Near-infrared spectroscopy noise reduction method for pesticide residue detection

Abstract

The invention discloses a near-infrared spectroscopy noise reduction method for pesticide residue detection. The method comprises the following steps: 1, a near-infrared spectrum of a target is acquired; 2, an EEMD method is used to decompose signals of the near-infrared spectrum, and a plurality of IMF components are obtained; 3, a threshold is adaptively acquired based on the IMF components, andthe IMF components containing noise are selected; 4, an improved L2 regularization method is used to perform noise reduction processing on the selected noise-containing IMF components; and 5, the components after noise reduction and effective information components are reconstructed to acquire noise reduction signals. The problems that the near-infrared spectrum itself contains noise and white noise is introduced by the EEMD are solved, and the signal-to-noise ratio of the near-infrared spectrum and the recognition accuracy of the classification in the pesticide residue detection are improved.

Description

technical field [0001] The invention belongs to the technical field of near-infrared spectrum non-destructive detection, and in particular relates to a near-infrared spectrum noise reduction method for pesticide residue detection. Background technique [0002] Near-infrared light refers to electromagnetic waves with a wavelength in the range of 780-2526nm. It is the first non-visible light region discovered in the absorption spectrum. Its spectral region can reflect the frequency multiplication, Combined frequency vibration absorption, from which the information of organic substances and some inorganic substances can be obtained. In recent years, near-infrared spectroscopy (NIR) analysis technology has developed rapidly due to its advantages of high efficiency, high precision, low cost, and non-destructive analysis, which has attracted widespread attention from the society. In the near-infrared spectrum range, different molecules correspond to the vibration frequencies that...

AI Technical Summary

Problems solved by technology

Traditional filters are only suitable for stable noise processing, but cannot satisfy the elimination of non-stationary noise

Average Empirical Mode Decomposition (EEMD) is an adaptive time-frequency localization analysis method proposed by N.E.Huang in 2009. By adding Gaussian white noise uniformly distributed in the entire time-frequency space, signal regions of different scales are automatically mapped to The appropriate scale related to the background white noise solves the mode aliasing problem of the empirical mode decomposition (EMD), but the original and newly added noise will be decomposed into each intrinsic mode function (IMF), It is difficult to effectively remove the original stained noise in the spectrum and the newly added white noise in the decomposition process by the noise reduction method of selecting several IMFs to reconstruct the spectrum through the threshold method

Images