Electrocardiogram (ECG) is the most successful physiological signal that is measured continuously in freely moving humans for a long time and R-R intervals obtained from ECG is the standard measure for analyzing heart rate variability. However, ECG signals under daily activities often contain various noises, including those caused by theoretically inevitable sources, such as electromyograms and cardiac axial fluctuations with respiration and postural changes. As the result, even automated ECG analyzers used for clinical purposes still require the careful editing and corrections of QRS detections errors by skilled operators, which causes both economical and time consuming burden. Given the recent wide-spread of wearable ECG monitoring and its potentially life-long longitudinal data collections, the development of highly reliable QRS wave detection algorithms has become increasingly important. Therefore, this study focused on improving QRS detection accuracy in electrocardiogram with noise mixed, focusing on the usefulness of adaptive correlation filter.