Research On Detection Algorithm Of Wheel Hub Grinding And Polishing Area Based On 3D Vision

In the automobile wheel workshop,most of the automobile wheels need to be polished and polished.There is a lot of dust and noise pollution in the processing process,which has caused great harm to the personal safety of workers.At present,relevant domestic manufacturers have begun to gradually use industrial robots to replace manual grinding,but most of the industrial robots currently used in the production field use manual teaching to plan their motion trajectories.This method consumes a lot of manpower and has a low level of automation.In order to improve the work efficiency of automobile wheel grinding,reduce production costs,and protect the personal safety of workers,this thesis designs a set of automobile wheel contour edge extraction system based on 3D vision to assist industrial robots to complete the grinding of automobile wheels.The main research content of this thesis is to select a suitable 3D camera as the acquisition tool according to the workshop environment and the characteristics of the automobile hub,and select the software and hardware to build an experimental platform.For the consideration of real-time production,the algorithm of contour edge extraction based on depth image is firstly studied,and an edge contour extraction method based on Hough circle is proposed according to the characteristics of automobile wheel.Then,due to the limitation of two-dimensional images,point cloud preprocessing is carried out on the three-dimensional point cloud data of automobile wheels,including point cloud filtering and denoising,voxel grid method downsampling,and target point cloud segmentation.Firstly,in order to obtain the point cloud data and depth images of the car wheel hub,this thesis analyzes the current mainstream acquisition methods,and selects a suitable 3D camera as the acquisition tool according to the workshop environment and the characteristics of the car wheel hub,and selects software and hardware to build a experiment platform.Secondly,this thesis studies the contour edge extraction algorithm based on depth image,and proposes an edge contour extraction method based on Hough circle according to the characteristics of the car wheel itself.After experiments,it analyzes the obtained effect and summarizes the method.deficiencies exist.At the same time,this thesis also studies the 3D point cloud solution,and preprocesses the 3D point cloud data of the car wheel,including point cloud filtering and denoising,voxel grid method downsampling,target point cloud segmentation,etc.process.On the target point cloud processed by the point cloud,the target contour edge extraction based on the normal estimation method is carried out,and the problems existing in this method are summarized according to the experimental results.In order to meet the real-time requirements in industrial production,an edge contour extraction algorithm based on Hough circle and normal estimation method is proposed in this thesis,which is used to extract the edge contour of automobile wheel hubs.Finally,experiments are designed according to the method proposed in this thesis,and experiments are carried out on different types of car wheels to verify the accuracy and realtime performance of the method in this thesis.The experiments show that the method in this thesis can obtain complete and clear contour edges,which can meet the precision of industrial production.Compared with the normal estimation method,the time efficiency is increased by15.3%,and the real-time performance is further improved.