Time-Domain Feature Parameter Extraction Algorithm of Pulse Wave Based on Morphological Features

Abstract

Pulse wave’s Time-domain characteristics have clear physiological and pathological significance, and are widely used in pulse recognition and disease prediction. At present, for different forms of pulse wave feature points, the accuracy of time-domain feature point extraction algorithms varies greatly. Therefore, this paper proposes a time-domain feature extraction algorithm of pulse wave based on morphological features. Firstly, the pulse wave is normalized and averaged by using normalization and correlation averaging methods. Secondly, the morphological characteristics of the main wave are divided into two states: extreme point and non-obvious feature. The characteristic morphology of the dicrotic front wave and the dicrotic wave are divided into three states: extreme point, non-obvious feature, feature fusion or disappearance. Finally, the location range of each feature point is determined, and the pulse wave feature points are identified by the curvature method and the difference method. The experiment shows that the algorithm can effectively identify the pulse wave characteristic points of common pulse waves, which creates conditions for the effective extraction of subsequent relevant human physiological and pathological information.