《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (8): 2699-2707.DOI: 10.11772/j.issn.1001-9081.2025070889
• 前沿与综合应用 • 上一篇
收稿日期:2025-08-04
修回日期:2025-10-29
接受日期:2025-10-29
发布日期:2025-12-22
出版日期:2026-08-10
通讯作者:
姚育东
作者简介:高艳(2000—),女,河南商丘人,硕士研究生,主要研究方向:深度学习、非接触式心率测量
Yan GAO1,2, Yudong YAO3(
)
Received:2025-08-04
Revised:2025-10-29
Accepted:2025-10-29
Online:2025-12-22
Published:2026-08-10
Contact:
Yudong YAO
About author:GAO Yan, born in 2000, M. S. candidate. Her research interests include deep learning, non-contact heart rate measurement.
摘要:
基于远程光电容积脉搏波(rPPG)技术的非接触式心率测量方法主要面临两个方面的挑战:一是准确选取面部感兴趣区域(RoI);二是从长视频序列中准确理解rPPG的周期模式。针对现有工作中面部RoI定位不精准和未考虑到不同信号源之间的时间滞后性的问题,提出一个端到端的rPPG信号提取模型RppgMamba,用于非接触式心率测量。在该模型中,设计的具有交叉扫描机制的分层时空Mamba模块从多尺度的时空感受场中挖掘弱信号的空间感知关系,以自适应地定位面部RoI,从而避免受到背景和运动噪声等带来的干扰;而设计的频域细化模块在频域中进一步学习rPPG信号的准周期模式,以提高输出信号质量,同时结合相位补偿器对输出信号进行相位校正,减少固有时间滞后性带来的影响。实验结果表明,所提模型在三个公开数据集上获得了最佳的性能,验证了该模型对心率测量良好的精确度和适用性。
中图分类号:
高艳, 姚育东. 基于时空Mamba模型的生理信号提取和心率测量[J]. 计算机应用, 2026, 46(8): 2699-2707.
Yan GAO, Yudong YAO. Physiological signal extraction and heart rate measurement based on spatial-temporal Mamba model[J]. Journal of Computer Applications, 2026, 46(8): 2699-2707.
| 模型 | UBFC-rPPG | PURE | UBFC-Phys | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | |
| TS-CAN[ | 1.70 | 2.72 | — | 0.99 | — | 2.48 | 9.01 | — | 0.92 | — | 20.71 | 26.36 | 22.11 | -0.28 | -14.62 |
| DeepPhys[ | 6.27 | 10.82 | — | 0.65 | — | 0.83 | 1.54 | — | 0.99 | — | 20.11 | 25.01 | 21.60 | -0.12 | -15.74 |
| EfficientPhys[ | 1.14 | 1.81 | — | 0.99 | — | — | — | — | — | — | 24.12 | 34.74 | 27.49 | -0.32 | -17.66 |
| PhysFormer[ | 0.50 | 0.71 | — | 0.99 | — | 1.10 | 1.75 | — | 0.99 | — | 16.17 | 22.24 | 18.59 | -0.08 | -10.76 |
| PhysNet[ | 2.95 | 3.67 | — | 0.97 | — | 2.10 | 2.60 | — | 0.99 | — | 7.55 | 11.56 | 9.10 | 0.42 | -9.55 |
| RhythmFormer[ | 0.50 | 0.78 | — | 0.99 | — | 0.27 | 0.47 | — | 0.99 | — | 4.96 | 7.59 | 6.26 | 0.60 | -6.79 |
| RhythmMamba[ | 0.50 | 0.75 | — | 0.99 | — | 0.23 | 0.34 | — | 0.99 | — | 6.54 | 10.62 | 8.61 | 0.49 | -5.34 |
| PhysMamba[ | 0.54 | 0.76 | — | 0.99 | — | 0.25 | 0.40 | — | 0.99 | — | 15.85 | 21.68 | 20.58 | 0.02 | -11.41 |
| RppgMamba | 0.37 | 0.44 | 0.27 | 0.99 | 8.52 | 0.20 | 0.30 | 0.15 | 0.99 | 18.10 | 5.13 | 6.97 | 6.03 | 0.75 | -5.50 |
表1 在UBFC-rPPG、 PURE和UBFC-Phys数据集上的实验结果
Tab. 1 Experimental results on UBFC-rPPG, PURE and UBFC-Phys datasets
| 模型 | UBFC-rPPG | PURE | UBFC-Phys | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | |
| TS-CAN[ | 1.70 | 2.72 | — | 0.99 | — | 2.48 | 9.01 | — | 0.92 | — | 20.71 | 26.36 | 22.11 | -0.28 | -14.62 |
| DeepPhys[ | 6.27 | 10.82 | — | 0.65 | — | 0.83 | 1.54 | — | 0.99 | — | 20.11 | 25.01 | 21.60 | -0.12 | -15.74 |
