A QRS detection algorithm based on Haar wavelet

An on-line QRS detection algorithm was developed. It is based on the Haar wavelet and implemented as a recursive filter. Because it works with fixed-point arithmetic and it avoids overflow error, there is no cumulative error. The performance of the detector was evaluated: 1) On a set of clinical records. 2) On a normal segment of lead II. Here, it was corrupted adding electromyographic and powerline interferences, baseline drift due to respiration and electrode motion artifacts. A detector easy to implement, fast, with performance of 99.23% of detections on maximal noise-corrupted ECG was obtained; moreover, this detector is able to support, without changes in its parameters, a considerable variation of the QRS shape, with a performance of 99.14% of detections on clinical records