Blind deconvolution infrared spectrogram super-resolution restoration method

Abstract

The invention discloses a blind deconvolution infrared spectrogram super-resolution restoration method. The method comprises the steps of performing normalization processing on infrared spectral data; calculating a first-order derivative of spectral intensity after normalization; estimating the noise intensity of a spectrum according to the first-order derivative of the spectrum so as to determine the value of a regularization parameter; establishing an infrared spectrum model; and solving a high-resolution infrared spectrum by utilizing a Huber-Markov field, a spectral deconvolution algorithm and an alternating minimum optimal iteration algorithm. According to the method, noise suppression is enhanced in a flat region of the spectrum, and spectral structure preservation is enhanced in a steep region of the spectrum; for the used Fourier infrared spectrum, the spatial-domain iterative blind deconvolution algorithm is good in effect; when a signal-to-noise ratio is higher than 30 dB, the spectral resolution is increased by about 30%; and priori knowledge of Huber-Markov adaptive constraints is effectively utilized, the application of the infrared spectrum to the fields of target detection and identification as well as agriculture, biology, medical science, meteorology, space technology and petrochemical industry is promoted, and technical support is provided.

Description

technical field [0001] The invention relates to the technical field of infrared spectrum data processing, in particular to a super-resolution restoration method for blind deconvolution infrared spectrum images. Background technique [0002] In recent decades, people have been actively researching on mid-wave infrared spectroscopy, which has made infrared spectroscopy widely used in the fields of target tracking, chemometric analysis, material analysis, and biomedicine. However, in the process of spectrometer measurement, infrared spectral data are often affected by factors such as spectral line collision broadening, natural broadening, and low-pass characteristics of circuits and detectors, resulting in band overlap and strong noise interference in spectral data. These interferences bring difficulties to the application of infrared spectroscopy, and even lead to serious substance identification and target recognition errors, which greatly limit the application of infrared sp...

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is to provide a blind deconvolution infrared spectrogram super-resolution recovery method for the defects in the prior art, which solves the problem that the infrared spectral data measured by the Fourier transform infrared spectrometer is often subject to random The problem of reduced spectral resolution due to the influence of noise and instrument error

Images