Investigation On Automatic Sorting Method Of Curved Edge Needle Based On Machine Vision

Knitting is the core components in knitting processing,which requires high accuracy and is directly related to the yield,quality and cost of knitted products.Knitting is fine and diverse,and it is easy to produce deformation in the left and right direction and the front and rear direction during the processing.Inspection is a necessary link in knitting production.At present,domestic needle-making enterprises are highly dependent on manual detection and manual sorting,which has low detection efficiency,poor detection accuracy and great influence of human factors.Manual detection has become increasingly difficult to adapt to the requirements of high-quality knitting production.Improving the detection methods and means and improving the detection level are important issues to be solved in the knitting industry.In this paper,the author applied machine vision technology to realize the automatic sorting task of curved edge needle by optimizing and adjusting the original hardware and software systems of the research group.The overall design improvement,camera calibration and image processing of the system platform are studied in depth.The prime contents and results of the study are as follows.Firstly,the overall design of the visual sorting system is carried out.Including optical imaging system design and automatic needle splitting mechanism design.By comparing104 slot needles,according to the shape characteristics of curved edge needles and system performance requirements,the illuminating source and lighting mode of the imaging system are analyzed and determined,and the camera and lens models are determined.According to the contour characteristics of curved edge needle,a new automatic needle feeding mechanism is developed.Secondly,through in-depth analysis of camera calibration theory of visual sorting system,parameters calibration and distortion correction are completed.By analyzing the principle of camera calibration,Zhang ' s camera calibration method is used to calibrate this system's cameras.The pixel equivalent is calculated and calibrated by using a rectangular metal block with known physical size,and the radial distortion of the image is corrected by using the internal and external parameters and distortion parameters of camera 1.Camera 2 adapted shot 2 uses a telecentric lens distortion that can be ignored.Then,the processing algorithm of the left and right images of the curved edge needle in the visual sorting system is designed.Aiming at the problem of defocus and blur caused by the shape characteristics of ‘bow' type bending of lace needle,the image sharpening is used after the adjustment of camera depth of field is useless.The nolinear unsharp masking algorithm is introduced for edge sharpening by comparing several sharpening algorithms.The left and right image segmentation and positioning are completed by using threshold segmentation,area screening and minimum enclosing rectangle.The gap value between the left and right lace image with long needle body and the inverted image is measured by segment detection.The gap value is used as a qualified judgment basis to complete the left and right deformation measurement.And through the experiment,it can be seen that the gap value without lens distortion is slightly larger than the gap value after removing distortion.The sharpened image can reduce the numerical jump range,reduce the left and right detection error,and improve the detection stability.Finally,the image processing algorithm before and after the curved edge needle of the visual sorting system is designed.By comparing the filtering effect,median filtering algorithm is used to reduce the damage of noise to image quality.According to the approximate pose range of curved edge needle in the image,the local filtering algorithm based on median filtering is introduced to improve the filtering speed.Removal of burrs and particles in curved edge needles and background parts of images before and after using closed operation of symmetrical structural elements.Through the shape-based template matching algorithm and contour comparison method,the deformation detection of the curved-edge needle is completed.The machine vision detection and sorting experiment of lace needle was carried out.The results show that the consistency of left and right,front and rear directions of curved edge needle after sorting system is good.The detection accuracy of the system can reach above 0.0135 mm,the detection speed can reach 70 pieces per minute.