Abstract
The images capture in a bad environment usually loses its fidelity and contrast. As the light rays travel towards its destination they get scattered several times due to the tiny particles of fog and pollutants in the environment, therefore the energy gets lost due to multiple scattering till it arrives its destination, and this degrades the images. So the images taken in bad weather appear in bad quality. Therefore, single image haze removal is quite a bit tough task. Significant research has been done in the haze removal algorithm but in all the techniques, the coefficient of scattering is taken as a constant according to the homogeneous atmosphere but in real time this does not happen. Therefore, this paper introduces a simple and efficient method so that the scattering coefficient becomes variable according to the inhomogeneous environment. Then, this research aims to remove the haze with the help of a fast and effective algorithm i.e., Prior Color Fading, according to the inhomogeneous environmental properties. Thereby, to filter the depth map, the authors used a weighted guided image filtering which removes the drawbacks of guided image filter. Afterwards the scattering coefficient is made variable according to the inhomogeneous atmosphere and then the Simple Color Balance Algorithm is applied so that the readability property of images can be increased. The proposed method tested on various general outdoor images and synthetic hazy images and analyzed on various parameters Mean Square Error (MSE), Root Mean Square Error (RMSE), Peak Signal to Noise Ratio (PSNR), Mean Structural Similarity (MSSIM) and the Universal Objective Quality Index (UQI). Experimental results for the proposed method show that the proposed approach provides better results as compared to the state-of-the-art haze removal algorithms.
Highlights
Images recorded in foggy and hazy weather can be corrupted by scattering of atmospheric particles, which decreases the contrast, changes the color, and transforms the features of images in such a way that are difficult to identify by human vision and some outdoor systems based on computer vision
The proposed method tested on various general outdoor images and synthetic hazy images and analyzed on various parameters Mean Square Error (MSE), Root Mean Square Error (RMSE), Peak Signal to Noise Ratio (PSNR), Mean Structural Similarity (MSSIM) and the Universal Objective Quality Index (UQI)
For showing a clearer comparison between these methods, Tabs. 1–4 show the analysis of the image quality containing MSE, RMSE, PSNR, MSSIM and the UQI
Summary
Images recorded in foggy and hazy weather can be corrupted by scattering of atmospheric particles, which decreases the contrast, changes the color, and transforms the features of images in such a way that are difficult to identify by human vision and some outdoor systems based on computer vision. The present research is based on the method proposed by Zhu et al [11] that uses a linear model for modeling the scene depth using the brightness and the saturation of the hazy image. The parameters of the linear model are learnt by a supervised learning method This method is simple providing efficient and good dehazing results, as required by some applications, it presents the drawback of considering the atmospheric scattering coefficient as constant, while atmospheric conditions may not be homogeneous [11]. The guided image filter used in this method does not remove halo artifacts from the image Another simple method is the one proposed by He et al [12], which uses a dark channel prior algorithm but does not work when there are white objects in the scene.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
