An apple variety identification method based on fuzzy clustering of spectral band optimization

Abstract

The invention discloses an apple variety identification method based on optimal fuzzy clustering of a spectral band. The method comprises the following steps: S1, acquiring Fourier near-infrared spectrums of different varieties of apple samples: for the different varieties of apple samples, detecting the apple samples by using a Fourier near-infrared spectrometer, obtaining Fourier near-infrared spectrum data of the apple samples, and storing the data in a computer; And S2, preprocessing the near infrared spectrum of the apple sample in the step S1 by using a standard normal variable change (SNV). And S3, carrying out waveband optimization on the near infrared spectrum in the step S2 by using BIPLSDA (Binary Interval Least Squares Analysis). S4, carrying out dimension reduction processingand identification information extraction on the apple near-infrared spectrum: compressing the apple near-infrared spectrum data in the step S3 by utilizing principal component analysis (PCA); And then utilizing linear discriminant analysis (LDA) to extract the identification information of the data. And S5, identifying the apple variety of the test sample containing the identification informationin the step S4 by using an improved fuzzy C-means clustering method.

Description

technical field [0001] The invention relates to a method for distinguishing apple varieties, in particular to a method for distinguishing apple varieties by fuzzy clustering with optimal spectrum bands. Background technique [0002] Apples are rich in nutrients, and it is one of the fruits that people often eat. The classification of apples is an important link in the commercialization of apples after harvest. Manual classification is not only time-consuming but also affected by subjective factors; while the traditional physical and chemical analysis of apples needs to detect sugar content and acidity, the experiment is complicated and time-consuming. Therefore, it is very necessary to study a simple, fast and non-destructive method for identifying apple varieties. [0003] Near-infrared spectroscopy technology is based on the self-vibration of the organic functional groups (O-H, C-H, N-H, S-H) inside the sample to absorb the energy of the corresponding wavelength in the n...

AI Technical Summary

Problems solved by technology

[0005] The present invention aims at the shortcoming that the existing fuzzy C-means clustering clustering result is affected by noise data when clustering the apple Fourier near-infrared spectral data containing noise, and when the full spectrum data is processed, the redundant data affects the The improvement plan proposed due to the shortcomings of the accuracy of data processing

The invention adopts the backward interval partial least squares discriminant analysis to optimize the spectral band, which solves the problem that redundant data affects the accuracy of data processing; adopts an improved fuzzy C-means clustering method to perform apple Fourier near-infrared Fuzzy clustering analysis of spectral data to identify apple varieties, solving the problem that the clustering results of fuzzy C-means clustering are affected by noisy data when clustering noisy near-infrared spectral data

Images