Figure 4.

Overview of the gold-standard SCG${}^{\mathrm{g}}$ and vision-based SCG${}^{\mathrm{v}}$ preprocessing. The output of the microphone was recorded simultaneously by both the camera and DAQ to synchronize the signals. The raw SCG${}^{\mathrm{g}}$ was bandpass filtered and the SCG${}^{\mathrm{v}}$ was resampled to 5000 Hz. Both signals were then segmented using the ECG R waves, and the ensemble averages of their segments were computed. The time domain similarity analysis was performed between the ensemble averages of the SCG segments. For the time–frequency analysis, the whole preprocessed signals were used.