Design And Implementation Of Weld Defect Detection System Based On X-ray Inspection Images

In recent years,with the rapid development of technology in the industrial field,non-destructive flaw detection technology,as an important means of welding product quality inspection,has been widely used in high-tech fields such as military,vehicle,aerospace,construction,and medical care.The tube ultrasonic flaw detector has developed into a computer for digital real-time imaging.In the field of industrial welding,X-ray inspection technology,as a commonly used non-destructive inspection method,can effectively obtain the film image inside the component,and through the artificial visual evaluation method,the weld defect can be detected and quality evaluated.This manual detection method is time-consuming and laborious,and the accuracy rate is not stable enough.It relies heavily on the professionalism and experience of the detection personnel,and the labor cost is relatively high.Therefore,studying an intelligent defect detection scheme for weld images can effectively assist the inspectors in their work and improve the inspection efficiency,which has important practical significance.The research content of this article is mainly divided into the following aspects.According to the characteristics of the weld image,this article has conducted in-depth investigations and experiments on image preprocessing and neural network technology.In terms of improving the image quality,this paper adopts the method of limiting the contrast adaptive histogram equalization to enhance the image,and the method of median filtering to reduce the noise of the image.In terms of effective area extraction,this paper uses filtering denoising,threshold segmentation,morphological processing,contour extraction,and rectangular box circumscribing methods to remove the interference information on both sides of the imaging.In terms of target detection,this article uses Faster RCNN and YOLO v3,two classic deep learning target detection models,to experiment on the welding image data set,and analyze the experimental results.Finally,a neural network model with accuracy and detection speed meeting the requirements is obtained.Based on the above research content,this article finally realized an X-ray image-based intelligent detection system for weld defects,including image area extraction,image processing,intelligent defect detection,result evaluation and other multi-directional integrated functions to assist researchers in their flaw detection work.Improve the accuracy and efficiency of detection.