Heavy Fog Image Enhancement Algorithm Based on Tophat Weighted Bilateral Filtering

Abstract

A Retinex algorithm based on Tophat weighted bilateral filtering is proposed to enhance the distortion of images in highly heavy foggy weather, which makes it difficult to recognize objects. Firstly, The Tophat operator is used to correct the bilateral filter, keeping the edge effects of the brightness component. Then, an adaptive Gamma transformation is applied to expand the details of the saturation and brightness of the dark tones, and the reflection component is obtained from the illumination component through the Retinex algorithm to further enhance the detail information of the heavy image. Finally, the enhanced image is obtained by converting the HSV color space model into the RGB color space. The enhancement effect was evaluated from both subjective and objective aspects. The subjective evaluation results showed that the gray-scale range of the image was more reasonable, and objects in the image could be recognized more clearly. In terms of objective evaluation, when compared with the Retinex algorithm with color restoration (MSRCR), guided filter Retinex algorithm (GFR), improved dark prior algorithm, and Frankle-McCann Retinex algorithm (FMR), the proposed method greatly improved in terms of information entropy, standard deviation, and average gradient. The defogging effect was significantly improved.