论文部分内容阅读
本文提出运用最大似然采样一致性准则解算遥感影像配准系数的方法。该方法基于极大似然估计理论,首先对初始匹配点的坐标残差进行概率建模,计算概率模型成立时的似然函数值并选择似然函数值最大时的参数为正确结果,最终剔除错误点保留正确匹配点。该方法较之传统的最小二乘方法更为准确地计算配准系数,并可以解决随机采样一致性准则解算配准参数时,对阈值的依赖问题。试验证明,该方法可提高配准参数解算的稳健性和精度。
In this paper, a method of calculating the registration coefficient of remote sensing images using the criterion of maximum likelihood sampling consistency is proposed. This method is based on the theory of maximum likelihood estimation. First, the probability of the initial residuals of the matching point is modeled. The probability function is established when the probability model is established, and the parameters with the maximum value of the likelihood function are chosen as the correct ones. Finally, The wrong point keeps the correct match point. Compared with the traditional least square method, this method can calculate the registration coefficient more accurately and can solve the problem of threshold dependence when the stochastic sampling consistency criterion solves the registration parameters. Experiments show that this method can improve the robustness and accuracy of the solver parameters.