Image Registration Source Codes with Algorithm Implementations

This image registration project utilizes a comprehensive set of source codes implementing various computer vision algorithms. The system architecture begins with a MATLAB GUI interface comprising proj.m for image selection window management and testListBoxlj.m for handling list box functionality. For optimization core, we implemented two key algorithms: PSO.m (Particle Swarm Optimization) for global search capabilities and POWELL.m implementing Powell's conjugate direction method for local optimization. The mutual information module contains six different metric implementations: MI.m (basic Mutual Information), EMI.m (Enhanced Mutual Information), RMI.m (Robust Mutual Information), HiMI.m (Hierarchical Mutual Information), GMI.m (Gradient-based Mutual Information), and FPMI.m (Fast Pyramid Mutual Information) - each designed for specific registration scenarios with varying computational efficiency and accuracy trade-offs. Supporting programs include oneDimSearch.m as a subroutine for POWELL.m implementing one-dimensional line search optimization, and myMI.m as a dedicated mutual information calculator for PSO.m. The validation suite features restore.m for assessing registration accuracy by comparing transformed images, while gobad.m performs essential image transformations including translation and rotation operations using affine transformation matrices. During the image registration workflow, these programs work cohesively: the GUI facilitates user interaction and image input, optimization algorithms iteratively adjust transformation parameters, mutual information metrics quantify alignment quality, and validation tools verify final registration accuracy through geometric transformations.