Research On X-ray Image Defect Detection Algorithm Of Radial Tire

As an important part of motor vehicle,tire is closely related to the safety of People’s Daily life.At present,domestic tire X-ray testing equipment mainly relies on imports,but due to the high price and the difference of defect standards,foreign equipment can hardly be well used in our industrial production,so most enterprises still rely on artificial detection of defects,low efficiency.With the development of machine vision,it has become a trend for machines to replace human eyes.Machine vision technology has great significance for automatic detection of tire defects.In this thesis,the concepts related to image gray distribution curve were first introduced,and the tire cord areas were differentiated and positioned.According to the characteristics of tire tire side area and tire crown area cord on gray distribution curve,the region positioning algorithm was designed.Firstly,a noise smoothing method of gray distribution curve was proposed to smooth the noise on gray distribution curve.Then,threshold was set by the amplitude of concave line segment of gray distribution curve to accurately locate the curtain line area.Then,the tire side cord is extracted,and four kinds of cord extraction algorithms are proposed: gray minimum extraction method of noise smoothing,Gaussian line detection method,independent and combined line segment method and gray minimum extraction method of filter.In this thesis,the results of four kinds of blind cord extraction algorithms are compared,and the extraction results of blind cord,as well as the efficiency and robustness of the algorithm are comprehensively analyzed.Finally,the gray minimum filtering method is used to extract blind cord.Finally,the common defects of tire X-ray images were detected in this paper.Based on the different characteristics of different defects on the gray distribution curve,the algorithm of thin opening,broken cord,impurity and bubble defect detection was designed and implemented.For thin opening of cord lines,based on the extraction of cord lines above and the spacing histogram of cord lines,threshold values can be set in the fetal side area according to different defect standards for detection,and adaptive threshold values can be set in the fetal crown area according to the 3σ criteria for detection.For broken lines,this paper detects based on the change of the distance between one curtain line and its adjacent curtain lines.For impurities,based on the gray value histogram of minimum gray value of concave line segment of gray distribution curve,this paper proposes two adaptive threshold defect detection methods,the 3σ criterion and the maximum distance method,and compares the results of the two detection methods,and finally determines the maximum distance method based on the histogram to detect impurity defects.For bubbles,the maximum spacing method is used to detect them based on the histogram of maximum gray value of convex segment of gray distribution curve.The tire defect detection algorithm proposed in this paper can detect both broken lines and impurity defects in the library of this subject,and the accuracy rate of tire side opening defects is 86.54%,crown opening defects 89.19% and bubble defects 91.67%,which has certain practical value and provides a new solution path for tire defect detection.