Haze removal algorithm based on single-images with chromatic properties
Introduction
Safe, reliable vehicles critically depend on autonomous driving and intelligent transportation systems. Most ADAS perception systems contain LIDAR-based primary detection blocks and vision-based road segmentation modules. The vision-based system is vulnerable to inclement weather such as fog and rain, which can cause a severe increase in traffic accident rates. Fog or rain come with large amounts of particles and water droplets which absorb and scatter light in the atmosphere, causing blurring, poor contrast, and degrade fidelity in outdoor images. ADAS are substantially limited in harsh weather conditions, to this effect. There is urgent demand for efficient and robust algorithms to benefit computational photography and computer vision applications such as stereo matching [1], segmentation [2], and recognition [3].
Haze (e.g., fog) removal is a challenging endeavor because the concentration of the haze is related to depth, which is difficult to assess in a given image. Early researchers on this topic attempted the use of multiple images or additional information [4], [5], [6], [7]. Although these methods perform well in capturing the depth of haze across multiple images, they are not suitable for real-time application due to excessive input requirements.
Recently, single-image-based haze removal has attracted a great deal of research interest due to its potential for much broader application. Many single-image methods are based on priors or assumptions for effective haze removal. Haze can be removed efficiently by maximizing the local contrast of a hazy image, and the depth information is gathered by learning strategies. These contrast enhancement methods above also perform well but over-saturation always occurs in white regions and results in abnormal color in local areas, such as sky or white tall buildings. For haze removal based on deep learning, they perform excellently in removing haze with fast speed; however, the results are affected by training dataset and may fail in heterogeneous fog conditions. In this study, we combined the DRM with a probabilistic framework based on single images (Fig. 1). This technique can significantly enhance scene contrast and visibility in images featuring dense fog. Instead of using the DCP or boundary constraint, we use the DRM [8] to initially estimate the fog map. Based on our observation that highlight and fog share the same chromatic properties, we assert that haze removal can be treated as a decomposition problem similar to highlight removal. The chromatic properties of the fog map can constrain the chromaticity of restored images and prevent over-saturation.
It is important to note, however, the slight differences between fog and highlight which introduce noise to the fog map and de-hazed images. In our previous work [11], we used a bilateral filter to remove noise and haze from images. However, noise sharply increases as image sizes increases; the bilateral filter no longer functions beyond a certain image size. In this study, we attempted to streamline the haze removal process by using haze state detection to identify which images in the ADAS require dehazing rather than doing this on every image. Then, instead of using a bilateral filter, we used a maximum a posteriori (MAP) with several iterations to refine the fog map and determine the transmission rate and variations in atmospheric light. This eliminates the over-dehazing problem caused by heterogeneous fog in restored images. Finally, we recovered high-quality, haze-free images using an atmospheric scattering model.
The remainder of this paper is organized as follows. Section 2 reviews the extant research and highlights previous contributions which inspired our approach. Section 3 discusses the atmospheric scattering model that is most commonly used for image de-hazing; the linear relationship between brightness and saturation as-applied to identifying haze states is presented, then the fog map is established based on chromatic properties. Section 4 describes the fog map optimization and scene radiance recovery processes. The experimental results are presented and analyzed in Section 5, and conclusions are provided in Section 6.
Section snippets
Related work
Early methods for haze removal mainly rely on additional information for a single image, such as histogram equalization [4], or selecting multiple images of the same scene under different weather conditions. These methods yield limited results because haze scatters in various directions and remains in most parts of any image even after the histogram algorithm is applied. Schechner et al. [5] and Shwartz et al. [6] captured images from different polarization angles to efficiently remove haze.
Haze detection and de-hazing model
It is important to initially identify the haze state in an image to ensure accurate and efficient haze removal. The haze state can be easily detected by mapping RGB color space in the hazy image to HSV color space, then combining the haze concentration with depth information. We use the atmospheric scattering model to describe the formation of a hazy image, and the DRM as prior to preliminarily estimate the fog map.
