Exploring Image Quality Assessment from a New Perspective: Pupil Size CCS: Computing methodologies Image processing
Yixuan Gao, Xiongkuo Min, Guangtao Zhai · 2025
Read Exploring Image Quality Assessment from a New Perspective: Pupil Size CCS: Computing methodologies Image processing by Yixuan Gao, Xiongkuo Min, Guangtao Zhai (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-13841.txt.
Introduction
Yixuan Gao, Xiongkuo Min, and Guangtao Zhai study image quality judgment through pupil size in a 2025 arXiv paper. They ask what happens in the mind when people rate image quality. Pupil size offers a window on that hidden process.
Subjects took part in two tasks: free viewing of images and scoring image quality. The team compared pupil size between the tasks. Scoring seemed to switch on a visual attention mechanism that free viewing did not. Pupil changes during scoring also tracked the quality of the image closely. The work gives theory for objective quality tools and a new way to gather honest quality impressions.
Catalog record
| CREATOR | Yixuan Gao, Xiongkuo Min, Guangtao Zhai |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-13841
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Exploring Image Quality Assessment from a New Perspective: Pupil Size CCS: Computing methodologies Image processing Yixuan Gao, Xiongkuo Min, Guangtao Zhai Abstract . This paper explores how the image quality assessment (IQA) task affects the cognitive processes of people from the perspective of pupil size and studies the relationship between pupil size and image quality. Specifically, we first invited subjects to participate in a subjective experiment, which includes two tasks: free observation and IQA. In the free observation task, subjects did not need to perform any action, and they only needed to observe images as they usually do with an album. In the IQA task, subjects were required to score images according to their overall impression of image quality. Then, by analyzing the difference in pupil size between the two tasks, we find that people may activate the visual attention mechanism when evaluating image quality. Meanwhile, we also find that the change in pupil size is closely related to image quality in the IQA task. For future research on IQA, this research can not only provide a theoretical basis for the objective IQA method and promote the development of more effective objective IQA methods, but also provide a new subjective IQA method for collecting the authentic subjective impression of image quality. 1. Introduction Recently, image quality assessment (IQA) has become a hot research field ( Wang et al. 2004 ; Zhang et al. 2011 ; Mittal et al. 2012 ; Min et al. 2018 ; Zhai et al. 2012 ; Gao et al. 2022b ) . Specifically, IQA can be divided into subjective IQA ( H. R. Sheikh, Z. Wang, L. Cormack, and A. C. Bovik 2005 ; Virtanen et al. 2014 ; Hosu et al. 2020 ) and objective IQA ( Radun et al. 2008 ; Gao et al. 2017 ; Min et al. 2017 ; Zhai and Min 2020 ; Gao et al. 2022a ) . Subjective IQA measures image quality by integrating subjective opinion scores given by subjects. Objective IQA computes image quality through some computational models, which generally model human vision. For example, the contrast sensitivity function, just noticeable difference model, and visual attention mechanism are frequently used to model the human visual system (HVS) to calculate image quality ( Fang et al. 2019 ; Gao et al. 2009 ; Liu and Heynderickx 2011 ; Cao et al. 2023 ) . The successful applications of these HVS-based objective IQA methods are benefited from sufficient research on human visual perception. Vision is an important way for people to obtain and perceive image information. HVS can obtain visual information by controlling the eye movement or the constriction and dilation of the pupil, and transmit the obtained information to the brain for further analysis. As the gateway through which people observe the world, the pupil plays an important role in human visual perception. The pupil is in the center of the iris. Changes in pupil size, that is, pupil constriction and dilation, can control the amount of light emitted to the retina, which reflect the basic physiological reactions of human vision caused by changes in ambient brightness ( Woodhouse and Campbell 1975 ) . In addition, changes in pupil size can also reflect the task-evoked cognitive processes of people. For example, Kahneman et al . ( Kahneman and Beatty 1966 ) analyzed the task-evoked pupil size in short-term memory tasks. They found that both the difficulty and number of tasks affect pupil size. Rieger et al . ( Rieger and Savin-Williams 2012 ) proposed that the pupil dilates in response to attractive social partners. Some studies on the relationship between emotional stimuli and pupil size showed that the dilated pupil indicates a positive stimulus and the constricted pupil indicates a negative stimulus ( Hess and Polt 1960 ; Mudd et al. 1990 ) . Authors in ( Sabatino DiCriscio et al. 2018 ) explored the differences in pupil responses to local and global information. They found that pupils constrict significantly when people focus on local information rather than global information. Generally speaking, changes in pupil size are different for different tasks. Therefore, this paper first explores how the IQA task affects the cognitive processes of people from the perspective of pupil size. The performance of objective IQA methods is usually measured by the correlation between the calculated image quality and the image quality obtained from subjective experiments. In other words, both subjective IQA methods and objective IQA methods are sensitive to the subjective opinion scores collected from subjects. Recently, researchers have found that the subjective opinion scores collected by the five-level rating scale proposed by ITU-R ( Union 2002 ) may have some errors in reflecting the authentic human visual perception brought by images ( Laparra-Hernández et al. 2009 ) . For example, some subjects may be dishonest when evaluating image quality. In order to solve this problem, researchers have collected people’s authentic impressions of image quality by studying human physiological signals. Al-Juboori et al . ( Al-Juboori et al. 2017 ) investigated the relationship between changes in the electroencephalogram and subjective quality scores for high dynamic range images. Gutu et al . ( Gutu et al. 2014 ) studied the relationship between facial electromyography and image quality. Although these methods are effective, it is inconvenient to carry out subjective experiments because of the heavy apparatuses. In contrast, pupil size can be measured with a portable eye tracker. In addition, some studies have proved that the pupil responds to both the low-level (color and structure in an image) and high-level (content semantics) processing in the HVS ( Hess 1972 ; Pappusetty et al. 2017 ) . Therefore, this paper also studies whether different image quality levels can be measured by changes in pupil size when people observe images. Our work has following contributions. First, this paper explores how the IQA task affects the cognitive processes of people from the perspective of pupil size, which can provide a solid theoretical basis for the objective IQA method and promote the development of more effective objective IQA methods. Second, the relationship between the change in pupil size and image quality is investigated, which can be used to guide the subjective IQA method to collect the authentic subjective impression of image quality using pupil size. The remaining framework of the paper is organized as follows. In Section 2 , we introduce the subjective experiment in detail. Section 3 analyzes the obtained pupil data. In Section 4 , we explore the relationship between image quality and pupil size. Finally, a summary of the paper is given in Section 5 . Figure 1. 20 reference images selected from the LIVE database. Figure 2. Distribution of quality of all selected distorted images provided by the LIVE database. 2. Subjective Experiment The subjective experiment includes two tasks: free observation and IQA. The details of the experiment, including stimulus materials, subjects, apparatus, and experimental procedures, are described as follows. 2.1. Stimulus The stimulus materials used in the subjective experiment were 100 distorted images selected from the LIVE database ( H. R. Sheikh, Z. Wang, L. Cormack, and A. C. Bovik 2005 ) . Specifically, we first selected 20 reference images from the LIVE database. Fig. 1 shows the