Image Quality Assessment Based on Local Pixel Correlation
The available image quality assessment methods are mostly based on statistical characteristic and consider very little the change of pixel correlation in conjunction with the quality assessment, which induces the quality assessment metric to be limited in
- PDF / 765,152 Bytes
- 10 Pages / 439.37 x 666.14 pts Page_size
- 30 Downloads / 209 Views
stract. The available image quality assessment methods are mostly based on statistical characteristic and consider very little the change of pixel correlation in conjunction with the quality assessment, which induces the quality assessment metric to be limited in the degradation of image quality caused by the change of pixel correlation. However, the pixel correlation change has a big effect on the image quality, so a novel image quality assessment based on the pixel correlation is proposed in this paper. Firstly image is parted based on mutual information, and then, three kinds of mutual information between the pixel intensity and the image patches are extracted to catch the variation of the pixel correlation. Finally the machine learning is utilized to learn the mapping from these differences space to image quality. The experimental results show that the proposed framework has good consistency with subjective perception. Keywords: Image quality assessment · Mutual information · Pixel correlation
1
Introduction
Image acts as the carrier of information, which conveys the vital information to everyone and becomes a ubiquitous part of modern life. However, the impairment of image effects the substantial information contained in the image, which will bring down the satisfaction of human perceived. It is essential to build image quality evaluation metrics for various image applications [1-2]. Subjective methods perceive image quality by many participators, which is expensive and time consuming. So we move to objective measurements which accomplish the image quality assessment task automatically. The state-of-the-art image quality assessment methods can be divided into two broad classes in which kind of information it used. The first is the image pixel domain based paradigm, where the pixel value changes between the reference and distorted signals is predicted as the image quality. Mean Squared Error [3] and Peak Signal-toNoise Ratio are the most widely used objective quality metrics due to their convenience and clear physical meaning. Zhou Wang et al. [4] propose a method based on Structural Similarity which measures the structure variation between the reference and distorted image. Corresponding to the pixel domain based methods, are the image © Springer-Verlag Berlin Heidelberg 2015 H. Zha et al. (Eds.): CCCV 2015, Part II, CCIS 547, pp. 266–275, 2015. DOI: 10.1007/978-3-662-48570-5_26
Image Quality Assessment Based on Local Pixel Correlation
267
transform domain based methods, where the transform domain coefficients information error between the reference and distorted signals is predicted as the image quality. Sheikh et al. [5] proposed a method named Visual information fidelity based on the Gaussian scale mixtures in the wavelet domain. Wang et al. [6] propose a method using a natural image statistic model in the wavelet domain, which measures the wavelet coefficients histogram difference between the reference and distorted signals to get image quality score. Ding et al. [7] propose a method using mutual info
Data Loading...