| EfficientPhys[ | 1.14 | 1.81 | — | 0.99 | — | — | — | — | — | — | 24.12 | 34.74 | 27.49 | -0.32 | -17.66 |
| PhysFormer[ | 0.50 | 0.71 | — | 0.99 | — | 1.10 | 1.75 | — | 0.99 | — | 16.17 | 22.24 | 18.59 | -0.08 | -10.76 |
| PhysNet[ | 2.95 | 3.67 | — | 0.97 | — | 2.10 | 2.60 | — | 0.99 | — | 7.55 | 11.56 | 9.10 | 0.42 | -9.55 |
| RhythmFormer[ | 0.50 | 0.78 | — | 0.99 | — | 0.27 | 0.47 | — | 0.99 | — | 4.96 | 7.59 | 6.26 | 0.60 | -6.79 |
| RhythmMamba[ | 0.50 | 0.75 | — | 0.99 | — | 0.23 | 0.34 | — | 0.99 | — | 6.54 | 10.62 | 8.61 | 0.49 | -5.34 |
| PhysMamba[ | 0.54 | 0.76 | — | 0.99 | — | 0.25 | 0.40 | — | 0.99 | — | 15.85 | 21.68 | 20.58 | 0.02 | -11.41 |
| RppgMamba | 0.37 | 0.44 | 0.27 | 0.99 | 8.52 | 0.20 | 0.30 | 0.15 | 0.99 | 18.10 | 5.13 | 6.97 | 6.03 | 0.75 | -5.50 |
| 模型 | 训练数据集 | 测试数据集 | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UBFC-rPPG | PURE | UBFC-Phys | ||||||||||||||
| MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | ||
| TS-CAN | UBFC-rPPG | — | — | — | — | — | 3.69 | 13.80 | 3.39 | 0.82 | 5.26 | 4.95 | 9.32 | 5.68 | 0.71 | -1.62 |
| PURE | 1.30 | 2.87 | 1.50 | 0.99 | 1.49 | — | — | — | — | — | 6.32 | 11.51 | 7.28 | 0.59 | -3.86 | |
| DeepPhys | UBFC-rPPG | — | — | — | — | — | 5.54 | 18.51 | 5.32 | 0.66 | 4.40 | 10.10 | 17.81 | 11.28 | 0.35 | -3.22 |
| PURE | 1.21 | 2.90 | 1.42 | 0.99 | 1.74 | — | — | — | — | — | 10.57 | 17.15 | 11.87 | 0.36 | -4.06 | |
| EfficientPhys | UBFC-rPPG | — | — | — | — | — | 5.47 | 17.04 | 5.40 | 0.71 | 4.09 | 8.12 | 17.67 | 8.80 | 0.32 | -1.37 |
| PURE | 2.07 | 6.32 | 2.10 | 0.94 | -0.12 | — | — | — | — | — | 7.78 | 17.23 | 8.55 | 0.36 | -2.21 | |
| PhysFormer | UBFC-rPPG | — | — | — | — | — | 12.92 | 24.36 | 23.92 | 0.47 | 2.16 | 5.11 | 8.79 | 6.35 | 0.70 | -2.71 |
| PURE | 1.44 | 3.77 | 1.66 | 0.98 | 0.18 | — | — | — | — | — | 4.20 | 8.31 | 4.97 | 0.72 | -1.38 | |
| PhysNet | UBFC-rPPG | — | — | — | — | — | 8.06 | 19.71 | 13.67 | 0.61 | 6.68 | 2.98 | 5.69 | 3.78 | 0.88 | -0.23 |
| PURE | 0.98 | 2.48 | 1.12 | 0.99 | 1.49 | — | — | — | — | — | 3.39 | 6.72 | 4.00 | 0.81 | -0.80 | |
| RhythmFormer | UBFC-rPPG | — | — | — | — | — | 0.97 | 3.36 | 1.60 | 0.99 | 12.01 | 3.69 | 6.64 | 4.69 | 0.84 | -1.08 |
| PURE | 0.89 | 1.83 | 0.97 | 0.99 | 6.05 | — | — | — | — | — | 3.94 | 6.89 | 5.06 | 0.83 | -0.47 | |
| RhythmMamba | UBFC-rPPG | — | — | — | — | — | 1.98 | 6.51 | 3.59 | 0.96 | 8.94 | 4.03 | 8.28 | 5.21 | 0.73 | -0.73 |
| PURE | 0.95 | 1.83 | 1.04 | 0.99 | 6.35 | — | — | — | — | — | 3.82 | 7.21 | 4.85 | 0.80 | 0.05 | |
| PhysMamba | UBFC-rPPG | — | — | — | — | — | 1.20 | 5.99 | — | 0.97 | — | 6.85 | 13.82 | 10.50 | 0.57 | -2.86 |
| PURE | 0.97 | 1.93 | — | 0.99 | — | — | — | — | — | — | 7.41 | 14.38 | 10.76 | 0.49 | -1.70 | |
