Reduced-Reference Image Quality Assessment by Structural Similarity Estimation

• Published in: IEEE transactions on image processing Vol. 21; no. 8; pp. 3378 - 3389
• Main Authors: Rehman, A., Zhou Wang
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.08.2012; Institute of Electrical and Electronics Engineers
• Subjects: Algorithms; Applied sciences; Data Interpretation, Statistical; Detection, estimation, filtering, equalization, prediction; Distortion measurement; Divisive normalization transform; Estimation; Exact sciences and technology; Feature extraction; image deblurring; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Image processing; Image quality; image repairing; Information, signal and communications theory; natural image statistics; Nonlinear distortion; Pattern Recognition, Automated - methods; Receivers; reduced-reference image quality assessment (RR-IQA); Reference Values; Reproducibility of Results; Sensitivity and Specificity; Signal and communications theory; Signal processing; Signal, noise; structural similarity; Subtraction Technique; Telecommunications and information theory; Transforms; Image processing; Similarity; image deblurring; structural similarity; Regression analysis; Image restoration; Divisive normalization transform; Blurred image; Subjective evaluation; Discretization method; Image quality; image repairing; reduced-reference image quality assessment (RR-IQA); Multiscale method; natural image statistics; Quality control; Database; Signal distortion; Image evaluation
• ISSN: 1057-7149; 1941-0042; 1941-0042
• DOI: 10.1109/TIP.2012.2197011

Reduced-reference image quality assessment (RR-IQA) provides a practical solution for automatic image quality evaluations in various applications where only partial information about the original reference image is accessible. In this paper, we propose an RR-IQA method by estimating the structural similarity index (SSIM), which is a widely used full-reference (FR) image quality measure shown to be a good indicator of perceptual image quality. Specifically, we extract statistical features from a multiscale multiorientation divisive normalization transform and develop a distortion measure by following the philosophy in the construction of SSIM. We find an interesting linear relationship between the FR SSIM measure and our RR estimate when the image distortion type is fixed. A regression-by-discretization method is then applied to normalize our measure across image distortion types. We use six publicly available subject-rated databases to test the proposed RR-SSIM method, which shows strong correlations with both SSIM and subjective quality evaluations. Finally, we introduce the novel idea of partially repairing an image using RR features and use deblurring as an example to demonstrate its application.