Single image dehazing based on single pixel energy minimization
- PDF / 2,974,076 Bytes
- 19 Pages / 439.37 x 666.142 pts Page_size
- 70 Downloads / 191 Views
Single image dehazing based on single pixel energy minimization Yakun Gao 1,2 & Yanbo Zhang 2 & Haibin Li 2
& Wenming Zhang
2
Received: 26 March 2019 / Revised: 12 April 2020 / Accepted: 22 April 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
The common dehazing algorithms always assume that the transmission values of all the pixels in an image block are the same (local consistency assumption). However, it is easy to appear “halo” for image regions where the depth changes obviously. In this paper, we calculate the transmission of each pixel separately without the local consistency assumption. First, we initialize a random transmission value for each pixel in the whole image. Then, we optimize the transmission values through several iterations by minimizing an energy function, which contains the data term and penalty term. In each iteration, we take two procedures of propagation and random search to optimize transmission values. Finally, we use the optimized transmission and the estimated atmospheric light to calculate the haze-free image. Comparison experiments show that our algorithm can remove haze effectively, and obtain the best performance. Keywords Image dehazing . Random initialization . Propagation . Random search . Energy minimization
1 Introduction Due to the absorption and scattering effects in hazy scenes, the quality of the captured image degrades seriously. Degraded images often suffer low contrast and fewer details, which are difficult to meet the needs of practical applications. Image dehazing has attracted attentions of many scholars. In order to solve these problems, many scholars have proposed their own solutions and the single image dehazing method has become the mainstream. The global
* Haibin Li [email protected]
1
School of Electrical Engineering and Automation, Henan Institute of Technology, Xinxiang 453003, China
2
School of Electrical Engineering, Yanshan University, Qinhuangdao 066004, China
Multimedia Tools and Applications
atmospheric light and the transmission map estimation are the two main problems of single image dehzing based on the atmospheric scattering model. The global atmospheric light determines the brightness perception of the dehazed image, and the transmission map decides whether there are halo artifacts, edge blur or hazy remains in the dehazed image. The common non-deep learning dehazing algorithms always assume that the transmission values of all the pixels in an image block are the same (local consistency assumption LCA). Under this assumption, it is easy to appear “halo” for the regions where the depth changes obviously. In this paper, to solve the “halo” effect, we propose a new haze removal algorithm which can estimate and refine the transmission map adaptively. First, we traverse the whole image and assign random initial transmission value to each pixel, and assume that in neighbor regions, there will be at least one pixel carries the transmission value which is very close to the truth one. Then, based on above idea, we
Data Loading...