| RppgMamba | UBFC-rPPG | — | — | — | — | — | 1.31 | 6.54 | 1.48 | 0.96 | 11.82 | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| PURE | 0.68 | 1.32 | 0.90 | 0.99 | 6.53 | — | — | — | — | — | 3.04 | 6.51 | 3.75 | 0.84 | 0.92 | |
表2 跨不同数据集的实验结果
Tab. 2 Experimental results across different datasets
| 模型 | 训练数据集 | 测试数据集 | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UBFC-rPPG | PURE | UBFC-Phys | ||||||||||||||
| MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | MAE | RMSE | MAPE/% | Corr | SNR/dB | ||
| TS-CAN | UBFC-rPPG | — | — | — | — | — | 3.69 | 13.80 | 3.39 | 0.82 | 5.26 | 4.95 | 9.32 | 5.68 | 0.71 | -1.62 |
| PURE | 1.30 | 2.87 | 1.50 | 0.99 | 1.49 | — | — | — | — | — | 6.32 | 11.51 | 7.28 | 0.59 | -3.86 | |
| DeepPhys | UBFC-rPPG | — | — | — | — | — | 5.54 | 18.51 | 5.32 | 0.66 | 4.40 | 10.10 | 17.81 | 11.28 | 0.35 | -3.22 |
| PURE | 1.21 | 2.90 | 1.42 | 0.99 | 1.74 | — | — | — | — | — | 10.57 | 17.15 | 11.87 | 0.36 | -4.06 | |
| EfficientPhys | UBFC-rPPG | — | — | — | — | — | 5.47 | 17.04 | 5.40 | 0.71 | 4.09 | 8.12 | 17.67 | 8.80 | 0.32 | -1.37 |
| PURE | 2.07 | 6.32 | 2.10 | 0.94 | -0.12 | — | — | — | — | — | 7.78 | 17.23 | 8.55 | 0.36 | -2.21 | |
| PhysFormer | UBFC-rPPG | — | — | — | — | — | 12.92 | 24.36 | 23.92 | 0.47 | 2.16 | 5.11 | 8.79 | 6.35 | 0.70 | -2.71 |
| PURE | 1.44 | 3.77 | 1.66 | 0.98 | 0.18 | — | — | — | — | — | 4.20 | 8.31 | 4.97 | 0.72 | -1.38 | |
| PhysNet | UBFC-rPPG | — | — | — | — | — | 8.06 | 19.71 | 13.67 | 0.61 | 6.68 | 2.98 | 5.69 | 3.78 | 0.88 | -0.23 |
| PURE | 0.98 | 2.48 | 1.12 | 0.99 | 1.49 | — | — | — | — | — | 3.39 | 6.72 | 4.00 | 0.81 | -0.80 | |
| RhythmFormer | UBFC-rPPG | — | — | — | — | — | 0.97 | 3.36 | 1.60 | 0.99 | 12.01 | 3.69 | 6.64 | 4.69 | 0.84 | -1.08 |
| PURE | 0.89 | 1.83 | 0.97 | 0.99 | 6.05 | — | — | — | — | — | 3.94 | 6.89 | 5.06 | 0.83 | -0.47 | |
| RhythmMamba | UBFC-rPPG | — | — | — | — | — | 1.98 | 6.51 | 3.59 | 0.96 | 8.94 | 4.03 | 8.28 | 5.21 | 0.73 | -0.73 |
| PURE | 0.95 | 1.83 | 1.04 | 0.99 | 6.35 | — | — | — | — | — | 3.82 | 7.21 | 4.85 | 0.80 | 0.05 | |
| PhysMamba | UBFC-rPPG | — | — | — | — | — | 1.20 | 5.99 | — | 0.97 | — | 6.85 | 13.82 | 10.50 | 0.57 | -2.86 |
| PURE | 0.97 | 1.93 | — | 0.99 | — | — | — | — | — | — | 7.41 | 14.38 | 10.76 | 0.49 | -1.70 | |
| RppgMamba | UBFC-rPPG | — | — | — | — | — | 1.31 | 6.54 | 1.48 | 0.96 | 11.82 | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| PURE | 0.68 | 1.32 | 0.90 | 0.99 | 6.53 | — | — | — | — | — | 3.04 | 6.51 | 3.75 | 0.84 | 0.92 | |
| 帧差融合块 | RSSB | 频域细化模块 | MAE | RMSE | MAPE/% | Corr | SNR/dB |
|---|---|---|---|---|---|---|---|
| × | √ | √ | 7.37 | 12.57 | 8.64 | 0.60 | -4.77 |
| √ | × | √ | 3.96 | 7.40 | 5.09 | 0.79 | -1.18 |
