Detection of Impurity and Bubble Defects in Tire X-Ray Image Based on Improved Extremum Filter and Locally Adaptive-threshold Binaryzation

The manufacturing technology of all steel radial tire is complex. Some defects will inevitably appear in the tire due to the complex production process. Impurity and bubble are two typical kinds of defects in the tire. In this study, a unified detection of impurity and bubble defects in tire X-ray images is proposed by means of improved extremum filter and improved locally adaptive-threshold binaryzation . Firstly, the tire image is divided into cords and background by extremum filter. Secondly, an improved locally adaptive-threshold binaryzation is used to separate defects from the background. Finally, image denoising and marking are processed. In the experiment, we tested the proposed approach by using 280 tires with various types of defects from a tire factory. For the detection of clear and blur impurities and clear bubbles, the precision of our method can reach 97%, and the recall is 95.7%. It is to be noted that the proposed method can only be used to detect those two defects in sidewall and shoulder of the tire which we refer as the carcass.