Visual Entropy Score (VES): A Perceptual Complexity Metric for Video Content Assessment Across Sensitive Populations
Year: 2026
Type: preprint
Abstract: Existing video content standards address photosensitive epilepsy risk through flash frequency thresholds (Harding PSE test; W3C WCAG 2.1), but no unified, publicly accessible metric characterizes the full perceptual complexity of video content across diverse sensitive populations — children, individuals on the autism spectrum, people with epilepsy, and the general adult population concerned with sleep quality and cognitive load. We propose the Visual Entropy Score (VES), a composite metric combining five complementary components: pixel change rate, flash score, motion entropy (derived from optical flow angular distribution), scene cut frequency, and contrast variance. A key contribution is the distinction between ordered motion (panning, tracking shots) and disordered motion (explosions, crowd chaos), achieved through Shannon entropy of the optical flow field rather than raw motion magnitude. The VES is computed per one-second window and aggregated across three summary statistics — median, 95th percentile, and cumulative high-entropy duration — to avoid dilution of intense but brief stimulation events. Population-specific weighting profiles are proposed for five use cases. We ground the metric in the theoretical framework of Fractal Mechanics of Consciousness (FMC), which predicts that high perceptual entropy disrupts attractor consolidation in the neural energy landscape and impairs theta-level coherence. Applications include pre-sleep content recommendations, autism-adapted media curation, pediatric content regulation, post-concussion recovery protocols, and advertising regulation on streaming platforms.
Author: Leroy, Rémi
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7222434166
Cites:
Cited by:
Related to: