Research on Intelligent Visual Detection Methods for Human Physiological Parameters

In recent years, the incidence and mortality rates of cardiovascular diseases have been increasing year by year, prompting greater attention to the monitoring of physiological parameters in daily life. This paper proposes a deep physiological parameter detection method based on Image Photoplethysmography (IPPG) technology, overcoming the discomfort of contact-based measurements and the limitation of measuring single physiological indices through intelligent visual means. A lightweight attention mechanism combined with the BHCNN model is designed to automatically extract waveform features of IPPG signals for estimating blood pressure and heart rate in the human body. The mean absolute errors (MAE) of the obtained systolic pressure, diastolic pressure, and heart rate are 6.60, 6.22, and 7.15 respectively, with standard deviations (STD) of 3.22, 2.61, and 3.98. The data consistency of both indices exceeds 95%, demonstrating high accuracy.