K. Kuzume, K. Niijima, and S. Takano (Japan)
Lifting wavelet transform, Learning, Signal detection, Electrocardiogram
This paper presents a signal detection method based on an integer-type Haar lifting wavelet transform. We apply it to electrocardiogram (ECG) signal processing. First, we propose a method for learning free parameters contained in this transform. We learn these parameters so that the ratio of a lifting wavelet coefficient to a corresponding old one is two. We utilize the value as an indicator to identify the target signal. As a practical application, we employ the method to detect characteristic waves in ECG signals. The proposed method is also suitable to make a real-time signal detection system with a small amount of hardware.
Important Links:
Go Back
