Image fusion-based spectroscopic analysis method with enhanced precision

Abstract

Spectroscopy and spectroscopic techniques provide precise information about chemical compositions and molecular structures of substances. However, high mechanical complexity and significant manual reading errors often hindered conventional spectrometers. In this study, a high-precision spectroscopic analysis method based on image fusion is proposed, integrating CMOS devices with high-precision angle sensors. A novel approach was established to constructing a fully closed-loop measurement system for adaptive machine vision. Our method combines the hybrid Hough transform algorithm with the extended Kalman filter (EKF) algorithms, utilizing spectral response calibration, Monte Carlo sensitivity analysis (a stochastic evaluation of system response uncertainty through 10 ⁴ iterations of random parameter sampling), and a real-time diffraction angle compensation algorithm. As a result, the absolute wavelength uncertainty of the constructed system was reduced to 0.38 nm, with a relative uncertainty of less than 0.09%, representing an order-of-magnitude improvement compared to conventional methods. In comparison with the traditional spectrometer systems, the relative uncertainty was reduced by 85.7%-94.6% (*p*<0.01), and the manual adjustment time was reduced by 94.7%. The research method delineated in this paper can be extended to a wide range of applications, including industrial measurements, biological detection, and other scenarios, through multi-sensor spectral fusion (400–900 nm) and environmental compensation algorithms.