How I Built a Real-Time rPPG Heart Rate Tracker in Python (POS Algorithm & Butterworth Filtering)

Real-time remote photoplethysmography (rPPG) heart rate monitoring can be achieved without specialized sensors using a Python-based framework called BioSense Tracker Pro. This system processes skin color variations caused by blood volume pulses, which can be captured through a standard laptop webcam. The challenge lies in overcoming ambient camera noise and specular reflections that mask the desired signal.

To address these issues, the system employs a combination of the Plane-Orthogonal-to-Skin (POS) algorithm and Butterworth bandpass filtering. The POS algorithm projects temporal RGB signals onto a plane orthogonal to the skin tone vector, separating intensity variations from pulsatile variations. This effectively eliminates specular reflections caused by lighting fluctuations.

Next, a Butterworth bandpass filter is applied to the processed signal. Human resting heart rates typically range from 45 BPM to 240 BPM, corresponding to frequencies between 0.75 Hz and 4.0 Hz. The Butterworth bandpass filter removes high-frequency sensor noise and low-frequency motion drift, ensuring a clean signal for further analysis.

The core processing pipeline is designed to run on standard CPU threads with low latency, making it suitable for real-time applications. The project utilizes the Python 3.10+ programming language along with popular libraries such as OpenCV, NumPy, and SciPy for image processing and signal manipulation tasks.

For ease of use, a standalone executable package is also available for download. This allows users to test the BioSense Tracker Pro system without the need for setting up complex Python environments or installing additional dependencies. The source code is openly accessible on the GitHub repository linked in the project description.