Research On The Surface Defect Detection Method Of Chamical Fiber Yarn Package Based On Machine Vision

As an important textile material,chemical fiber yarn package is widely used in the fields of fabric,clothing and architectural interior decoration.The quality of chemical fiber yarn package affects the quality of textile.In the production line of yarn package,the chemical fiber yarn package can not leave the factory until it goes through several processes,such as winding yarn,dropping cylinder,transportation,storage,detection and classification,packaging,etc.,therefore,the surface defect detection of yarn package is an important link for chemical fiber enterprises to carry out quality control.At present,the defects of yarn package are mainly detected by artificial means,which has great influence on it.In addition,the labor intensity is high,and the production efficiency and accuracy are also very low.With the rapid development of machine vision technology,it is possible to realize automatic defect detection of chemical fiber yarn package.In this paper,machine vision technology is applied to the defect detection of chemical fiber yarn package,and the following working is mainly completed:(1)Construction of visual inspection system hardware platform of chemical fiber yarn package.According to the requirements of image acquisition of chemical fiber yarn package,the hardware parts of the visual detection system are constructed,and the selection of industrial camera and lens,light source and lighting mode,structure design of camera fixed bracket and model selection of industrial PC are analyzed and introduced respectively.(2)Image preprocessing of chemical fiber yarn package.After obtaining the image of chemical fiber yarn package,it is necessary to conduct image preprocessing to remove the interference to the image and improve the image quality.By analyzing the characteristics of the image of chemical fiber yarn package,homomorphic filtering algorithm is first used to remove the uneven illumination in the image.The adaptive median filter algorithm is selected to remove the unwanted interference information such as noise on the image.(3)Defect detection algorithm of chemical fiber yarn package.According to the characteristics of less defect samples of all kinds of chemical fiber yarn package in this paper,a defection algorithm based on multi-scale and multi-direction Gabor wavelet filters were proposed.Firstly,Gabor wavelet filter was obtained by fusion of Gabor transform and wavelet transform,and the parameters of Gabor wavelet filter were optimized by semi-peak tangent method.Then,the preprocessed yarn package image was filtered by Gabor wavelet filters at three scales and six directions.The filtered sub-images are processed by the image fusion algorithm to get the final fusion image.Finally,the fused images are segmented and discriminated by threshold segmentation.The experimental results show that the detection rate of this method can reach 98.5% and the algorithm execution time is 0.0916 s,which can meet the actual requirements.(4)Defect classification algorithm of chemical fiber yarn package is presented.According to the characteristics that there are few samples of various defects in the image of chemical fiber yarn package in this paper,a one-to-one SVM multi-classifier with radial basis function as the kernel function is designed to identify and classify various defects of chemical fiber yarn package.GGCM is used in feature extraction to extract texture features of fusion image and combine them with shape feature of binary image to input them into SVM for classification.Experimental results show that the final defect recognition rate of this classification method can reach 94.7%,which has good robustness and can meet the requirements for the system.