An Optimized Stationary Wavelet Fusion Technique for Image de-hazing

Nowadays, the amount of smoke and dust in the air is increasing significantly due to industrialization. The smoke and dust particles accumulate in the relatively dry air and cause haze in the surrounding area, impairs visibility. This haze also affects photography, which reduces the images' quality and looks unnatural. The hazy atmosphere affects even pictures taken with a cell phone in everyday life. There are many methods to remove this haze content from the image, but they have not yielded great results. The long-time and short-time shots constantly differed while attempting to eliminate atmospheric haze from the images. To solve this problem, a fusion rule was proposed to fuse the luminance and dark channel prior (DCP) methods. The transmission estimated with the DCP method contributes mainly to the foreground regions, while the luminance model deals with the celestial regions. The fusion technique is a pixel-level fusion approach in the transform domain. The proposed approach combines the transmittance values obtained from the dark channel in front of the foreground region (background) and the luminance model for the sky region in the transform domain using the Stationary Wavelet Transform (SWT) with the optimized level of decomposition. The proposed algorithm was subjected to quantitative analysis of some statistical measures. The result shows that the proposed method successfully maintains the maximum visual truth content by effectively removing atmospheric haze from the images.