Remote Sensing Image Haze Removal Using Gamma-Correction-Based Dehazing Model
OAPA Haze is evident in most remote sensing (RS) images, particularly for the RS scenes captured in inclement weather, which severely hinders image interpretation. In this paper, two simple yet effective visibility restoration formulas are proposed for RGB-channel RS (RRS) images and multispectral RS (MSRS) images, respectively. More specifically, a robust gamma-correction-based dehazing model (RGDM) is firstly defined, which can better address the non-uniform illumination problem in hazy images. Then, the scene albedo restoration formula (SARF) used for RRS images is obtained by imposing the existing prior knowledge on this RGDM, which enables us to simultaneously eliminate the interferences of haze and non-uniform illumination. In subsequence, according to Rayleigh’s law, an expanded restoration formula (E-SARF) is further developed for MSRS data. Using the proposed E-SARF, the spatially varying haze in each band can be thoroughly removed without using any extra information. Experiments are conducted on challenging RRS and MSRS images, including images with non-uniform illumination, nonuniform haze distribution, and heavy haze. The results reveal that SARF and E-SARF are superior to most other state-of-arts techniques in terms of the both the recover quality and implementation efficiency.
Please use this identifier to cite or link to this item: