| The workpiece size measurement methods are divided into contact and non-contact in the production of industrial parts.Among them,the contact method may cause damage to the object surface,which measurement efficiency is relatively low.Moreover,the measurement result is sensitive to subjective factors.In order to improve the accuracy and efficiency of it,non-contact measurement methods came into being.The traditional stereo vision measurement,as one of them,is widely used.By using this method,multiple 3D points of the object surface can be obtained at the same time.Thus,the measurement effectiveness can be improved.However,it is sensitive to the scene lighting and dependent on the object surface texture feature heavily.As another typical non-contact measurement technology,structured light measurement technology has the advantages of simplicity,strong anti-interference ability.The most important thing is that,unlike the traditional stereo vision measurement technology,in the structured light vision measurement system,the laser projects laser stripes on the measured object surface actively,so the feature points can be accurately located without being affected by the surface texture of the object.Based on the above reasons,this technology can solve the fundamental problem of feature points matching in the conventional stereo vision measurement.As a result,the high accuracy measurement results can be obtained.With the continuous increasing of industrial production standard,only three-dimensional measurement can no longer satisfy it.How to give consideration to the surface defect detection while workpiece size measurement is an urgent problem to be solved.Therefore,in this paper,a workpiece surface defect detection model is introduced to the structured light stereo vision measurement system.By doing that,the workpiece size accuracy and quality can be meet simultaneously in the industrial production process.Based on the above purposes,the research contents of this paper mainly include:(1)Camera calibration.The model of camera imaging is studied in this paper as well as the principles of Zhang's calibration method.At the same time,a camera calibration method based on dot pattern is introduced.Firstly,the Canny operator is used to extract the edge of the dots,and then the initial center coordinates of them are obtained by the SURF feature extraction algorithm.Finally,in order to further located the initial center coordinates precisely,they are fined at the sub-pixel level.The experiment result show that when the error reaches a stable value,the calibration images of Zhang's calibration method are 24,while the images of this paper are only 9.Moreover,compared with the reprojection error of 0.20760 pixels by Zhang's calibration method,the proposed method is 0.10392 pixels.(2)Extraction of structured-light center stripe.Firstly,by using LBP,HOG and Gabor transformation,three stripe feature images can be obtained.Then,the fused stripe feature image can be determined by integrating the three images.Finally,the centerline of it is extracted by skeleton thinning method.(3)The defect detection of workpiece surface.In this paper,the workpiece surface defect detection models are introduced into structured light stereo vision measurement system,which can provide a pre-screening function before size measurement.It is a further extension and supplement of pure stereo vision measurement.These models are YOLOV3 model,Deeplab V3 model and Mask r-CNN model,respectively.Compared with first two models,the Mask r-CNN model can not only locate defects accurately but also implement subdivision of defects,and it's Io U is 72.35%.(4)3D reconstruction and measurement accuracy analysis.The experiment result show that,compared with traditional visual measurement system,the 3D point cloud of the workpiece surface obtained by structured light stereo visual measurement technology is denser.And the measurement accuracy of it is 0.2709 mm.It should be noted that the workpiece is a 0-level workpiece,and its size is 100mm×35mm. |
PDF Full Text Request