Rigid Image Registration using Mutual Informationand Wavelet Transform
Sabri Halheit, Souad Benabdelkader · 2018
Mutual information similarity metric has become the leading technique in multimodal image registration in a variety of applications, for several years. The real advantage of this method is its ability to measure the statistical dependency between sensed and reference image without any prior segmentation or constraints in the image content of the modalities involved. In this paper, we propose a new rigid image registration scheme based jointly on mutual information and wavelet transform. The main point of our work is to generate, in a one-level multiresolution representation, the adequate initial mapping transform needed to align the reference and sensed image. The algorithm proposed aims to minimize two main drawbacks of mutual information registration, namely time consuming and local maxima.