最大对齐度算法在多模态图像配准中正负像无关性分析

Independent of Positive or Negative Image in Multi-modal Image Registration Using Alignment Metric

基于对齐度的配准是一种重要的多模态图像配准方法,其中的图像预处理是关键。在对红外热像仪图像和可见光图像进行配准的过程中,首先进行图像预处理,其中一个环节是将红外热像仪图像取负像,这样可使之与相应的可见光灰度图像比较相近。实验结果表明,对齐度配准算法与正负像无关,即可见光图像与相应的红外热像仪图像正像和负像的对齐度配准结果相同,理论证明也表明该结论的正确性。这一规律对于提高最大对齐度配准算法的计算效率具有重要意义,同时也具有一定的理论意义。

Alignment metric plays an important role in multi-modal images registration,where the preprocessing is a key step.In the process of registration between infrared thermal images and visual light images,one possible preprocessing is to convert the infrared thermal images to be their negative so that the converted images seem to be more alike with the corresponding visual light images.However,practical computations indicate that the registration based on alignment metric is independent of positive or negative images,i.e.,the registration of visual light image with the corresponding infrared thermal image or its negative is the same according to alignment metric,and this conclusion can also be proved by the theoretical proof.This rule would be helpful in multi-modal images registration.