Mutual Information-Based Image Registration

Application Background: Mutual Information (MI)-based registration methods utilize the statistical information of image grayscale to measure similarity between images. By employing search algorithms to find optimal spatial transformations that maximize similarity, these methods achieve accurate image alignment. MI measures the generalized distance between the joint probability distribution of two variables and their completely independent distributions as a similarity metric. Key Technology: This approach adopts rigid body transformation in 2D image space, primarily determining the rotation angle and x/y-axis displacements of floating images that maximize mutual information.