| √ | √ | vanilla | 5.71 | 11.19 | 6.52 | 0.62 | -2.84 |
| √ | √ | √ | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
表3 主要模块的消融实验结果
Tab. 3 Ablation experiment results of main modules
| 帧差融合块 | RSSB | 频域细化模块 | MAE | RMSE | MAPE/% | Corr | SNR/dB |
|---|---|---|---|---|---|---|---|
| × | √ | √ | 7.37 | 12.57 | 8.64 | 0.60 | -4.77 |
| √ | × | √ | 3.96 | 7.40 | 5.09 | 0.79 | -1.18 |
| √ | √ | vanilla | 5.71 | 11.19 | 6.52 | 0.62 | -2.84 |
| √ | √ | √ | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| RSSB | MAE | RMSE | MAPE/% | Corr | SNR/dB |
|---|---|---|---|---|---|
| 双向扫描 | 3.06 | 4.89 | 4.01 | 0.85 | 0.67 |
| 交叉扫描 | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| 局部扫描 | 2.95 | 6.05 | 3.84 | 0.66 | 0.53 |
| Z字型扫描 | 3.18 | 7.20 | 3.73 | 0.72 | 0.37 |
表4 RSSB中不同扫描方式的有效性
Tab. 4 Effectiveness of different scanning methods in RSSB
| RSSB | MAE | RMSE | MAPE/% | Corr | SNR/dB |
|---|---|---|---|---|---|
| 双向扫描 | 3.06 | 4.89 | 4.01 | 0.85 | 0.67 |
| 交叉扫描 | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| 局部扫描 | 2.95 | 6.05 | 3.84 | 0.66 | 0.53 |
| Z字型扫描 | 3.18 | 7.20 | 3.73 | 0.72 | 0.37 |
| 频域细化模块 | MAE | RMSE | MAPE/% | Corr | SNR/dB |
|---|---|---|---|---|---|
| vanilla | 5.71 | 11.19 | 6.52 | 0.62 | -2.84 |
| CML | 2.55 | 5.91 | 3.10 | 0.88 | 0.65 |
| USC | 3.17 | 7.09 | 3.75 | 0.85 | -0.14 |
| CML+USC | 3.11 | 6.68 | 3.76 | 0.87 | 0.01 |
| CML&USC,k=8 | 3.46 | 7.43 | 4.01 | 0.81 | -0.46 |
| CML&USC,k=12 | 2.67 | 6.78 | 4.01 | 0.83 | 0.75 |
| CML&USC,k=16 | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| CML&USC,k=20 | 2.53 | 5.21 | 3.11 | 0.89 | 0.95 |
| CML&USC,k=24 | 2.53 | 5.32 | 3.58 | 0.88 | 0.87 |
表5 频域细化模块的有效性
Tab. 5 Effectiveness of frequency domain refinement module
| 频域细化模块 | MAE | RMSE | MAPE/% | Corr | SNR/dB |
|---|---|---|---|---|---|
| vanilla | 5.71 | 11.19 | 6.52 | 0.62 | -2.84 |
| CML | 2.55 | 5.91 | 3.10 | 0.88 | 0.65 |
| USC | 3.17 | 7.09 | 3.75 | 0.85 | -0.14 |
| CML+USC | 3.11 | 6.68 | 3.76 | 0.87 | 0.01 |
| CML&USC,k=8 | 3.46 | 7.43 | 4.01 | 0.81 | -0.46 |
| CML&USC,k=12 | 2.67 | 6.78 | 4.01 | 0.83 | 0.75 |
| CML&USC,k=16 | 2.52 | 4.70 | 3.51 | 0.89 | 1.03 |
| CML&USC,k=20 | 2.53 | 5.21 | 3.11 | 0.89 | 0.95 |
| CML&USC,k=24 | 2.53 | 5.32 | 3.58 | 0.88 | 0.87 |
| 数据集 | 左脸颊 | 右脸颊 | 额头 | 下巴 | 总体 |
|---|---|---|---|---|---|
| UBFC-rPPG | 25.2 | 33.8 | 14.7 | 17.5 | 91.2 |
| PURE | 32.5 | 28.6 | 16.3 | 13.4 | 90.8 |
表6 不同皮肤区域的精确率 (%)
Tab. 6 Precisions of different skin areas
| 数据集 | 左脸颊 | 右脸颊 | 额头 | 下巴 | 总体 |
|---|---|---|---|---|---|
| UBFC-rPPG | 25.2 | 33.8 | 14.7 | 17.5 | 91.2 |
| PURE | 32.5 | 28.6 | 16.3 | 13.4 | 90.8 |
| 模型 | 输入(T×C×H×W) | 参数量/106 | 乘法累加运算量(GMACs) | FLOPs/109 | 在GPU上的推理时间/ms |
|---|---|---|---|---|---|
