Publication: Electrocardiogram (ecg) recognition system based on sparse representation classifier
Loading...
Date
2020-07-01
Authors
Tan, Li Wen
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
It is well known that ECG signal is important for heart disease monitoring. However, a new discovery reported that the ECG signal can be used as biometric features for its unique properties among individual. The uniqueness is due to different individuals have different shapes and sizes of heart. Consequently, the ECG signals produced are different. As the biometric research based on ECG is still at infancy and has potential for investigation, this project explores on the ways of feature extraction of ECG signals. Biometrics is best described as a personal identity verification system, meaning that your personal identity is verified by using individual and unique physical characteristics such as your fingerprints, hand geometry, speech, eyes, handwriting, or facial recognition. All human being has electrical activity in our own respective heart, and it can be measured by a test called electrocardiogram (ECG). Automated recognition of ECG is becoming more and more vital in the aspect of biological research or environmental monitoring as scientist today still unable to determine whose ECG reading that belongs to by the just listening to the sound of the heartbeat. Even if that is possible, it is going to take decades to find out. Besides, a traditional method of authentication method can be falsify easily, which causes the system to be not reliable and not safe anymore. Hence, ECG recognition based on the sound of the heartbeat is a very important topic to enhance in the aspect of human biological research. This project aims to develop an ECG recognition system through analysing human heartbeat sound. In the data acquisition stage, databases from Physionet ECG-ID database have been used to evaluate the performance of the system. Raw Human ECG files are processed using Mel-Frequency Cepstral Coefficient (MFCC) technique to extract features that will be needed in testing and training the system. In this project, the classifier used is
Sparse Representation Classifier (SRC) and Kernel Sparse Representation Classifier (KSRC). Performance between SRC and KSRC is compared and discussed in this project. Two experiments were done in this project , both using SRC and KSRC. In short, KSRC (87.03%) has a higher performance in accuracy compared to SRC (83.35%).