To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
A New ECG-based Identification System using Morphology Scaling and Statistical Parameters |
|
Author |
Hamid Jannati† and Mohammad Hossein DoostMohammadi |
| Citation |
Vol. 26 No. 5 pp. 35-40 |
|
Abstract |
The uniqueness of electrocardiographic (ECG) signals and robustness against copying or forgery has led to use them for constructing ECG-based identification algorithms. However, one of the challenges in the design of ECG based identification algorithms is the feature extraction from these signals. This procedure typically is performed by techniques based on conversion or waveform (i.e. morphology). Diagnosis and determination of parameters of morphology techniques are troublesome tasks and there would be high possibility of deficiencies and error while using morphology parameters. In addition, due to the changes in these parameters in different cycles of heart rate, relative similarities between these parameters from different people may be established, which would reduce the performance of morphological characters for identification purpose. Therefore, a new algorithm based on statistical parameters of the signal and parameters scaling is proposed to overcome the mentioned problems in this study. Performance of the proposed method is evaluated in a statistical community including 20 to 69 people, with population growth step of 5%. Results show that in all cases, the whole community is identified with 100% accuracy. This shows that compared with the morphological parameters method, the proposed algorithm improves the accuracy from 5% up to 22/64%. This shows that the proposed approach has better performance especially in large statistical communities. |
|
Keywords |
Electrocardiogram (ECG); Identification; Morphology scaling; Statistical parameter |
|
URL |
