Electronic Vision Detection System For Raw Silk Quality Based On Machine Vision

A silkworm can spit silk from 1000 to 1500 meters in diameter from fine to smooth to fine.The silkworm gland secretes silk proteins and the process of spinning is easily affected by the external environment,causing various defects.22.2/24.4dtex raw silk commonly used in silk products is composed of 7 to 9 cocoons of cocoon silk,and its fineness is expressed in intervals in the entire textile industry and is unique.Silk fabrics are skin-friendly,and the gloss is noble and elegant.They are commonly used in luxury goods,but the appearance quality indicators such as raw silk uniformity,roundness,and defects have extremely important effects on silk products.In recent years,people have invested a lot of research on how to electronically monitor the appearance quality of raw silk.Among them,much research has been done on raw silk fineness and uniformity,but relatively little research has been done on the flatness of raw silk.Based on the analysis of existing raw silk appearance quality detection methods,this paper proposes an electronic vision detection system for raw silk appearance quality based on machine vision.This system is not only intuitive,but also highly accurate.The research on the uniformity,flatness and defects of raw silk was done.The entire raw silk appearance quality detection system is composed of image acquisition and image processing.The image acquisition part mainly includes two CCD cameras,one LED point light source,two six-times telecentric lenses,two image acquisition cards,and signal generation.And other hardware devices such as computer image workstations;in addition,the image processing part is mainly image processing software.Thesoftware is a program written in C++ based on the OpenCV algorithm library.It mainly processes image blurring,thresholding,and morphology processing and then acquires images target information.The paper first verifies the accuracy,stability and repeatability of the system through experiments after the entire system is completely set up.The results prove that the system can effectively guarantee the scientificity of the appearance quality of raw silk.Secondly,the raw silk is image collected from two orthogonal directions through the image acquisition part to obtain the front light image and the back light image of the raw silk,and then the diameter value of the raw silk in two orthogonal directions is obtained through image processing.Assuming that the cross section of the raw silk is elliptical Based on the two diameter values as the long and short diameters of the elliptical section,the flatness of the raw silk was calculated,and the test results were analyzed to obtain the flatness detection sampling capacity.In addition,the flatness before and after soaking of the raw silk was studied.The experimental results show that The flatness of raw silk before and after soaking is significantly related,and the flatness of raw silk is increased after soaking and rewinding,and the raw silk is more rounded.In addition,the front light CV%value,back light CV%value,and cross-sectional area CV%of raw silk in each length segment are calculated.The correlation between the flatness of raw silk and each uniformity is analyzed.The experimental analysis shows that the flatness of raw silk has a certain correlation with the CV%value of the cross-sectional area of the raw silk,and the correlation with the unidirectional CV%value is not obvious.At the same time,the effects of different silk sheath lengths on the flatness and uniformity of raw silk were studied and analyzed.The results showed that the selection of a suitable silk sheath length in the process of slicing raw silk was important to the roundness and uniformity of raw silk.Through this system,rapid detection of flatness and evenness of raw silk can be achieved,which has great reference value for the electronic detection of the appearance quality of raw silk,and it is also helpful for the research of shaped fibers.Finally,the system divides raw silk defects according to the area method with reference to the 2008 national raw silk quality inspection standards.There are five main categories,namely,small defects,small defects,common defects,large defects,and extra large defects.A defect detection experiment was performed.In addition,this paper proposes a method based on histogram of orientated gradient histogram(Histogram of Oriented Gradient HOG)feature extraction and Support Vector Machine(SVM)recognition and classification to identify and classify ring cracks,spirals,nodes and rough classes of raw silk defects.The experiment proves that the method has a high recognition rate and can realize fast and efficient raw silk appearance quality detection.