New SAC metric predicts codec robustness for rPPG heart-rate monitoring

A new study led by Achraf Ben Ahmed introduces Spatial Artifact Coherence (SAC), a physical metric that determines when spatial decomposition methods outperform global-projection approaches for remote photoplethysmography (rPPG) under video codec compression. SAC measures the ratio of off-diagonal to diagonal energy in the 4x4 inter-patch Green-channel covariance matrix (bandpass 0.75-2.5 Hz). With 280 subjects across three public datasets and 11 codec degradation variants — including MPEG-4, H.265, H.264, JPEG, and chroma subsampling — the analysis of 13 algorithms via 904 Wilcoxon tests (BH-FDR, q < 0.05) shows SAC explains 93.8% of between-variant variance in PCA advantage (r = +0.969).

Key findings include zero overlap between codec families: non-MPEG-4 variants cluster at SAC 0.10-0.18 with 84-90% PCA win rates, while MPEG-4 variants cluster at SAC 0.48-0.59 with only 61% win rate and a 5.8x reduction in mean improvement. Within subjects, 78% confirm the expected pattern (p < 10^-22, dz = 0.73). The PatchPCA algorithm family and P-Hybrid are identified as the most deployment-robust approaches. Two necessary conditions for PatchPCA advantage are SAC < 0.30 and low-to-moderate motion, directly ruling out raw-to-MPEG-4 transcoding pipelines. This provides a physically grounded metric for codec-aware rPPG algorithm selection in clinical remote monitoring systems.