An Improved Image Quality Assessment in Gradient Domain
The available image quality assessment (IQA) methods based on gradient calculation are mostly implemented without considering visual perception threshold (VPT) and color information. However, incorporating VPT with IQA model can reduce redundant informati
- PDF / 3,607,739 Bytes
- 9 Pages / 439.37 x 666.14 pts Page_size
- 62 Downloads / 243 Views
stract. The available image quality assessment (IQA) methods based on gradient calculation are mostly implemented without considering visual perception threshold (VPT) and color information. However, incorporating VPT with IQA model can reduce redundant information and human visual system (HVS) is extremely sensitive to color variation. An improved image quality assessment in gradient domain is proposed which utilizes minimum amount of gradient coefficients to capture the color and structure distortion of degraded image by applying a VPT to remove the unperceived gradient coefficients. The difference of perceived gradient coefficients between distorted and reference image is measured to acquire image quality score. Experimental results on two benchmarking databases (LIVEII and TID2008) indicate the rationality and validity of the proposed method. Keywords: Image quality assessment · Human visual system · Gradient calculation · Visual perception threshold
1
Introduction
Objective image quality assessment is designed with the aim of interpreting the quality of distorted image automatically and responding consistently with the behavior of the HVS [1-2]. A huge number of IQA algorithms have been emerged with the evolution of image processing technology, which can be divided into two categories, namely HVS based paradigm and non-HVS based metrics. The traditional peak signal to noise ratio (PSNR) [3] just measure the pixel difference between degraded and reference image to obtain the image quality score, which doesn’t accord with the way of human perceive information. The perfect IQA model is required to simulate the actual process of HVS perceive image. However, the HVS is extremely complex and the research on it is limited, which lead to the mainstream IQA methods are designed based on certain properties of HVS. The Multi-Scale structural similarity (MS-SSIM) [4] assumes that HVS is sensitive to structure information in an image when perceiving the image quality. Motivated by SSIM, the gradient SSIM (G-SSIM) [5] is built by Chen et al, which first compute the gradient of distorted image and reference image and then measure the luminance similarity, contrast similarity and structural similarity of gradient maps. Given the gradient magnitude maps, the gradient orientation © Springer-Verlag Berlin Heidelberg 2015 H. Zha et al. (Eds.): CCCV 2015, Part II, CCIS 547, pp. 293–301, 2015. DOI: 10.1007/978-3-662-48570-5_29
294
Y. Ren et al.
maps and contrasts of reference and distorted image, the similarity among them is computed in geometric structure distortion (GSD) [6] method to acquire the image quality score. RR-VIF [7] constructs the IQA model by measuring the change of visual information fidelity in the distorted image. GMSD [8] explores the use of global variation of gradient based local quality map for overall image quality prediction. The available IQA methods based on gradient calculation are mostly implemented ignoring VPT and color information. However, incorporating VPT with IQA model can reduce redundant in
Data Loading...
