基于3D视觉的汽车轮胎胎面花纹缺陷检测系统
Automotive Tire Tread Pattern Defect Detection System Based on 3D Vision

目前，国内对于首胎胎纹的识别验证，普遍还是采用人工的方式，这种方式存在耗时长，精度低，容易遗漏等缺点。为了解决人工识别出现的种种问题，本文设计了基于3D视觉技术汽车轮胎胎面花纹缺陷检测系统，该系统收集到的是轮胎胎面花纹的点云数据，经过处理可得到轮胎胎面花纹的二维深度图，再由训练好的语义分割模型进行检测识别。实验结果表明，该系统能够快速地准确地检测出轮胎胎面花纹存在的鼓泡、划痕和花纹错位等缺陷，具有实效性、高精度性的优点。
At present, the identification and verification of the tread of the first tire in China is generally manual, which has the shortcomings of long time-consuming, low precision, and easy omission. In order to solve the problems of manual recognition, this paper designs a vehicle tire tread pattern defect detection system based on 3D vision technology, which collects the point cloud data of tire tread pattern, and after processing, a two-dimensional depth map of tire tread pattern can be obtained, and then detected and recognized by the trained semantic segmentation model. The experimental results show that the system can quickly and accurately detect the defects such as bubbling, scratches and misalignment of the tire tread pattern, which has the advantages of effectiveness and high precision.