Warp-knitted Fabric Defect Detection Based On Machine Vision

Along with the rapid development of the social economy technology and the computer electronic science,the traditional industrial manufacturing industry has also ushered in its fourth industrial revolution—— "the Industry 4".China has also made a timely manner,"Made in China 2025" plans to join the ranks of industrial transformation and upgrading.As a traditional manufacturing industry in our country,the textile industry has no doubt that it has entered the industrial transformation.The production which has high quality and high performance products make an enterprise have chips in the fierce competition in the industry.Monitoring the quality of the cloth production is an important means to ensure the quality of the cloth,and using the new technology of fabric quality supervision and inspection is a direction of our industrial reform.At present,there have been some online defect detection equipment has been running in the market,but these devices detected by the cloth is mainly for non-woven cloth,twill and plain cloth and other relatively simple,such as the warp-knitted fabric defect detection still exist some problems.In this paper,it is mainly to solve the problem of the warp-knitted fabric defect detection.For the warp-knitted fabric defect detection,we can divide it into three categories: the surface fabric flaw detection,the bottom fabric flaw detection,and the weft fabric flaw detection.In this paper,the method of fabric defect detection is proposed,and the original image processing method is further optimized and improved: Considering that the image to be detected is rectangular,the perspective transform is applied to image rectification.We choose the width of 50 pixels of the sliding window on the image of each row of data for the low HAT transform processing,to determine the defects of each line,and then determine the veil defects.We first use the integral image data into a one-dimensional form,and then select the width of 30 pixels of the sliding window of the median filter data processing,by calculating the data before and after filtering to determine the situation of the bottom yarn defects.Through the observation of the weft fabric image,we find that the distance between the weft yarn in a certain range,can be approximated as periodic,weft relatively bright,can be found by calculating the extremum of weft position,at the same time in order to further accelerate the rate of the algorithm,reduce the amount of calculation,introduces the method of image segmentation will be filtered out weft data at the same time,the calculated weft position,using the flaw of single period,double weft attribute judgment cycle.The whole algorithm has high speed,accuracy,and stability,anti noise and other aspects have outstanding performance,in the practical application of industrial production,has been well verified,which can meet the needs of industrial production.