| TS-CAN[ | 180×3×72×72 | 2.23 | 40.87 | 81.98 | 10.298 |
| DeepPhys[ | 180×3×72×72 | 2.23 | 40.87 | 81.98 | 9.767 |
| EfficientPhys[ | 180×3×72×72 | 2.16 | 20.64 | 41.42 | 6.098 |
| PhysFormer[ | 160×3×128×128 | 7.38 | 50.28 | 101.16 | 25.088 |
| PhysNet[ | 128×3×72×72 | 0.77 | 17.56 | 35.26 | 3.396 |
| RhythmFormer[ | 160×3×128×128 | 3.33 | 38.16 | 76.88 | 16.690 |
| RhythmMamba[ | 160×3×128×128 | 5.14 | 13.17 | 26.48 | 268.971 |
| PhysMamba[ | 128×3×128×128 | 0.73 | 48.50 | 97.51 | 21.842 |
| RppgMamba | 128×3×128×128 | 1.34 | 30.36 | 67.22 | 24.958 |
表7 计算复杂度比较
Tab. 7 Comparison of computational complexity
| 模型 | 输入(T×C×H×W) | 参数量/106 | 乘法累加运算量(GMACs) | FLOPs/109 | 在GPU上的推理时间/ms |
|---|---|---|---|---|---|
| TS-CAN[ | 180×3×72×72 | 2.23 | 40.87 | 81.98 | 10.298 |
| DeepPhys[ | 180×3×72×72 | 2.23 | 40.87 | 81.98 | 9.767 |
| EfficientPhys[ | 180×3×72×72 | 2.16 | 20.64 | 41.42 | 6.098 |
| PhysFormer[ | 160×3×128×128 | 7.38 | 50.28 | 101.16 | 25.088 |
| PhysNet[ | 128×3×72×72 | 0.77 | 17.56 | 35.26 | 3.396 |
| RhythmFormer[ | 160×3×128×128 | 3.33 | 38.16 | 76.88 | 16.690 |
| RhythmMamba[ | 160×3×128×128 | 5.14 | 13.17 | 26.48 | 268.971 |
| PhysMamba[ | 128×3×128×128 | 0.73 | 48.50 | 97.51 | 21.842 |
| RppgMamba | 128×3×128×128 | 1.34 | 30.36 | 67.22 | 24.958 |
| [1] | Xiang G, Ou Y, Li J, et al. Recovering physiological signals from facial videos: recent advances and applications in intelligent vehicles[J]. IEEE Transactions on Intelligent Vehicles, 2024, 9(10): 6576-6598 . |
| [2] | Huang B, Hu S, Liu Z, et al. Challenges and prospects of visual contactless physiological monitoring in clinical study[J]. npj Digital Medicine, 2023, 6(1): No.231. |
| [3] | Zhang X, Hu M, Zhang Y, et al. Recent progress of optical imaging approaches for noncontact physiological signal measurement: a review[J]. Advanced Intelligent Systems, 2023, 5(9): No.2200345. |
| [4] | Lewandowska M, Rumiński J, Kocejko T, et al. Measuring pulse rate with a webcam - a non-contact method for evaluating cardiac activity[C]// FedCSIS 2011. Piscataway: IEEE, 2011: 405-410. |
| [5] | Poh M Z, Mcduff D J, Picard R W. Advancements in noncontact, multiparameter physiological measurements using a webcam[J]. IEEE Transactions on Biomedical Engineering, 2010, 58(1): 7-11. |
| [6] | Verkruysse W, Svaasand L O, Nelson J S. Remote plethysmographic imaging using ambient light[J]. Optics Express, 2008, 16(26): 21434-21445. |
| [7] | de Haan G, Jeanne V. Robust pulse rate from chrominance-based rPPG[J]. IEEE Transactions on Biomedical Engineering, 2013, 60(10): 2878-2886. |
| [8] | de Haan G, van Leest A. Improved motion robustness of remote-PPG by using the blood volume pulse signature[J]. Physiological Measurement, 2014, 35(9): No.1913. |
