Research On Yarn Evenness Detection Algorithm Based On Image Processing

Yarn evenness irregularity is one of the important indices to measure yarn appearance quality.The traditional methods of detecting yarn evenness irregularity are mainly included artificial visual method,capacitive method and photoelectric method.The detection results of these methods are susceptible to environmental factors and artificial subjective factors.Therefore,using digital image processing technology to overcome the disadvantage the traditional detection method has important significance to improve the level of yarn quality and the stability of textile production process.In this paper,the image processing technology is used to detect the yarn evenness irregularity fast and accurately.The main contents of the study include three aspects: The first is the pretreatment of the yarn image,which includes the filtering pretreatment and the tilted correction of the yarn image;The second is the yarn image segmentation and yarn evenness image extraction;The third is the measurement and analysis of the yarn evenness.During the process of yarn image pretreatment,the mean filter and median filter are compared,and the improved smoothing filter algorithm is proposed.The research shows that the improved smoothing filter can filter the noise and interference information of the yarn image effectively,and avoid the fuzzy distortion of the yarn image,so the edge details and shape information of the yarn evenness is well preserved.In order to get the complete,clear and accurate yarn evenness image,this paper studies and compares four methods of extracting yarn evenness.Study on 21 S,28S,32 S,36S combed cotton yarn,the segmentation accuracy and average processing time of each image of four kinds of algorithms are evaluated as the performance index.Compared with other methods,the segmentation accuracy of the method based on OTSU and morphological filter is relatively poor,and the method based on top-hat transform and FCM has a longer processing time.The method based on saliency algorithm has higher processing speed and accuracy than other methods.At the end of the paper,the irregularity is calculated and analyzed according to the obtained image.Research indicates that the diameter value measured by this method is very close to the theoretical value,and the results of yarn evenness test are in good consistency with those of Uster Classimat 5.The yarn defect detection and the classification results are also in good agreement with the manual visual methods.The method of image processing is used to detect the yarn evenness in this paper.The method overcome the disadvantage the traditional detection method,and improve the detection accuracy.It is of great theoretical significance for improving the yarn quality and the stability of textile production process,and has a good application prospect.