Research On Detection Method Of Quality Defect Of Glue Based On 3D Vision

In the context of Made in China 2025 and Industry 4.0,with the development of technologies such as intelligent manufacturing,high-end manufacturing,and high-precision processing,many industries have put forward higher requirements for non-contact defect detection methods.As a key procedure in the manufacturing industry,the quality of the glue directly determines the performance,quality and appearance of the product,and may also affect the service life of production equipment and even the user’s personal safety.Traditional manual detection methods and visual detection methods based on two-dimensional images have certain limitations.Because of the characteristics of fast speed,high precision,good stability and strong anti-interference ability,the three-dimensional visual inspection method based on line laser has become a technical trend..Based on the 3D visual inspection method of line laser,this paper has carried out in-depth research on the quality defect detection problem of glue.Aiming at the problem of obtaining high-quality 3D point cloud data,this paper builds a 3D scanning system with line laser sensor as the core,and designs the hardware and software structure.The possible causes of noise in the point cloud data are analyzed in detail.Aiming at the problem of low processing efficiency of scattered point clouds,a point cloud topology is established based on KD tree and Octree.The point cloud neighborhood search method is studied,which lays the foundation for related algorithms such as point cloud differential information estimation and point cloud feature description.Aiming at the noise problem in point cloud data,this paper proposes a point cloud filtering algorithm based on combination.The algorithm first divides point cloud noise into large-scale noise and small-range noise,and then use statistical filtering algorithm,radius filtering algorithm and bilateral filtering algorithm in turn to achieve efficient removal of point cloud noise.Aiming at the problem of large amount of point cloud data,this paper proposes a voxel downsampling algorithm that combines space partition and normal.The algorithm uses the constraint relationship between voxel and normal to downsample the point cloud.In order to further reduce the amount of point cloud data,this paper removes the planar background point cloud data in the point cloud based on RANSAC algorithm.Experimental results show that the algorithm of this paper can greatly reduce the number of point clouds while maintaining the surface features of the point clouds as much as possible.Finally,on the basis of point cloud preprocessing and improved point cloud registration algorithm,this paper deeply studies the method of glue quality defect detection based on point pair deviation and the method of glue quality defect detection based on curve fitting.In this paper,the detection of convex glue,concave glue and broken glue is realized.The results of comparison experiments further verify the feasibility and effectiveness of algorithms in this paper.