BLUE: A Stale-Pixel Optical-Flow Compositor for Entropy-Efficient Surveillance Video Encoding

Shubham Baid, Akash James, Sahil Chachra, Nishant Sinha, Kunal Kislay

Continuous-recording surveillance systems face a storage problem that codec tuning alone cannot fully solve: even at aggressive CRF settings, a static-camera scene spends most of its bits re-encoding a background that has not changed. We present BLUE, a pre-encode compositor that exploits this structure by maintaining a persistent seed frame of the background and substituting background pixels with seed pixels before the encoder runs. The encoder then emits near-free SKIP macroblocks for the frozen background, while live pixels in foreground regions are carried unchanged at full quality. We evaluate BLUE on all 308 annotated short subclips from the VIRAT Ground Surveillance Release 2.0 dataset using a six-point CRF sweep with both x264 and x265. At CRF 28, BLUE reduces file size by a mean of 34.6% (x264) / 39.4% (x265) on 95.8% / 99.4% of clips respectively. Foreground-region PSNR, computed only over VIRAT object-annotation bounding boxes, is preserved or improved on 60.7% of clips (+0.36 dB mean, +5.48 dB maximum). Full-frame perceptual quality (VMAF) drops by a median of 6.75-8.59 points; we quantify and disclose this trade-off explicitly. A lightweight deployment gate measuring the compositor's own VMAF on a 2-second prefix identifies the 40% of clips where even full-frame quality degradation is near-imperceptible (Delta VMAF <= -2.9), enabling a selective-activation strategy that retains both the storage benefit and acceptable perceptual fidelity.