Polarimetric Dual-Channel Multiscale Decomposition Dehazing
Tongwei Ma, Jianping Zhou, Lilian Zhang, Chen Fan, Bo Sun, Ruilei Xue, Jun Mao · IEEE Sensors Journal · 2025
In this article, the problem of simultaneous image dehazing of near and far scenes in hazy weather is addressed. We propose the polarimetric dual-channel multiscale decomposition algorithm to solve this problem. The main strategy is to extract the near and far scenes of the image separately, and then combine them into a dehazed image based on the fusion principle. First, the cuckoo search-contrast-limited adaptive histogram equalization (CS-CLAHE) method is presented to obtain the contrast-enhanced polarization contrast-limited adaptive histogram equalization image as the channel of the near scene. Meanwhile, the incomplete normalized degree of polarization (INDoP) image is obtained as the channel of the far scene. Then, we introduce a rolling guidance filter and bilateral filter to decompose the dual channels of near and far scenes into the base layers and the detail layers. To fusion the base layers, we use the Sobel gradient map as the fusion weight of the base layers, and we proposed a fusion rule for the base layers by taking account of the Sobel gradient map. This strategy can retain the low-frequency information of near and far scenes. To preserve the textures of near and far scenes in detail layer fusion, we design a novel weighted least-squares optimization method to fuse the detail layers, which effectively improves the quality of the fused image. To evaluate the proposed method, we create a polarized hazy image dataset. Experimental results demonstrate that the method has significant advantages in qualitative and quantitative evaluations compared with other state-of-the-art dehazing methods.