Textile classification and identification method based on infrared spectrum detection technology

Abstract

The invention provides a textile classification and identification method based on an infrared spectrum detection technology. The textile classification and identification method comprises the following steps: S1, firstly, carrying out moisture absorption on dried samples of different types of textiles; for the same sample, preparing sub-samples with different water contents according to differentmoisture absorption time; s2, collecting near infrared spectrums of all the sample sub-samples; s3, performing two-dimensional correlation spectrum analysis on a series of sub-sample near-infrared spectra of each sample to obtain a synchronous two-dimensional correlation spectrogram and an asynchronous two-dimensional correlation spectrogram, and then fusing the synchronous two-dimensional correlation spectrogram and the asynchronous two-dimensional correlation spectrogram to obtain a mixed two-dimensional correlation spectrogram; and S4, establishing an image classification model by using aGoogLeNet image recognition model based on a deep artificial neural network in combination with a transfer learning method according to the spectral images of the different types of samples in the step S3, thereby realizing classification and recognition of the different types of textile samples. The method can effectively identify and classify different types of textiles with complex components and highly similar compositions.

Description

technical field [0001] The invention relates to the technical field of textile detection and recognition, in particular to a textile classification and recognition method based on infrared spectrum detection technology. Background technique [0002] Different substances have different molecular structures, and their near-infrared spectra are also different. Using the pattern recognition method, according to the difference of the near-infrared spectral characteristics of the sample, the classification and discrimination of the sample can be realized, mainly relying on the difference in the chemical information. The infrared spectrum discriminant analysis process includes steps such as spectral preprocessing, feature extraction and pattern recognition, and classifies and recognizes by comparing the spectral distance between the sample to be tested and known samples of different types. For different types of samples with obvious differences in spectra, the infrared spectrum de...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a textile classification and recognition method based on infrared spectrum detection technology to solve the technical problem that traditional infrared spectrum detection methods cannot effectively identify and classify different types of samples with complex components and highly similar compositions in textiles

Images