| [9] | Wang W, Stuijk S, de Haan G. A novel algorithm for remote photoplethysmography: spatial subspace rotation[J]. IEEE Transactions on Biomedical Engineering, 2016, 63(9): 1974-1984. |
| [10] | Pilz C S, Zaunseder S, Krajewski J, et al. Local group invariance for heart rate estimation from face videos in the wild[C]// CVPRW 2018. Piscataway: IEEE, 2018: 1335-1358. |
| [11] | Wang W, den Brinker A C, Stuijk S, et al. Algorithmic principles of remote PPG[J]. IEEE Transactions on Biomedical Engineering, 2017, 64(7): 1479-1491. |
| [12] | Niu X, Yu Z, Han H, et al. Video-based remote physiological measurement via cross-verified feature disentangling[C]// ECCV 2020, LNCS 12347. Cham: Springer, 2020: 295-310. |
| [13] | Liu X, Zhang Y, Yu Z, et al. rPPG-MAE: Self-supervised pretraining with masked autoencoders for remote physiological measurements[J]. IEEE Transactions on Multimedia, 2024, 26: 7278-7293. |
| [14] | Zhong R, Zhou Y, Gou C. EST-TSANet: video-based remote heart rate measurement using temporal shift attention network and ESTmap[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 73: No.5002114. |
| [15] | Qian W, Guo D, Li K, et al. Dual-path TokenLearner for remote photoplethysmography-based physiological measurement with facial videos[J]. IEEE Transactions on Computational Social Systems, 2024, 11(3): 4465-4477. |
| [16] | Lu H, Han H, Zhou S K. Dual-GAN: joint BVP and noise modeling for remote physiological measurement[C]// CVPR 2021. Piscataway: IEEE, 2021: 12399-12408. |
| [17] | Yu Z, Li X, Zhao G. Remote photoplethysmograph signal measurement from facial videos using spatio-temporal networks[C]// BMVC 2019. Durham: BMVA Press, 2019: No.277. |
| [18] | Chen W, McDuff D. DeepPhys: video-based physiological measurement using convolutional attention networks[C]// ECCV 2018, LNCS 11206. Cham: Springer, 2018: 356-373. |
| [19] | Liu X, Fromm J, Patel S, et al. Multi-task temporal shift attention networks for on-device contactless vitals measurement[C]// NeurIPS 2020. Red Hook: Curran Associates Inc., 2020: 19400-19411. |
| [20] | Liu X, Hill B, Jiang Z, et al. EfficientPhys: enabling simple, fast and accurate camera-based cardiac measurement[C]// WACV 2023. Piscataway: IEEE, 2023: 4997-5006. |
| [21] | Liu Z, Lin Y, Cao Y, et al. Swin Transformer: hierarchical Vision Transformer using shifted windows[C]// ICCV 2021. Piscataway: IEEE, 2021: 9992-10002. |
| [22] | Yu Z, Li X, Niu X, et al. Autohr: a strong end-to-end baseline for remote heart rate measurement with neural searching[J]. IEEE Signal Processing Letters, 2020, 27: 1245-1249. |
| [23] | Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]// NeurIPS 2017. Red Hook: Curran Associates Inc., 2017: 6000-6010. |
