Search

This article proposes a motion-robust video-based respiratory rate (RR) monitoring method. Specifically, the chest and abdomen regions of the human body are detected and tracked as the region of interest (ROI), which are further divided into patches. The dense inverse search (DIS) is then employed to calculate the optical flow within each patch. The phase signal of the optical flow is taken as the raw respiratory signal, which is denoised by a normalized least mean square (NLMS) adaptive filtering with the horizontal flow signal as the motion noise reference. Finally, a quality-guided RR pulse selection method in the time–frequency domain is designed to fuse the denoised RR pulses from all patches. The proposed approach is evaluated on the in-house BSIPL-RR dataset and the public COHFACE dataset. Experimental results indicate that this method outperforms the state-of-the-art (SOTA) comparison methods. This research effectively extends the scope of video-based RR measurements in real-world applications.

The video-based frequency measurement method for bridge cable vibration offers advantages such as speed, efficiency, and noncontact compared to traditional sensor-based methods. However, the presence of complex backgrounds in video images can affect the accuracy of cable frequency measurement. To address the problem, a novel phase-based frequency measurement method is proposed, which focuses on extracting cable edge vibration from background noise in the spatial and temporal domains. First, in the spatial domain, to process the vibration signals more precisely, each video sequence is divided into multiple subregions. To enhance the edge vibration within the subregions while initially suppressing background noise, the Otsu threshold segmentation method (OTSM) is employed for subregion categorization. Subsequently, the phase-based vibration estimation method is utilized to build the spatial domain vibration representation of the subregions based on the phase differences between adjacent frames while maintaining optical flow consistency. Then, the temporal vibrational waveforms are extracted, which may still include noise from the background edges. To restore the cable vibration, a combination of singular spectrum analysis and nonnegative matrix factorization (NMF) is further designed for characterizing cable vibrations and attenuating the noise in the temporal domain. Finally, the cable vibration restored from all subregions is synthesized, forming the ultimate cable signal. The proposed method has been evaluated through extensive testing in outdoor environments, and it has exhibited remarkable enhancements in measuring cable vibration frequencies when dealing with complex background interference compared to the existing methods.

Remote photoplethysmography (rPPG) enables heart rate (HR) measurement under stable illumination and low noise conditions. However, challenges arise due to rapid lighting changes, significant head movements, and vehicle vibrations during driving, impacting the recovery of clear rPPG signals and rendering denoising techniques alone inadequate in entirely eliminating noise interference. To address these challenges, we introduce an innovative approach, “quality-guided spectrum peak screening” (QSPS), for monitoring the driver’s HR during driving. First, we developed a signal evaluation framework for assessing signal quality across multiple facial regions following wavelet filtering. Subsequently, by leveraging quality information, a spectral peak screening algorithm is applied to signals from multiple regions, mitigating the impact of residual noise on rPPG. The precise HR is determined by integrating quality scores and short-term stability from various facial regions. Our method is employed in a commercially available driver monitoring system (DMS) equipped with a monochrome camera. Test results show QSPS’s robustness compared with established near-infrared (NIR)-based HR detection methods. In driving scenarios, QSPS achieves a mean absolute error (MAE) of 4.32 bpm, a root mean square error (RMSE) of 6.15 bpm, and a percentage of time with HR estimation errors smaller than 6 bpm of 68.8%. Furthermore, QSPS exhibits excellent performance during extended 10-min HR monitoring, with a Pearson correlation coefficient of 0.943 in the night test and 0.906 in the day test.