Research And Implementation Of Machine Vision System For Defect Detecting In Dyeing Bobbin

As an essential material in the process of yarn dyeing,the dyeing bobbin must be high-quality.For not only does it affect itself recycling,bust also it makes a big difference to the quality of the dyed yarn.Supposing it enables to detect the bobbin's defects in real-time in the production process,both high-quality output and less trouble to dyeing would be achievable.However,the detection of it is mostly based on humans after production,which has poor efficiency and low accuracy.Instead,the detection based on machine vision has a series of advantages,such as far speed,high accuracy and automation.Therefore,detecting the dyeing bobbin based on machine vision seems a good idea.According to the project requirements,this paper developed a much more efficient system for dyeing bobbin's detection after having studied correlative theories and algorithms.And the main work and innovations are as follows:(1)At present,there are several real-time online detections based on machine vision for dyeing bobbin to detect whether there is any cut-length defect,dent,burr,unsmooth cut edge or less hole defect made during the production process.Based on it,the paper carried on the research and implementation of machine vision system for defect detecting in dyeing bobbin.(2)In order to detect the defects of the dyeing bobbin,the paper studied the image processing include image preprocessing,image segmentation and defect target recognition.In the image segmentation stage,the Otsu algorithm is improved by shortening the traversal range of the threshold from[0,L-1]to[[?_T/2],L-1].Where ?T represents the average of gray image.Also,based on it,a new method is proposed to segment the dyeing bobbin image.While in the defect recognition stage,different algorithms are designed to test different defects according to the actual detection environment.For recognition of the cut length defect,a method based on a base line is designed.For dent detection,a method that measures the distance of the"poles" instead of the whole wall of the dyeing bobbin is proposed.As for recognition of the burr and unsmooth cut edge,a method based on a sliding window strategy is designed.Finally,the recognition for numbers of the hole is solved by the method based on a fast positioning strategy.(3)The paper gives an introduction about the design and implementation of the detection system around hardware and software.The hardware includes an image acquisition module and a control module.The software consists of an image processing module,an alarm feedback module and an interactive interface module.(4)In the end,the functional testing of the detection system based on factory environment is carried out from three aspects that is the test of detection accuracy,the test of database and the test of the whole system.The testing result shows that the system designed can run stably and its accuracy can reach 90%.So it meets the design requirements and completes the expected goal.