Implementation of Dehazing Based on Dark Channel Prior

This program primarily implements a dehazing algorithm based on the dark channel prior principle. The dark channel prior refers to utilizing statistical knowledge from the dark channels of images during the dehazing process, where the dark channel is obtained by taking the minimum intensity across color channels for each pixel. By analyzing the dark channel statistics, the algorithm estimates atmospheric light and transmission maps to recover haze-free images. The implementation typically involves key steps such as dark channel computation using minimum filters, atmospheric light estimation from the brightest pixels in the dark channel, and transmission map refinement through guided filtering. This research represents one of the significant directions in image processing, and the application of this program is quite extensive across various domains including computer vision and photography enhancement. The code structure includes modular functions for dark channel extraction, atmospheric light calculation, and transmission refinement, allowing researchers to easily modify parameters and adapt the algorithm for specific scenarios. We hope this program serves as an effective tool for researchers engaged in this field, enabling them to better study and apply dehazing techniques through clear code organization and practical implementation examples.