A Method for Dehazed Image Quality Assessment
The development of general purpose no-reference approaches to dehazed image quality evaluation still lags in recent advances in image dehazing methods. While a number of image dehazing methods have been established and have shown to perform well, these ar
- PDF / 465,733 Bytes
- 5 Pages / 439.37 x 666.142 pts Page_size
- 56 Downloads / 208 Views
Abstract The development of general purpose no-reference approaches to dehazed image quality evaluation still lags in recent advances in image dehazing methods. While a number of image dehazing methods have been established and have shown to perform well, these are correlating highly with subjective evaluation of image quality. Toward ameliorating this we introduce the DIAS (Dehazed Image Assessment using Statistics) which is a no-reference approach to dehazed image quality assessment (DIQA) that does not assume a specific type of distortion of the image. It is based on detecting dehazed image quality based on Circularly Symmetric Gaussian Normalization Procedure Visible Edges Feature and it requires no training. The method is shown to correlate highly with human perception of quality. Our contribution in this direction is the development of dehazed image quality assessment method based on Circularly Symmetric Gaussian Normalization Procedure Visible Edges Feature which does not require exposure to distorted images priori and training.
Keywords Dehazed image quality assessment Circularly symmetric Gaussian Visible edges feature
Z. Hu Q. Liu (&) Intelligent Transport Systems Research Center, Wuhan University of Technology, Wuhan 430063, Hubei, People’s Republic of China e-mail: [email protected] Z. Hu e-mail: [email protected] Q. Liu School of Automation, Wuhan University of Technology, Wuhan 430070, Hubei, People’s Republic of China
Z. Wen and T. Li (eds.), Practical Applications of Intelligent Systems, Advances in Intelligent Systems and Computing 279, DOI: 10.1007/978-3-642-54927-4_87, Springer-Verlag Berlin Heidelberg 2014
909
910
Z. Hu and Q. Liu
1 Introduction Due to the influence of atmosphere particles such scattering effect in haze weather, it causes different degrees of reduction of the scene visibility. Quality of outdoor images is usually degraded which loose the contrast and color fidelity. Thus it greatly affects the processing of the video image effects and postanalysis. Therefore, many scholars have further studied in image dehazing [1–3]. However, there exist only few methods specifically for dehazed image quality assessment. At present, there exist approaches to DIQA research are varied and commonly follow one of two trends. One is quality assessment by image contrast, and the other is quality assessment by consolidated image contrast and color. The former is the most widely used method of Tarel and Hautière which based on the visible edge for no-reference image quality evaluation [4, 5]. The latter, such as in [6–8], this approach extracts features of global or local contrast from images to measure the grade of image enhancement and combine the three indexes of hue, RGB element and histogram similarity in order to evaluate the quality of image color. We seek to observe that the emphasis of DIQA is the detailed and clarity of the image. So, in this paper, we present the DIQA method based on visible edges feature of the images.
2 Feature Extraction 2.1 Normalization Procedure Nowaday
Data Loading...
