Cotton grade classification method and device based on distance metric learning

Abstract

Cotton grade grading method and device based on distance metric learning, including collecting digital images of cotton samples, extracting image calculation features, and obtaining 21 calculation features of "color", "impurity" and "milling"; using large-boundary nearest neighbors The algorithm is trained on the training data to obtain a Mahalanobis distance metric matrix; and a k-nearest neighbor classifier is trained; for the cotton samples to be classified, digital images are collected and calculated features are extracted and input to the classifier, and the class output by the classifier is That is, cotton grade. The device includes a light box with a camera and two standard D65 light sources directly above it, and a computer control system. The calculation features selected by the present invention for cotton automatic grading have a high degree of coincidence with the perceptual features selected by manual grading, which ensures the correctness of grading, and the use of distance measurement learning further improves the accuracy of system grading, completely No human intervention is required.

Description

technical field [0001] The invention relates to a method and device for automatically classifying cotton appearance grades, in particular to a method and device for classifying cotton grades based on distance measurement learning, and belongs to the fields of machine vision and intelligent electronics. Background technique [0002] In recent years, the rapid development of the domestic cotton textile industry has driven the increasing demand for cotton, and the gap in the domestic cotton market is increasing day by day. However, due to the continuous decline in the domestic cotton planting area, a large amount of cotton needs to be imported every year. Since the inspection of cotton, its appearance grade has been one of the important factors reflecting the quality of cotton and one of the determinants of cotton price, which has an important influence in import and export trade. [0003] The grade of imported cotton is determined by three factors: "color", "impurity" and "gi...

AI Technical Summary

Problems solved by technology

But its use is limited by its inability to test cotton grades from countries other than the United States

In addition, HVI's judgment of cotton appearance and grade does not match well with the results given by manual inspection, which also affects its practicability

Furthermore, the cost of HVI testing is too high, and it has the characteristics of expensive equipment, easy damage, and high maintenance costs.

[0004] Image processing is an increasingly mature technology, but there is no method and device for cotton grade classification using image processing technology

Images