Research on Feature Extraction of Human ECG Identification

Abstract

ECG, a kind of identity information in human identification, can be easily affected by detector's movement and the change of mood. Traditional ECG feature extraction methods require a lot of calculation which increases the computation time in the process of identification. In this paper, we mainly discuss how to seize the main, easy to extract and stable ECG feature, to complete the identification successfully and quickly. And we propose a method of feature extraction that as for intervals, we only choose QRS interval which always maintain the same. Experiments show that the identification speed has been improved when the accuracy of recognition has been little affected.