Research And Design On Automatic Recognition System Of Tire-code

Tire-code is a serial number that printed in the side of the body at vulcanization and molding. It notes various kinds of information of the tire and as an ID card of the tire. The tire enterprises can index tire-code and set up a tire management system. At present, tire-codes are imported into database in a typical method. First, a workers copy out the tire-codes, then another worker input them into computer. In this way, misinput is inevitably. Moreover, this means low efficiency. With the tire production rapidly development, manual input can't meet modern tire enterprise's needs. It's an effective innovation method by using automatic recognition technology that solves the problem of date input commendably.The article analyzes the imaging feature of tire-code deeply and designs a tire-code image collection machine. It can collect image real time. The collected images are suitable for further dealing with. Then the article puts forward a detailed recognition scheme. It includes two steps: image pre-processing and feature extracting and tire-code recognition. The aim of image pre-processing and feature extracting is to get feature vectors from tire-code characters. It concerns characters orientation, image enhancement, characters slope adjustment, single character segmentation, normalization, feature selecting and extracting. At characters recognition, using BP Neural Networks to recognize the characters and improving on BP algorithm to obtain better adaptive to noise, higher recognition rate, fast recognition speed. Finally, based on Visual C++6.0 development environment, using C++ and MFC, with Object Oriented Programming method, developing software of Tire-code Automatic Recognition System. The experiment results this system has a higher recognition rate and favorable anti-noise capability.