Determination Of Yarn Blended Ratio Based On Image Processing

At present, in the textile quality supervision, microscopic projection method and thechemical dissolved method is used to determine the yarn blended ratio, this method istime-consuming, easy to cause the visual fatigue of inspectors, and it can make thetest results mix up with many subjective factors. In the wake of teeming with the newtype fiber, blended product varieties is also increased, this brings great distress todetermine the yarn blended ratio. Face multifarious fiber types, the traditional fiberidentification method has already can’t fully competent, which is caused in the processof textile inspection, identification of the fibers can produce error, not conducive tothe practical production of textiles, makes the fiber in the yarn category for identitywill become a very important and complex work.In recent years, the application rangeof the image processing technology is expanding constantly, and has been deeply useto the textile industry. Digital image analysis techniques can reflect fiber’smorphology intuitively, and it can extract measuring result of fiber in a reasonableamount of time, at the same time it can carry out a large amount of data processing.This paper is based on cotton/Tencel blended yarn as experiment object, andmainly study fiber’s longitudinal characteristics analysis, fiber image preprocessing,fiber eigenvalue extraction, main extract the fiber diameter of CV value, rectangularand smoothness. Research the correlation of three eigenvalues, the study resultsshowed that the diameter of the CV value and smoothness combination of the two ismore advantageous to distinguish between cotton fiber and Tencel fiber. Design theBP neural network, the training sample parameter, use the test sample to test thenetwork’s practical, identify the cotton fiber and Tencel fiber, and work out the yarnblended ratio.