3D Shape Reconstruction MATLAB Code Using Shape from Defocus with Divergence

This MATLAB code implements 3D shape reconstruction using the shape-from-defocus technique enhanced with divergence analysis. The core algorithm processes two distinct depth images captured with different focal settings to reconstruct detailed 3D models. The implementation includes key functions for point spread function (PSF) modeling, defocus parameter estimation, and depth map calculation through divergence-based optimization. The code structure begins with image preprocessing where input images are normalized and filtered to reduce noise. The main reconstruction module employs Fourier transform analysis to estimate defocus parameters by comparing high-frequency content variations between the two images. The divergence component calculates depth gradients using Laplace operators, enhancing edge preservation and surface continuity. Key MATLAB functions include: - psfGeneration() - creates point spread functions for different depth levels - depthFromDefocus() - computes depth maps using defocus blur analysis - divergenceOptimization() - refines depth estimates through gradient-based optimization - meshReconstruction() - converts depth maps to 3D mesh models using triangulation The algorithm's efficiency comes from its sparse matrix operations and parallel processing capabilities, making it suitable for medium-resolution images. Users can adjust parameters like PSF size and optimization thresholds to balance reconstruction accuracy and computational speed. The output includes both point cloud data and surface mesh representations compatible with standard 3D visualization tools. This implementation demonstrates robust performance on various objects while maintaining computational efficiency, making it accessible for researchers and practitioners without requiring extensive background in computational photography. The code includes comprehensive documentation and example datasets for easy validation and adaptation.