Gabor feature extraction for electrocardiogram signals

In this paper, the useful features for clinical diagnosis from electrocardiogram (ECG) signals have been extracted to speed up the diagnosis decision from doctors. Based on the background information of ECG signals, after analyzing some presented methods for ECG feature extraction, algorithm for each feature extraction have been proposed. The major methods for feature extraction we proposed contain short-time Fourier transform (STFT), Gabor filter, and matching process using Gaussian models with various scales. According to the experimental results, less comparative error shows that the proposed algorithm surpasses state-of-arts as stated in the literature for extracting features on ECG signals.