Research On Automatic Tracking And Detection Of Grinding Workpiece Based On Machine Vision

With the rapid development of grinding robots,computer vision and other technologies,the intelligentization,automation,and low cost of the processing and manufacturing industry will become a trend in the new era.How to realize the automation and intelligence of the manufacturing process and then achieve the integration of management and control will be an unprecedented challenge to the current development of the company.In this paper,the technologies of position tracking,computer vision detection and so on are studied,which are urgent to be solved in the current grinding and polishing technology.Some effective and rapid solutions are also proposed,mainly including workpiece location tracking,workpiece defect detection,and surface Roughness testing,etc.Their purposes are mainly to realize the integration of automatic grinding and visual detection of workpieces and to fully integrate traditional CNC machine tools,industrial robots and computer vision.To polish the workpiece,a visual feedback system is established on the basis of the control of the end of the six-degree-of-freedom robot arm.And a new technique is developed,which combines the automatic grinding process with the visual detection of the new technology idea.The main research contents are as follows:(1)Introduce the current status of the visual detection of workpieces and analysis the automatic tracking,defect detection and roughness deeply.In addition,some current visual detection methods are elaborated.Lastly,construct experimental environment according to the reality situation;(2)In order to detect the moving position of the workpiece in real time,an improved adaptive scale estimation correlation filter tracking algorithm is proposed.It can return the position of workpiece to us in real time.The method realizes adaptive scale estimation of the target on the basis of original correlated filtering KCF tracking algorithm.And it carries out the estimation of the centroid position of the workpiece in combination with the mobile platform.This ensures a certain tracking efficiency while estimating the scale.Furthermore,the method also adds a centroid correction relocation mechanism to enhance the tracking stability;(3)To process workpiece surface's defects,the Hough method based on fuzzy adaptation is used to extract the tilt angle of the workpiece and the affine transformation.Besides,an adaptive threshold edge detection algorithm is used to obtain the surface defect profile of the workpiece.It can feed back the defect information in real time.Moreover,in order to detect the small flaws on the surface of the workpiece,a local enlarged region defect detection algorithm is proposed to suppress the anisotropic texture of the enlarged image of the metal workpiece and extract the defect's main contour structure;(4)Finally,the traditional contact stylus measurement method may damage the surface of the object easily,and it is difficult to apply to the visual detection of the shortcomings,this paper uses non-contact light-section microscopy combined with the CCD camera,mobile platform and other equipment for the workpiece surface roughness detection.During testing,the standard workpiece with known roughness is detected by the light-cut method.A relation curve between the calculated value and the standard value is established and the parameters of the roughness detection algorithm are calculated.Then,the least squares method is used to fit the curve to reduce the error and obtain the surface roughness value.In summary,in the process of automatic grinding of workpieces,a visual feedback control system is established in combination with industrial robots,mobile platforms,and other equipment.It can achieve automatic integration of workpiece grinding and computer vision detection to improve workpiece quality and detection efficiency.