MATLAB Implementation of Multi-Scale Retinex Algorithm

I have discovered another MATLAB implementation of the Multi-Scale Retinex algorithm that produces excellent results. Retinex is an image enhancement algorithm that improves image contrast and brightness by simulating human visual perception mechanisms. This algorithm works by decomposing an image into multiple scales and applying filters with varying intensities to mimic the human eye's perception system. The MATLAB code provides a straightforward approach to implement the Retinex algorithm, featuring Gaussian filters at different scales (typically using sigma values of 15, 80, and 250) to handle both local and global contrast enhancement. Key functions include logarithmic transformation for dynamic range compression and center/surround operations for reflectance estimation. Users can easily adapt the code to different images by adjusting parameters such as the number of scales, Gaussian kernel sizes, and gain/offset values for final output normalization. The implementation includes efficient matrix operations for fast processing and handles both color and grayscale images through channel-wise processing. If you need to enhance your image enhancement techniques, I strongly recommend trying this code, which offers both basic functionality for beginners and customization options for advanced users seeking optimal results for specific application requirements.