基于支持向量回归的光度配准算法

在空域内将图像之间的光度对应假设为线性的、或一些简单的非线性模型进行处理．配准后的图像仍存在较大的光度差异。将支持向量回归技术应用到光度配准中，利用支持向量回归估计图像问的光度对应关系，再根据估计得到的回归曲线对图像进行光度变换，消除图像间的光度差异，针对实际图像的对比实验表明，基于支持向量回归的光度配准算法。优于常用的空域内回归估计光度配准算法。

A Photometric Registration Algorithm Based on Support Vector Regression

The traditional photometric registration algorithms in the space domain estimate the photometric relations between images on the assumption that they are linear or other simple relations, which results in the big photometric difference between images even after the photometric registration. Estimates the photometric relations between images by support vector regression, and transforms the image based on the estimated function, can effectively eliminate the photometric difference. The experiments with real images show that the photometric registration algorithm based on support vector regression is superior to the traditional photometric registration algorithms in the space domain.