Shadow Removal Using Retinex Center-Surround Algorithm

Resource Overview

Implementation of shadow removal through Retinex center-surround algorithm with detailed technical specifications and corresponding program presentation slides

Detailed Documentation

In our processing workflow, we employed the Retinex center-surround algorithm for effective shadow removal. This algorithm operates by simulating human visual perception mechanisms through multi-scale Gaussian surround functions and reflectance estimation. The implementation involves calculating illumination components using convolution with Gaussian kernels of varying sizes (typically σ values ranging from 15 to 80 pixels), followed by reflectance recovery through logarithmic domain operations. The core MATLAB functions include imgaussfilt() for Gaussian filtering and imadjust() for final image enhancement. We developed comprehensive PowerPoint documentation that illustrates the algorithm workflow, mathematical formulations, and performance comparisons across different parameter configurations.