The Tire Tread Detection System Based On Machine Vision

With the development of industrial automation,modern production process has put forward higher requirements on detection methods,accuracy and speed.Widely used machine vision inspection systems and technology can not only meet the needs of industrial automation and effectively improve the degree of industrial automation,detection efficiency and accuracy.In the production of rubber tires,tread length needs to be tested online.Most existing detection methods use manually on-line sampling to detect and record detection results,which has low detection efficiency and can not effectively guarantee detection accuracy.In this paper,according to requirements of rubber tire tread production line,combined with machine vision detection technology,a tire tread detection system is designed.This system can realize online detection and recognition of geometric information(length and inclination angle)and characters on tread in the production line,then save the detected results to PC database in real time.This paper mainly completed the following work:(1)design of overall program for tire tread detection system.The detection system was divided into three modules including image acquisition,information processing,information implementation and assigned their tasks,described the key technologies and basic principles of modules.(2)design of geometric detection methods in detection system.Based on machine vision technology,the length detection method based on shape feature was realized and improved by image pre-processing,threshold segmentation,tread area extraction and edge detection;a method based on Hough transform for detecting the inclination angle was designed.According to the detection accuracy and time,the performance of the methods was analyzed.(3)design of character detection methods in detection system.Based on machine vision software and machine learning methods,the system's character detection was realized through image pre-processing,character training,character classifier,character recognition.According to the detection accuracy and time,the influence of different classifiers on character detection was analyzed.(4)According to requirements of the detection system,combined withcommunication protocol of the lower computer and MySQL database,complete functions of the detection system was realized through programming on VS2010 and Halcon.The accuracy,real-time and length detection precision of the system were tested in the testing site.