Processing And Recognition Of Defects In X-ray Tire Image

With the regard to the security of traffic and transportation, tire inspection has already become an issue cared by every factory which manufactures tires. At present, the tire factories in our Country import all the X-ray tire inspection system they needed. But the system isn't suitable for domestic product line, and the cost is great. So developing our own X-ray tire inspection system with lower cost and good performance is a matter on edge.This subject mainly discusses how to identify the tire defects in the X-ray image automatically with digital image processing and recognition technology. To set forth the subject clearly, this paper makes a general introduction of the structure and operation principle of the X-ray system in the first place. After that, this research tries to use the conventional template matching method to detect the tire defects, but it doesn't work here. With this fail experience, a new way, which is to design special algorithm for each kind of defect in each region of interest, is brought forward. A whole tire image can be divided into three regions of interest consist of bead area, sidewall area and tread area. On the basis of different kind of defects existing in different area, recognition algorithms are developed accordingly.The abnormal array of the body cords and blister in sidewall area are the two kinds of defects mainly studied here. The former includes overlapped cord, open ply splices, wide spacing, crossed cord, broken cord and bending cord, and these defects can be identified by Characteristic parameters, which can be obtained after s series of operation on the original image, including filtering, segmentation, thinning and so on. For the blister, a new approach based on the gray-scale distributing has been proposed. The general position of the blister is found by searching maximal gray-scale points, then the region labeling and region growing method are used to determine blister.This paper adopts the Microsoft Visual C++6.0 as the developing environment, and all the algorithms designed have been implemented in it. A mass of X-ray tire images from the practical production have been tested, and the experiment proves that the algorithms are feasible and effective.