Dehazed Image Quality Assessment by Haze-Line Theory

6th conference on Advances in Optoelectronics and Micro/nano-optics

Song, Yingchao^1,2,3,4 ; Luo, Haibo^1,3,4 ; Lu, Rongrong^1,2,3,4 ; Ma, Junkai^1,2,3,4

Shenyang Institute of Automation CAS, Shenyang

110016, China^1

University of Chinese Academy of Sciences, Beijing

100049, China^2

Key Laboratory of Opto-Electronic Information Processing CAS, Shenyang

110016, China^3

Key Lab of Image Understanding and Computer Vision, Liaoning Province, Shenyang

110016, China^4

关键词: Color clusters;    Contrast Enhancement;    Evaluation metrics;    Human judgments;    Image contrasts;    Image quality assessment;    Inter clusters;    State of the art;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/844/1/012045/pdf
DOI  :  10.1088/1742-6596/844/1/012045

来源: IOP

【 摘 要 】

Images captured in bad weather suffer from low contrast and faint color. Recently, plenty of dehazing algorithms have been proposed to enhance visibility and restore color. However, there is a lack of evaluation metrics to assess the performance of these algorithms or rate them. In this paper, an indicator of contrast enhancement is proposed basing on the newly proposed haze-line theory. The theory assumes that colors of a haze-free image are well approximated by a few hundred distinct colors, which form tight clusters in RGB space. The presence of haze makes each color cluster forms a line, which is named haze-line. By using these haze-lines, we assess performance of dehazing algorithms designed to enhance the contrast by measuring the inter-cluster deviations between different colors of dehazed image. Experimental results demonstrated that the proposed Color Contrast (CC) index correlates well with human judgments of image contrast taken in a subjective test on various scene of dehazed images and performs better than state-of-the-art metrics.