Pretreatment of Aircraft Spectrum in Visible and Near Infrared Band Based on Wavelet Transform

Abstract

In recent years, the rapid improvement of spectral resolution has brought about the explosive growth of spectral data, and the spectral characteristics of ground objects have been expressed in more detail. Different substances can be distinguished according to the spectral curves. However, the spectral curve obtained from the measurement is often accompanied by inevitable noise interference. In this paper, the pretreatment method of aircraft spectral curve based on wavelet transform was studied to find out the best combination of wavelet parameters suitable for aircraft spectral data to remove dark current noise, and to suppress the interference of environmental noise by envelope removal method. In this paper, a new threshold function was proposed, which avoided the loss of features caused by constant compression and the difficulty of parameter selection in traditional threshold denoising. Combining with minimaxi threshold criterion, 4-layer decomposition level and sym8 wavelet, the new threshold function achieved optimal denoising performance which was much better than that of the four common filtering methods.