This paper presents an emerging research area of using eye response for video quality evaluation. Experiments were conducted to study how eye movements such as saccades and fixations in a video, which represents visual information processing, react to changes in video quality. Discussed in this paper are two sets of experiments on eye movements. The first experiment had multiple original movie video clips with quality variation applied across the entire scene. The second round of tests had multiple original movie video clips with transitioning between extreme qualities half way through within the video. The key contributions of this paper are an exploration of gaze response to video quality variation and results of the experiment show a significant difference in Fixations/Saccades ratio between low and high-quality videos. Gaze response to video quality shows a potential to be considered as one of the quality evaluation metrics for future studies.
Subjective video quality evaluation techniques usually involves a subject voluntarily attending to several regions in a video in order to scrutinize its quality. These techniques often tend to over-estimate the visual thresholds and in cases of non-uniform quality/coding, subjects mostly perceive the underlying video content in worse quality. This occurs due to a process known as Attentional Modulation occurring in the higher visual cortical areas V2 to V4. Examining disruptions in free viewing gaze patterns on the other hand, are said to be a more naturalistic method to measure the perceived video quality in such cases. To explore the feasibility of such a gaze disruption based quality metric, we examine the dependency between the two indicators: gaze disruptions and perceived subjective quality, obtained from a carefully controlled subjective test. By the examination of eye-tracking data and subsequent statistical analysis of difference opinion scores given by users, we are able to see that disruptions are indeed excellent indicators of perceived quality achieving a correlation 0.84. Several state of the art objective video quality metrics like SSIM, VIFp, VQM and PSNR-HVS (designed mostly for evaluation of uniform-quality) on the other hand, only produce a correlation ranging from 0.01 to 0.10. We conclude therefore that gaze disruptions may be used as excellent natural indicators of perceived quality in cases where quality is non-uniform, and may serve as new ground truth indicators for objective algorithms like Scanpath disruption(prediction) metrics that measure video quality in a more real-world like(naturalistic) manner.
The relationship between fixations and saccades that characterize eye movements have suggested the existence of two scene processing strategies- ambient and focal - the former is believed to capture the context of the scene whereas the latter helps in the detailed examination of the scene. Until recently, detailed analysis of these processing strategies have been performed only in young adults. Recently, however, these modes were found to exist in the viewing behavior of children as well. In the current study, we investigated two undiscovered aspects related to these processing modes- first, we studied the combined impact of visual processing modes and observer's age on the distribution of gaze locations, and second, we investigated how bottom up features of a scene influence the gaze behavior during ambient and focal modes across age groups. These analysis were performed over the eye tracking data collected over 50 observers of different age groups while viewing naturalistic scenes. Explorativeness and bit rate measures were developed to investigate the changes in gaze distribution. The result showed that an observer is more explorative during the focal mode than during the ambient processing mode, while the information processing rate follows a reverse trend. The influence of bottom up features was investigated using the area under the curve (AUC) metric and it was found that the bottom-up influence was more dominant during focal than the ambient mode for all age groups. We also investigated whether human face attracted attention and gaze differently during ambient and focal mode. The result showed that face were equally gazed during both processing modes.
To understand if art experts and novices view paintings differently, we conducted a series of experiments where we asked participants to look at digital images of paintings while we recorded their eye movements. The expert participants were recruited among students and faculty studying or practicing art. Novice participants did not study art-related disciplines. Half of the participants in each group received a free viewing instruction. The second half was told that they would be asked questions about the images they viewed. Gaze trajectories (scanpaths) were recorded using an SMI remote red 250 eye tracker. To analyze the differences between art viewing patterns of experts and novices, and for different instruction conditions, we employed Recurrence Quantification Analysis (RQA), which was successfully used in previous research [5] to reveal the influence of expertise in medical image viewing. Our results indicate that expertise was, indeed, a significant factor influencing eye viewing patterns in terms of several extracted RQA measures. The instruction condition and painting type were also significant.