FPGA-Compatible Fog Rectification and Contrast Enhancement for Improved Vision Algorithms in Foggy Environments

abstract-Fog rectification is a crucial preprocessing step in enhancing image quality for applications like autonomous driving and object recognition, particularly in foggy and hazy conditions. This paper presents a streaming fixed-point implementation of a fog rectification and contrast enhancement algorithm tailored for FPGA deployment aiming to enhance visibility and contrast in foggy images. The fog rectification involves the dark channel of the image calculating the airlight map, refining it with filters and restoring the defogged image. Contrast enhancement is achieved through RGB to gray conversion and contrast stretching. The proposed algorithm is designed to efficiently process foggy RGB images, removing haze and enhancing contrast. The implementation utilizes fixed-point blocks from Vision HDL Toolbox for FPGA compatibility and employs a streaming pixel format for real-time processing. Simulation results demonstrate the effectiveness of the algorithm in improving image quality for vision algorithms in foggy environments.