| [24] | Yu Z, Shen Y, Shi J, et al. PhysFormer: facial video-based physiological measurement with temporal difference Transformer[C]// CVPR 2022. Piscataway: IEEE, 2022: 4176-4186. |
| [25] | Yu Z, Shen Y, Shi J, et al. PhysFormer++: facial video-based physiological measurement with SlowFast temporal difference Transformer[J]. International Journal of Computer Vision, 2023, 131(6): 1307-1330. |
| [26] | Zou B, Guo Z, Chen J, et al. RhythmFormer: extracting patterned rPPG signals based on periodic sparse attention[J]. Pattern Recognition, 2025, 164: No.111511. |
| [27] | Gu A, Dao T. Mamba: linear-time sequence modeling with selective state spaces[PP/OL]. V2. arXiv (2024-05-31) [2026-05-31].. |
| [28] | Zou B, Guo Z, Hu X, et al. RhythmMamba: fast, lightweight, and accurate remote physiological measurement[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2025, 39(10): 11077-11085. |
| [29] | Luo C, Xie Y, Yu Z. PhysMamba: efficient remote physiological measurement with SlowFast temporal difference Mamba[C]// CCBR 2024, LNCS 15353. Singapore: Springer, 2024: 248-259. |
| [30] | Yu W, Wang X. MambaOut: do we really need Mamba for vision?[C]// CVPR 2025. Piscataway: IEEE, 2025: 4484-4496. |
| [31] | Guo H, Li J, Dai T, et al. MambaIR: a simple baseline for image restoration with state-space model[C]// ECCV 2024, LNCS 15076. Cham: Springer, 2025: 222-241. |
| [32] | Hu V T, Baumann S A, Gui M, et al. ZigMa: a DiT-style Zigzag Mamba diffusion model[C]// ECCV 2024, LNCS 15079. Cham: Springer, 2025: 148-166. |
| [33] | Huang T, Pei X, You S, et al. LocalMamba: visual state space model with windowed selective scan[C]// ECCVW 2024, LNCS 15633. Cham: Springer, 2025: 12-22. |
| [34] | Yi K, Zhang Q, Fan W, et al. Frequency-domain MLPs are more effective learners in time series forecasting[C]// NeurIPS 2023. Red Hook: Curran Associates Inc., 2023: 76656-76679. |
| [35] | Stricker R, Müller S, Gross H M. Non-contact video-based pulse rate measurement on a mobile service robot[C]// ROMAN 2014. Piscataway: IEEE, 2014: 1056-1062. |
| [36] | Bobbia S, Macwan R, Benezeth Y, et al. Unsupervised skin tissue segmentation for remote photoplethysmography[J]. Pattern Recognition Letters, 2019, 124: 82-90. |
| [37] | Sabour R M, Benezeth Y, de Oliveira P, et al. UBFC-Phys: a multimodal database for psychophysiological studies of social stress[J]. IEEE Transactions on Affective Computing, 2023, 14(1): 622-636. |
| [38] | Liu X, Narayanswamy G, Paruchuri A, et al. rPPG-Toolbox: deep remote PPG toolbox[C]// NeurIPS 2023. Red Hook: Curran Associates Inc., 2023: 68485-68510. |
| [1] | 苏培权, 许亮, 梁永坚. 基于欧拉影像放大的非接触式心率测量方法[J]. 计算机应用, 2018, 38(3): 916-922. |
| 阅读次数 | ||||||
|
全文 |
|
|||||
|
摘要 |
|
|||||