What the y label "n.u."mean?

[TouShienn]

Hello,I see your code ,the result figure is ploted,but what the meaning of the y label "n.u."? And as your opinion,Is your methods could be seen as choosing different areas between every frames according to SNR? Is that OK for IPPG predict? Because as I see, different areas' changing frequency is not same, it could cause some troubles for calculate BPM. Hope your response ,thank you very much.

[frederic-bousefsaf]

Hello, n.u. stands for "normalized units" For the second part of your question, did you mean computing iPPG signals from different ROIs? Sincerely

[TouShienn]

From your methods, It seemed continuously frames using different skin-mask, I mean could it be seen as using different roi in the continuously frames? Such as the first frames using forehead roi ,the second frames using cheek roi ,is it OK?

[frederic-bousefsaf]

Maybe it can work if the ROIs strongly overlap, otherwise I would say no because this approach will ultimately produce noisy PPG signals (abrupt changes the moment you switch from one ROI to another).

[TouShienn]

Thank you for your advices. By the way, I used the single green channel for predict heart beat. My own video was recorded by a 30fps, 1080p camera . I use a filter and fft to find the max peak which *60 can be seen as heart beat, the result is correct. But the raw ippg signal is not good as sample from your another repository .The single green channel raw ippg signal is showed as the picture below,one is your sample raw ippg signal ,the other is mine.I want to know what cause my raw signal so worse than yours,did you use some color enhancement method in your video? Hope your response,thank you very much.

[frederic-bousefsaf]

It depends on the video: mainly the recording and the analysis process employed to compute the ppg signals... Hard for me to comment without manipulating the data.