Multi-Scale Retinex Theory Implementation

This source code provides a comprehensive implementation of multi-scale Retinex theory for image enhancement. The algorithm employs Gaussian filtering at multiple scales to separate illumination and reflectance components, supporting both grayscale and color image processing. Key implementation features include scale-space decomposition using convolution operations with different sigma values, logarithmic domain transformations for dynamic range compression, and color restoration mechanisms for chromatic consistency. The code structure allows users to customize scale parameters and weighting factors to optimize enhancement results for various image types. Through multi-scale processing, the implementation effectively improves image visibility, contrast adjustment, and color fidelity while maintaining computational efficiency. This approach enables researchers and developers to apply advanced Retinex-based enhancement to diverse imaging applications with configurable parameters for optimal results.