The Atmospheric Vertical Detection of Large Area Regions Based on Interference Signal Denoising of Weighted Adaptive Kalman Filter

Abstract

In comparison with traditional space infrared spectroscopy technology, the interference signals of a large focal plane array (FPA) can be used to obtain spectra over a larger area range and rapidly achieve large-scale coverage of hyperspectral remote sensing. However, the low signal-to-noise ratio of the interference signals limits the application of spectral data, especially when atmospheric detection occurs in the long-wavelength infrared (LWIR) band. In this paper, we construct an LWIR hyperspectral system of a Fourier transform spectrometer composed of a HgCdTe photovoltaic IR FPA and a Michelson interferometer. The LWIR interference signals are obtained by a high-frequency oversampling technique. We use the Kalman filter (KF) and its improved weighted adaptive Kalman filter (WAKF) to reduce the noise of multiple measured data of each pixel. The effect of overshoot and ringing artifacts on the objective signals is reduced by the WAKF. The applicability is studied by the interference signals from the different sampling frequencies and different pixels. The effectiveness is also verified by comparing the spectra of denoised interferograms with the reference spectrum. The experimental results show that the WAKF algorithm has excellent noise suppression, and the standard deviation of the interferogram can be reduced by 39.50% compared with that of KF. The WAKF is more advantageous in improving the signal-to-noise ratio of the interferogram and spectra. The results indicate that our system can be applied to atmospheric vertical detection and hyperspectral remote sensing over large area ranges because our denoised technique is suitable for large LWIR FPA.