Isyarat ECG itu penting dalam pemantauan sakit jantung. Namun demikian, penemuan terbaru melaporkan bahawa isyarat ECG juga boleh digunakan sebagai ciri- ciri biometrik. Keunikan isyarat ECG disebabkan oleh perbezaan dari segi bentuk dan saiz antara individu yang lain. Hal ini menyebabkan isyarat ECG yang dihasilkan adalah berbeza. Oleh sebab penyelidikan biometrik yang berdasarkan ECG mempunyai ruang untuk penyasatan, projek ini menerokai cara-cara untuk mengekstrak ciri-ciri isyarat ECG. Untuk membina sistem biometrik yang baik, teknik-teknik pra-proses iaitu penapisan dan pembahagian isyarat ECG dikajikan terdahulu. Kemudian, transformasi wavelet dikemukakan sebagai teknik pemetikan ciri-ciri isyarat ECG. Wavelet ibu “db4”, “sym3” dan “coif2” digunakan dalam transformasi isyarat ECG. Pekali yang dihasilkan dianggapkan sebagai ciri-ciri. Seterusnya, Support Vector Machine (SVM) dengan fungsi kernelnya Fungsi Radial Basis digunakan sebagai pengkelas. Beberapa variasi dari segi ketepatan, kelajuan dan praktik dibandingkan dan dinilaikan. Sistem biometrik ini dinilaikan dengan menggunakan 200 isyarat ECG daripada 20 orang sukarela. Sistem tersebut mencapai Kadar Sama Ralat (EER) pada 2.0069%. Di samping itu, sistem ini juga mencapai Kadar Penerimaan Tulen (GAR) pada 97% apabila Kadar Penerimaan Penyamar (FAR) pada 1%. Keputusan sedemikian bukan sahaja membuktikan isyarat ECG mempunyai sifat yang berpotensi dalam bidang biometrik namun juga membuktikan sistem yang dibinakan ialah praktikal.
______________________________________________________________________________________
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. This 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. In order to develop a good biometric system, the pre-processing techniques i.e. ECG signal filtering and segmentation are first studied and developed. Then, the wavelet transform is proposed as feature extraction technique. Mother wavelets “db4”, “sym3” and “coif2” are used to transform the segmented ECG signal and the coefficients produced are employed as features. In classification stage, Support Vector Machine (SVM) is used as classifier with Radial Basis Function executed as kernel function. A few variations in terms of methods and parameters of the system are built and their performances including accuracy, speed and practicality are compared and evaluated. The developed biometric system is evaluated against a total of 200 ECG signals from 20 volunteers. The system hits an Equal Error Rate (EER) of 2.0069%. The system achieves 97% Genuine Acceptance Rate (GAR) at 1% False Acceptance Rate (FAR). The experimented results not only prove that the ECG signal is a promising trait to be used as biometric features but also prove that the developed system is practical and viable.
Heartbeat biometric system based on wavelet transform algorithm_Chee Kai Jye_E3_2014_NI