Tire Surface Character Detection System Based On Active Vision

With the continuous advancement of the modernization process,people’s living standards are constantly improving,and cars have become the main model of travel.As an important part of cars,the demand for tires is also increasing.The characters on the side of tires contain a lot of important information,so it is necessary to detect the quality of tire characters.However,traditional manual detection methods have some problems such as low efficiency and accuracy.In order to solve the problem of manual detection,this paper presents a tire surface character detection system based on active vision in order to achieve the goal of cost reduction and efficiency.The main research contents of this paper are as follows:A character detection system for tire surface is built based on the principle of laser triangulation.The system is built by selecting the appropriate field of view size and working distance according to the actual size of the detected object,and then selecting the camera,lens and laser according to the accuracy requirements.In order to obtain all the morphological features of the object surface,it is necessary to design a motion control system.In order to better control tire rotation with PLC as the core,the power,motor and communication modules are designed.After the system is built,the whole system needs to be calibrated.The calibration of the system is mainly divided into three parts.The experimental results show that the re-projection error of the camera calibration is 0.07 pixels,the error of the laser plane calibration is 0.07 mm,and the error of the mobile pose calibration is within 2 μ m,indicating the accuracy of the calibration.The accuracy of laser stripe centerline extraction is the key to three-dimensional reconstruction.Before extracting the centerline,the image needs to be preprocessed according to the characteristics of the laser stripes to obtain the laser stripe image,which is mainly divided into three steps: channel separation,filtering and image segmentation.In order to solve the shortcomings of traditional methods,the gray barycenter method based on normal direction is proposed.The experimental results show that the algorithm is superior to the traditional algorithms such as gray barycenter method and direction template method in terms of extraction effect and accuracy.The average root mean square error is 0.19 pixels,which lays a good foundation for subsequent character detection.Tire characters are detected based on geometric features and related experiments are carried out.After obtaining the tire point cloud data,filtering and voxelization operations are required to achieve the purpose of noise reduction and downsampling.Since the tire characters are some bumps,it is necessary to first judge the convex voxels,then divide the bumps areas according to the connectivity principle,and finally determine whether the tire characters are defective according to the geometric characteristics of the bump areas.According to the principle of experimental analysis,it is concluded that the positive detection rate of the system can reach 98.810%.