Problem formulation
The observed hazy image can be modeled as a linear superimposition of the desired background layer and the hazy layer (fog map) , such that:
The goal of haze removal is to decompose the B and H from a given input image O. As previously stated, however, this problem is ill-posed. To solve it, we use the MAP to maximize the joint probability of the haze-free layer and hazy layer [41]. The object function is as follows: where
Experimental results
We tested the effectiveness of the proposed method on various hazy images by comparison against the methods proposed by Wang et al. [24], AODNet [9], DehazeNet [10] and SVR [35]. All algorithms were implemented in MatlabR2014a and hazy images came from two categories: natural and synthetic. All natural images were downloaded from KITTI-Road dataset [44] and Google with the following parameters: ; (k-means cluster) .
Conclusion
In this study, we established a novel method for solving the decomposition problem of background scenes and hazy layers in single images. Haze state detection serves to identify whether haze exists in the ADAS system image, which streamlines the de-hazing process by determining which images need processing as opposed to simply processing all of them. We propose a prior which imposes constraints on the hazy layer based on the DRM, wherein fog and highlight share the same chromaticity properties.
Acknowledgment
This work was supported by a grant from the National Natural Science Foundation of China (NSFC, No. 61504032).
References (48)
- et al.
Fast single image dehazing based on a regression model
Neurocomputing
(2017) - et al.
Stereo matching with improved radiometric invariant matching cost and disparity refinement
- et al.
Feature learning for image classification via multiobjective genetic programming
IEEE Trans. Neural Netw. Learn. Syst.
(2014) - et al.
Decomposition-based transfer distance metric learning for image classification
IEEE Trans. Image Process.
(2014) - et al.
An advanced contrast enhancement using partially overlapped sub-block histogram equalization
IEEE Trans. Circuits Syst. Video Technol.
(2001) - Y.Y. Schechner, S.G. Narasimhan, S.K. Nayar, Instant dehazing of images using polarization, in: Proc. IEEE Conf....
- S. Shwartz, E. Namer, Y.Y. Schechner, Blind haze separation, in: Proc. IEEE Conf. Comput. Vis. Pattern Recognit., CVPR,...
- J. Kopf, B. Neubert, B. Chen, M. Cohen, D. Cohen-Or, O. Deussen, M. Uyttendaele, D. Lischinski, Deep photo: model-based...
- et al.
Real-time specular highlight removal using bilateral filtering
- Boyi Li, Xiulian Peng, Zhangyang Wang, Jizheng Xu, Dan Feng, AOD-Net: All-in-one dehazing network, in: IEEE...
DehazeNet: an end-to-end system for single image haze removal
IEEE Trans. Image Process.
An efficient haze removal algorithm using chromatic properties
Guided image filtering
IEEE Trans. Pattern Anal. Mach. Intell.
Nighttime haze removal using color transfer pre-processing and dark channel prior
Stress for dehazing
Enhanced variational image dehazing
SIAM J. Imaging Sci.
Fusion-based variational image dehazing
IEEE Signal Process. Lett.
Cited by (7)
An efficient dehazing method of single image using multi-scale fusion technique
2023, Journal of Ambient Intelligence and Humanized ComputingImplementation of Satellite Road Image Denoising using Iterative Domain Guided Image Filtering with Gray World Optimization
2022, Journal of CommunicationsRemote Sensing and Natural Image Dehazing using DCP based IDERS Framework
2022, 2022 1st International Conference on Electrical, Electronics, Information and Communication Technologies, ICEEICT 2022STRASS Dehazing: Spatio-Temporal Retinex-Inspired Dehazing by an Averaging of Stochastic Samples
2022, Journal of Renewable MaterialsSingle Image Haze Removal With Haze Map Optimization for Various Haze Concentrations
2022, IEEE Transactions on Circuits and Systems for Video TechnologyReview of passive polarimetric dehazing methods
2021, Optical Engineering