Study And Development Of The Knitting Needle Detection System Based On Machine Vision

Currently,the knitting needle detection mainly relied on the manual detection.The detection speed was relatively slow,about 6/min,and the detected knitting needle had poor consistency,high labor intensity,high business cost.It was difficult to match the requirements of modern industrial production.To improve the detection efficiency and consistency,on the platform of LabVIEW software system,this paper developed a universal knitting needle detection system with NI visual development module and related hardware,which can achieve dimension measurement and contour consistency detection.The detection efficiency is about 60/min,and accuracy rate can reach 99.7%.The system can realize the automatic feeding and sorting of the knitting needle,especially improving the detection efficiency and reliability.The main contents of this paper are as followed:(1)The functional and performance requirements of the camera,lens,light source,and other components in the hardware of the detection system are studied.The hardware models are selected.According to the characteristics of the knitting needles and the functional requirements of the system,the hardware platform of the entire detection system is built,including automatic needle feeding device,needle guiding device,needle positioning device,machine vision module and sorting device.(2)The software part of the detection system was developed.The image processing module of the knitting needle and the function module of the detection system were mainly studied.Among them,the image processing module mainly includes image filtering,binary image,edge detection and morphological processing.By comparing and analyzing various images processing method,finally,this paper selected the image processing method suitable for this paper needle parts.The functional modules of the software are designed to realize the needle size measurement,size consistency detection and sorting.(3)The system is calibrated,and the geometric dimension measurement of the left and right direction of the knitting needle and the front and back direction of the knitting needle are completed,and the result of the measurement is analyzed.(4)The paper introduced the detection principle of the deformation of the knitting needles and made a large number of experiments on contour consistency inspection of the knitting needles.Finally,the experimental results were tested by using the method of the manual projection.In the detection of the left and right deformation,the level camera collected image of the knitting needle that were placed on the glass plate.The paper can detect the value of the gap between the knitting needle and the image and the overall thickness to determine the eligibility of the left and right deformation.The experiment found that the theoretical maximum deformation is the same as the actual maximum deformation.In the detection of the before and after deformation,using the morphological treatment of expansion and erosion operations to make a template,the profile of the needle bar upper edge and the butt right edge in the image of the knitting needle coincide with the profile of the template.Then it is detected whether the contour of the needle hook in the image of the knitting needle to be measured is within the contour of the template.The experimental results show that,when the number of the expanded template pixels and the corrosive template pixels are both 2,the actual maximum deformation in the before and after direction of the knitting needles is 0.030?,and this meets the requirements of the consistency.