Improved Intensity-Based Image Registration via Archimedes Optimization Algorithm
Mohamed Kmich, Hicham Karmouni, Inssaf Harrade, Mohammed Jamil Ouazzani, Hassan Qjidaa, Mhamed Sayyouri · 2024
Image registration is the process of superimposing two or more images to align them correctly. The aim is generally to minimize spatial discrepancies between images. In this context, we propose a new method for intensity-based image registration (IBIR) based on the optimization of the parameters of the 2D affine geometric transformation needed to align the images via the Archimedes Optimisation Algorithm (ArOA), and on the normalization of the sum of squared differences (NSSD) between different images taken at different times or from different views and the desired image as an objective function. The proposed method, based on ArOA, is robust to variations and perturbations in the image data. It can avoid local optima and find optimal solutions even when elements such as noise, distortions, or artefacts are present in the images. Simulations show the efficiency of the proposed method in terms of matching aligned images and computation time compared with other methods based on the Sine-Cosine Algorithm (SCA) and the Equilibrium Optimizer (EA).