Baseline wander is a common noise in electrocardiogram (ECG) results. To effectively correct the baseline and to preserve more underlying components of an ECG signal

we propose a simple and novel filtering method based on a statistical weighted moving average filter. Supposed

a

and

b

are the minimum and maximum of all sample values within a moving window

respectively. First

the whole region [

a

b

] is divided into

M

equal sub-regions without overlap. Second

three sub-regions with the largest sample distribution probabilities are chosen (except

M

<

3) and incorporated into one region

denoted as [

a

0

b

0

] for simplicity. Third

for every sample point in the moving window

its weight is set to 1 if its value falls in [

a

0

b

0

]; otherwise

its weight is 0. Last

all sample points with weight 1 are averaged to estimate the baseline. The algorithm was tested by simulated ECG signal and real ECG signal from

www.physionet.org

www.physionet.org

. The results showed that the proposed filter could more effectively extract baseline wander from ECG signal and affect the morphological feature of ECG signal considerably less than both the traditional moving average filter and wavelet package translation did.