Dark channel defogging enhancement algorithm for haze image in complex environment

Abstract

Under the lighting conditions of artificial light sources, point lighting causes the image to receive uneven light, leading to the characteristics of color distortion and poor dehazing enhancement effect when the existing dark channel dehazing algorithm is used for dehazing enhancement of haze images. Aiming to solve the above problems, this paper proposes a dark channel defogging enhancement algorithm for haze image in complex environment. First, by calculating the intensity histogram of the initial image, the corresponding mean, median and maximum value are obtained, and a novel brightness factor is constructed accordingly. Then, the over-bright areas in the image are extracted through the brightness factor and processed by down-value. Finally, the down-value image is used to estimate the atmospheric light value to enhance the accuracy of the atmospheric light value, so as to solve the problems of color distortion and poor dehazing effect. The experimental results show that the proposed algorithm has good applicability to dust fog images under non-uniform illumination. Compared with other advanced algorithms, the no-reference evaluation index ranks high, and the qualitative results are more realistic and natural.