A Video-Based Technique for Heart Rate and Eye Blinks Rate Estimation: A Potential Solution for Telemonitoring and Remote Healthcare

Figures

• 

  Figure 1. Overview of the experimental settings. Laboratory and video–based equipment was employed to address the objectives of the study. Other acquisition devices were present although they were not used for the purposes of this study.

• 

  Figure 2. Main steps of the video-signal processing for Eye Blinks Rate (EBR) estimation. The distance between the eyelids is computed frame by frame. Then, filtering and the quadratic mean threshold are applied for obtaining the eye blinks number from the raw data.

• 

  Figure 3. Main steps of the video-signal processing for heart rate (HR) estimation. Starting from the bottom left, the facial video is recorded by mean of a PC webcam and the regions of interest (ROIs) are selected. Then, the R, G and B components are selected by mean of Principal Component Analysis (PCA) algorithm. The Heart Rate (HR) frequency is extracted after detrending, filtering and fast Fourier transformation. Finally, the HR values in time domain are obtained after z-score normalization.

• 

  Figure 4. The normalized EBR (EBR’) values evaluated through the video—based technique and the laboratory sensor during the n-Back (NB) (left image) and the Webcall (WEB) (right image) tasks did not statistically differ (all p > 0.05).

• 

  Figure 5. The normalized HR (HR’) values evaluated through the video—based technique and the laboratory sensor during the NB (left image) and the WE (right image) tasks did not statistically differ (all p > 0.05).

• 

  Figure 6. Results of the repeated measure correlation analysis on the EBR (left image) and HR (right image) estimated by the laboratory and video-based technique every 60 s.

• 

  Figure 7. The average NASA-TLX scores during the nominal (blue bar) and non-nominal (red bar) conditions (left image), and the average Self Assessment Manikin (SAM) scores in terms of arousal during the nominal and non-nominal conditions (right image). * indicates a statistical difference between the represented parameters.

• 

  Figure 8. The normalized EBR (left image) and HR (right image) values during the nominal (blue bar) and the non-nominal (red bar) conditions of the NB task. * indicates a statistical difference between the represented parameters.

• 

  Figure 9. The normalized EBR (left image) and HR (right image) values during the nominal (blue bar) and the non-nominal (red bar) conditions of the WEB task. * indicates a statistical difference between the represented parameters.