Robust infrared spectral deconvolution for image segmentation with spatial information regularization

Abstract

Spectral distortion often occurs in spectral data due to the influence of the point spread function of the infrared spectrometer. However, the infrared spectrum commonly exists the issues of random noises and band overlap. Resolution enhancement is usually the first step in the preprocessing procedure of material classification and segmentation. In this article, we have developed a resolution- enhancement algorithm with total variation (TV-norm) constraints for the degraded IR spectrum due to overlap and noises degradation. The point spread function is calculated according to infrared spectrometer system and Fourier-optical theory. Introducing the adaptive total variation, the constraint regularization, the proposed model can not only remove noises well but also estimate the point spread function simultaneously. This model is examined by a set of simulated IR spectrum with Poisson noises and a series of real IR spectra. Performance comparison with other state-of-the-art methods is made. The enhanced infrared spectrum can provide highly useful information for image segmentation.