Maximizing Mutual Information: Optimal Keypoint Selection using Golf Optimizer for Improved Multimodal Image Registration
Kannedari Uday Kiran, T. Swati, W. Yasmeen · Advances in engineering research/Advances in Engineering Research · 2025
One of the eminent image processing tools for performing tasks including classification, detection, recognition, and other analysis tasks is the image registration process.This technique is useful for solving a huge variety of real-world problems like surveillance, medical image processing, geophysics, computer vision, remote sensing, etc.The application of optimization techniques has acquired considerable attention in the last decades in the multimodal image registration domain.Thus, this work focuses on suggesting a new optimizer for promoting the performance of multimodal image registration to aid the medical industry.Initially, the multimodal medical images are gathered from the standard datasets to carry out the research.As the medical images are acquired at different locations and environments, it is required to perform the pre-processing stage for enhancing the image registration efficiency.Here, the Histogram Equalization (HE) is applied for preprocessing and then the difference of Gaussian (DoG) is employed for identifying the key points in the pre-processed images.As an innovation to this concept, the percentage of key points is optimized using a recently suggested Golf Optimization Algorithm (GOA).Therefore, it is required to optimize and determine the optimal percentage of key points for promoting the performance of image registration in terms of similarity measures.Thus, this work adopts mutual information as a similarity measure for evaluating the improvements in multimodal image registration.