Band centers and track finding for large two-dimensional infrared spectroscopic arrays

An efficient two-dimensional (2D) peak-finding algorithm is proposed to find peak maps that specify the peak centers of all bands in two-dimensional arrays of time-series infrared spectral data. The algorithm combines the second-derivative method with the intrinsic characteristics of 2D infrared reaction spectral data. Initially, the second-derivative method is used to detect all possible peak center positions, and then three criteria drawn from characteristics of 2D continuous spectral data are employed to filter peak positions. Four 2D peak maps are generated in a sequential order, with better and better approximations to the peak center positions being obtained in each. The 2D peak-finding algorithm has been successfully applied to both simulated spectra (to initially evaluate the algorithm) and then real 2D experimental spectra. The resulting peak maps exhibit very good estimates of the peak center positions. An ordering from the most significant to the least significant bands is obtained. The final peak maps can be used as starting parameters for various applications including the computationally intensive curve-fitting of time-series data.