Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Perceptual Constancy Constrained Single Opinion Score Calibration for Image Quality Assessment

Lei Wang, Desen Yuan

TL;DR

Subjective image quality labels are costly and noisy when collected from many raters. The paper introduces perceptual constancy constrained calibration (PC3), which estimates the mean opinion score (MOS) from a single opinion score (SOS) by modeling the SOS likelihood as a normal distribution with a mean given by a reference MOS plus a learnable relative quality $S_theta(x,r_x)$. A Siamese CONTRIQUE-based backbone computes the relative quality, and MOS values are updated via Newton's method while the relative quality is updated by backpropagation under a Lagrangian that enforces perceptual constancy across references. Key contributions include the PC3 framework, alternating optimization for MOS and the relative quality, and extensive validation across four IQA datasets showing improved SOS calibration and better IQA learning when only SOSs are available, along with strong generalization to few opinion scores. This approach reduces the need for multi-subject MOS labeling while delivering accurate MOS estimates and enhancing downstream IQA performance.

Abstract

In this paper, we propose a highly efficient method to estimate an image's mean opinion score (MOS) from a single opinion score (SOS). Assuming that each SOS is the observed sample of a normal distribution and the MOS is its unknown expectation, the MOS inference is formulated as a maximum likelihood estimation problem, where the perceptual correlation of pairwise images is considered in modeling the likelihood of SOS. More specifically, by means of the quality-aware representations learned from the self-supervised backbone, we introduce a learnable relative quality measure to predict the MOS difference between two images. Then, the current image's maximum likelihood estimation towards MOS is represented by the sum of another reference image's estimated MOS and their relative quality. Ideally, no matter which image is selected as the reference, the MOS of the current image should remain unchanged, which is termed perceptual cons tancy constrained calibration (PC3). Finally, we alternatively optimize the relative quality measure's parameter and the current image's estimated MOS via backpropagation and Newton's method respectively. Experiments show that the proposed method is efficient in calibrating the biased SOS and significantly improves IQA model learning when only SOSs are available.

Perceptual Constancy Constrained Single Opinion Score Calibration for Image Quality Assessment

TL;DR

Subjective image quality labels are costly and noisy when collected from many raters. The paper introduces perceptual constancy constrained calibration (PC3), which estimates the mean opinion score (MOS) from a single opinion score (SOS) by modeling the SOS likelihood as a normal distribution with a mean given by a reference MOS plus a learnable relative quality . A Siamese CONTRIQUE-based backbone computes the relative quality, and MOS values are updated via Newton's method while the relative quality is updated by backpropagation under a Lagrangian that enforces perceptual constancy across references. Key contributions include the PC3 framework, alternating optimization for MOS and the relative quality, and extensive validation across four IQA datasets showing improved SOS calibration and better IQA learning when only SOSs are available, along with strong generalization to few opinion scores. This approach reduces the need for multi-subject MOS labeling while delivering accurate MOS estimates and enhancing downstream IQA performance.

Abstract

In this paper, we propose a highly efficient method to estimate an image's mean opinion score (MOS) from a single opinion score (SOS). Assuming that each SOS is the observed sample of a normal distribution and the MOS is its unknown expectation, the MOS inference is formulated as a maximum likelihood estimation problem, where the perceptual correlation of pairwise images is considered in modeling the likelihood of SOS. More specifically, by means of the quality-aware representations learned from the self-supervised backbone, we introduce a learnable relative quality measure to predict the MOS difference between two images. Then, the current image's maximum likelihood estimation towards MOS is represented by the sum of another reference image's estimated MOS and their relative quality. Ideally, no matter which image is selected as the reference, the MOS of the current image should remain unchanged, which is termed perceptual cons tancy constrained calibration (PC3). Finally, we alternatively optimize the relative quality measure's parameter and the current image's estimated MOS via backpropagation and Newton's method respectively. Experiments show that the proposed method is efficient in calibrating the biased SOS and significantly improves IQA model learning when only SOSs are available.
Paper Structure (11 sections, 10 equations, 4 figures, 4 tables)

This paper contains 11 sections, 10 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Proposed Method
  4. Experiments
  5. Datasets and Settings
  6. Implementation Details
  7. Performance of SOS Calibration
  8. Impact for IQA Model Learning
  9. Generalization to Few Opinion Scores
  10. Effect of $\alpha$
  11. Conclusions

Figures (4)

  • Figure 1: The overall framework of the proposed method. The input includes an image $x_n$ and its SOS $y_n$. We calibrate $y_n$ by iteratively updating the estimated MOS $\mu_{x_n}$ via our PC3, which requires a randomly selected reference image.
  • Figure 2: The variation of estimated MOS in the iterative update process of PC3.
  • Figure 3: Error bar between MOS and SOS/PC3 in VCL dataset.
  • Figure 4: Performance comparison of FOS with different numbers of subjects.