A new method based on singular value decomposition (SVD) was applied to the denoising of the time-resolved spectral matrix (TRSM), which was obtained by a streak camera. The least informative principal components (PCs) were filtered out using the Tikhonov regularization principle. A method for determining the quasi-optimal regularization parameter was suggested. The SVD method was compared with the moving average time direction (MATD) smoothing of the TRSM. Numerical simulations and experimental study of the 3-aminophthalimide (3AP) solution in acetonitrile showed that the SVD smoothing allowed for accurate measurement of signals in peak maximum. Fitting the polynomial modified Gaussian to the smoothed fluorescent bands decreased the measurement errors of the maximum peak position and its width. This method allowed us to obtain kinetics data of the fluorescence band parameters of the 3AP spectrum in an acetonitrile solution. Flexible selection of the quasi-optimal processing parameters and a better time resolution are the main advantages of the SVD method over MATD.
Read full abstract