本文提出了一种基于学习的相似性度量方

法, 即将图像配准的度量问题转化为模式分类问题, 由基于机器学习设计的分类器自动检验图像是否配准. 本文对400 组图像进行了配准检验, 实验结果显示了该方法的可行性和可靠性.-This paper proposes a similarity measure based on learning methods, about the measurement problems of image registration into the pattern classification problem, the design of machine learning-based automatic test whether the image classifier registration. This set of images were of 400 registration test, test results show the feasibility and reliability.