Research And Development Of An Machine-vision-based Tire Online Classification And Identification System

The potential of Chinese auto market is huge, so good performance and high productivity automated equipment is particularly important, many automotive factories have only one tire transportation workshop,so the equipment that is able to classify different types of tires and deliver it to the corresponding line is essential. Considered the low efficiency and fatigue defects of the artificial classification, we researched and developed a vision-based online identification and classification system based on the technology of machine vision and PLC control.First, the related technology at home and abroad is researched, according to the characteristics of this project, the overall scheme of the online classification system is designed. On the basis of studying the characteristics of the optical imaging,then the front illumination of LED is adopted. According to the accuracy of the inspection and the requirement of the system, the selection of camera and lens are implemented.Secondly, we researched the common image processing such as image preprocessing, segmentation, and compared the effect of different image processing algorithm appling to tires.Then we proposed and implemented image processing methods applicabling to the system.Through researching on machine learning algorithm and its application in the classification,and analyzing the advantages and disadvantages of each method, we put forward a improved algorithm.By creating an improved decision tree structure,we can realize tire classification, and sent the corresponding information to the PLC, delivering it to the corresponding line accordingly.Finally the online identification and classification system for tire based on machine vision is developed. The system has been used in the production line, the application proves the algorithm can meet quite well with the requirement of speed and accuracy, and the system operates stable and reliable.