Spline wavelet overlapped peaks analysis

The proposed method of spline wavelet overlapped peaks analysis is based on the model of wavelet multifrequency channel decomposition. If the suitable optimal wavelet basis and frequency scale value are selected, the useful information can be extracted from noised overlapped signal. Then, according to the linear relationship between individual components in blurred version, multiple linear regression in low frequency domain is employed to resolve highly overlapped peaks. The results show that even signal-to-noise (S/N) decreases to 0.2, relative errors of peak intensity are less than 12% and the sum of square deviations Q is less than 0.3.