Brief Paper: LTE-QA: Luminance and Texture Entropy for CCTV Image Quality Assessment

Real world CCTV imagery is often perceived as low quality even without synthetic distortion, since operational conditions such as poor illumination and sparse texture limit visibility. Conventional no-reference image quality assessment (NR-IQA) methods, designed for natural images with artificial distortions, do not capture this surveillance-specific behavior well. We analyze human perceptual quality on a CCTV IQA benchmark and find that two interpretable cues, luminance entropy and texture entropy, are strongly correlated with subjective mean opinion scores (MOS). Motivated by this, we propose LTE-QA, a transformer-based NR-IQA model that takes spatial maps of these entropy measures as direct input, achieving higher correlation with subjective ratings than representative NR-IQA baselines.