Single Image Dehazing Using Wavelet-Based Haze-Lines and Denoising

Haze reduces the contrast of an image and causes loss in colors, which has a negative effect on subsequent object detection; therefore, single dehazing is challenging visual task. In addition, defects exist previous existing approaches: Pixel-based approaches are likely to result insufficient information estimate transmission, whereas patch-based ones prone generate shadows. They both also tend induce color deviations. Therefore, this study proposes novel method based multi-scale wavelet non-local dehazing. A hazy first decomposed into low-frequency three high-frequency sub-images by transform. Non-local denoising then employed remove haze noise, respectively. Finally, haze-free obtained from reconstruction sub-images. The proposed focuses images respectively, through details can be well preserved. Experimental results indicate that superior state-of-the-art quantitative qualitative evaluation synthetic real-world datasets.