Study of human identification by electrocardiogram waveform morph

  • Authors:

  • Gang Zheng;Zhong-Yi Li;Tong-Tong Liu;Min Dai

  • Affiliations:

  • Laboratory of bio signal and intelligent processing, Tianjin University of Technology, Tianjin, China and School of Computer and Communication Engineering;Laboratory of bio signal and intelligent processing, Tianjin University of Technology, Tianjin, China and School of Computer and Communication Engineering;Laboratory of bio signal and intelligent processing, Tianjin University of Technology, Tianjin, China and School of Computer and Communication Engineering;Laboratory of bio signal and intelligent processing, Tianjin University of Technology, Tianjin, China and School of Computer and Communication Engineering

  • Venue:
  • CCBR'11 Proceedings of the 6th Chinese conference on Biometric recognition
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

A new biometric recognition material electrocardiogram (ECG) waveform was developed rapidly in recent ten years. Except the common feature of biometric recognition, its unique "aliveness" and individual difference of heart geometric structure, This make ECG waveform has becoming a kind of high security level biometric. The paper proposed a similarity measurement strategy to do recognition work by ECG waveform. It uses the ECG waveforms collected from one person as his or her ECG waveform sample set. These ECG waveforms were partitioned into single ECG waveform (which is generate by one heart beat) firstly. And the discrete points that formed single ECG waveforms were overlaid into a two dimension coordinate. The points which have same coordinate will be accumulated, and the color of this coordinate will be changed into a 8 bit color system according to the overlay number of point. After the procedures, it presents a hierarchy color changing image, which can be used as a tunnel like ECG morph. Moreover, the inclusive degree can be computed by color clustering status. In the end, the morph will become the morph model of the person to judge the belonging of new ECG data. The ECG data used in the paper is from MIT/BIH ECG standard data set. From the results, the recognition accurate of ECG recognition can reach to 95.1% averagely.