Study On X-ray Image Recognition Technology Of Weld Defect

The X-ray photographic inspection technology of weld defects is realized by using the weld film obtained by transmission and the observation and judgment of the reviewer.This method has the disadvantages of complicated operation,high error rate and consuming storage space.With the development of transmission flaw detection imaging equipment and computer technology,it has become a mainstream method to detect weld defects by means of X-ray digital image produced by transmission flaw detection equipment and digital image processing recognition technology.The method based on X-ray digital image and image processing technology has the advantages of high efficiency,simple operation and less manpower consumption.However,due to the difference of detection environment,equipment and operation level of inspectors,the diversification of X-ray digital image of weld is caused,which brings the problems of low recognition rate and poor versatility to image processing.Therefore,how to improve the recognition rate of weld defects has become one of the hotspots of current research.In this paper,according to the development of related technology at home and abroad,Xray image detection and recognition of weld defects are completed through image acquisition,image preprocessing,image segmentation,defect feature extraction,defect quantification and classification.Firstly,XRS-4 ray source and Q345 welding plate were used as experimental equipment,and 91 X-ray images of welds were obtained through 300 mm focal length.The Smax=7 adaptive median filter with better effect is determined by experiments in spatial and frequency domain filtering.Meanwhile,the linear gray transformation enhancement method is determined to make the gray distribution of the image stretch evenly in contrast enhancement.Secondly,the image is segmented and labeled.An adaptive seam detection method based on connected region judgment is proposed for edge detection and segmentation.Compared with traditional edge detection algorithm,this method has good segmentation effect and produces only a few noise points.Then 11 eigenvalues are extracted for the segmented defects,and the eigenvectors are optimized to 3 dimensions successfully by using dimension reduction algorithm.Then,the quantitative method of weld defect using image quality meter and transmission geometry is studied.The simulation results show that the error of image quality meter is mainly between 1% and 3%.Then,the parameters of SVM are optimized by GA algorithm.The collected sample data are trained and identified,and the average recognition rate is 92%.Finally,a set of X-ray image quantitative and qualitative detection and recognition system for weld defects is established with the software,and the expected results are achieved.