The experiment shows that the new model and direct solution is better than the usual used polynomial model and improves the accuracy of geometric correction of image.

关 键 词：遥感图像 几何校正 非线性有理逼近 图像处理 参数估计 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]