Dissertation > Excellent graduate degree dissertation topics show

Human Identification System Research Using Electrocardiograms Based on Wavelet and Dynamic Time Warping

Author: GuoBao
Tutor: ShiLi
School: Zhengzhou University
Course: Systems Engineering
Keywords: ECG Identification Wavelet transform Dynamic Time Warping Software Design and Implementation
CLC: TP391.41
Type: Master's thesis
Year: 2010
Downloads: 109
Quote: 0
Read: Download Dissertation

Abstract

Identification or authentication is that people have to face everyday problems, as opposed to passwords, keys and other traditional identification methods, the use of biometric technology is the body's own physiological or behavioral characteristics and for identification, with security coefficient is high, easy to lose and difficult to forget, etc., so it is increasingly being applied to security, case detection, access control, time and attendance, medical insurance and other social fields. Currently, a new biometric identification technology - the use of electrocardiogram (ECG) characteristics for identification gradually attracted wide interest of researchers at home and abroad, ECG identification appears, can more effectively prevent forgery similar to fingerprints, voice imitation other existing biometric technology the defects, but also for the current biometric identification system, it will be an effective complement. Based on this background, this paper studies the existing ECG based on the identification, we propose a new method for ECG identification and conduct a preliminary study and discussion. Details are as follows: (1) ECG single feature point extraction. According to the multi-resolution wavelet transform characteristics and clinical ECG signal singularity analysis, proposed ECG signal quadratic spline wavelet decomposition algorithm by atrous, and in the four-scale wavelet decomposition to extract the R-wave peak point strategy. Experimental results show that the algorithm can accurately extract the R-wave peak point. (2) ECG recognition feature extraction. Accurate extraction of the R-wave peak point, based on the extraction of QRS spread throughout the cardiac cycle waveform as the identifying characteristics. The QRS wave affected by changes in heart rate is small, and therefore the starting point R-wave peak point, forward and backward respectively fixed length data extracted as recognition features QRS wave, but for the entire cardiac cycle waveform extraction, the heart rate changes according to the length Variable extraction strategy, so that the waveform can be extracted including the complete P-QRS-T waves. Extracting a plurality of spread in the QRS waveform based on the cardiac cycle, the average superimposed to obtain the identifier of a solid. (3) The method for fast identification. First, according to policy ECG recognition feature extraction to extract and create a template database identification, followed by extraction of the individual to be identified ECG recognition feature, the identification of individuals according to characteristics of the QRS wave of the QRS wave template library template analysis of correlation between, combined threshold method, narrowing the scope of recognition, then the individual will be identified cardiac cycle waveform characteristics and to determine the identification of the cardiac cycle within each template by dynamic time warping algorithm (DTW) for optimal matching to select individuals with a minimum matching distance as recognition results. In the experiment, respectively, in single-lead ECG and body six combinations for identification, experimental results show that, DTW algorithm effectively solved the heart rate changes brought about by the two feature vectors length matching problem in case of inconsistency, and The proposed identification method effectively combines the fast correlation analysis and the advantages of high accuracy DTW algorithm to achieve a rapid and accurate identification, and further proof that single-lead identification under feasibility. (4) ECG identification software design and implementation. Using SQL Server 2000 database to establish a platform for ECG identity template data, and using Visual C 6.0 level programming language implementation of the system software, and finally demonstrates the ECG identification processes and recognition effect.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net Mobile