Application Study On The Yarn Hairiness Detection Based On Image Processing Technology

In the process of textile processing, hairiness is an important evaluating index reflecting the quality of yarns, the number of hairs per unit length and the hairiness length directly not only affects the appearance, quality and usability of finished products, but also influences the subsequent processing. At present, the realization of yarn hairiness detection in the textile industry mainly adopts artificial visual blackboard method, this method is inefficient, multi-stepped, and low-accuracy, and can't provide quantitatively analysis for hairiness. The yarn hairiness detection method which based on image processing technology can overcome the deficiencies of conventional method to obtain scientific detection results.In this paper, a new hairiness detection method with image processing technology is presented, which based on hairiness physical properties and traditional manual method. The detection process of yarn hairiness is: inputs the obtained image into computer firstly; and does gray level transformation, skew correction, noise reduction, edge sharpening and a series of preprocessing operations for hairiness image; and then strengthens the yarn edge using morphological operations; finally extracts the perimeter, area, shape factor and other characteristic parameters of yarn hairiness from binary image, completes the automatic detection of yarn hairiness.Various frequently-used algorithms of image preprocessing and image segmentation are presented and analyzed in this paper, through comparing the image processing results, we can choose treatment plan which suits yarn hairiness feature extraction, and by mathematical morphological filtering, highlight hairiness edge. Experiments results show that this image processing scheme provide convenience for subsequent hairiness feature extracting.This paper presents a multi-region contour tracking algorithm, this algorithm is adaptive and high-accurate for geometrical characteristics extraction, and overcomes the contour tracking problem which tracking only used for single yarn region of binary image. In this method, one contour tracking and extraction can process more yarn regions, and improve the system efficiency.This paper presents the yarn hairiness detection method which used image processing technology, through this method, we can directly obtain hairiness morphological characteristics, the number of hairiness which has different length and the yarn hairiness index, at the same time, this method is convenient, fast and relatively low cost, alleviate the artificial measurement work strength, and improve the hairiness detection accuracy and efficiency.