Adaptive signal enhancement for overlapped peaks based on weighting factor selection

Abstract

Peak-sharpening is an effective method for the peak position detection of overlapped spectra. However, the weighing factor parameter strongly affects the sharpening performance, and the derivative adopted in the peak-sharpening method is sensitive to noise. In this paper, an adaptive peak-sharpening method based on weighting factor selection is proposed. The relationship between the sharpening ratio and weighting factor is studied. In addition, the Savitzky–Golay filter is adopted due to its excellent noise reduction and peak shape retention abilities. First, the smoothed signal and second-order derivative signal are obtained by the Savitzky–Golay filter. Then, the parameters of the overlapped peaks are estimated for the weighting factor selection. Next, the peak position is detected by the peak-sharpening method. After that step, the estimated parameters are updated, and the above steps are iterated until the detection of the peak position converges. Finally, the converged results are considered to be the final detection results. The experimental results using a simulated dataset, a virtual mass spectra dataset and a polarography dataset show that the proposed method is effective for peak position detection.