PROJECT TITLE :

Hazy Image Decolorization With Color Contrast Restoration

ABSTRACT:

Because the colour contrast field in a foggy image is warped, converting it to grayscale is difficult. We propose a technique for decolorizing hazy images into a distortion-free grayscale picture in this work. A CIELab colour space representation of the restored contrast and its distorted input is used to show the relationship between the restored contrast and its distorted inputs. Nonlinear optimization problems are used to build the grayscale image from this restoration. With an expansion of the Huber loss function, a new differentiable approximation approach is given to handle this problem. Using the suggested technique, we were able to achieve a gray-scale colour contrast that is very similar to the matching ground truth gray-scale one while still maintaining global luminance consistency.