Weld Defect Detection and Recognition System Based on Static Point Cloud Technology

Abstract

AbstractAt present, welding technology has been widely used in the industrial field, and high-quality welding is very important in the entire industrial process. Therefore, the detection and identification of welding quality are of great significance to the development of the industrial industry. Weld defects are mainly divided into internal defects and external defects. External defects can be seen directly with the naked eye, but internal defects cannot be seen directly. Non-destructive testing is required. Currently, the main non-destructive testing is manual evaluation, which is relatively subjective and limited by factors such as the technical level of technicians. This application is aimed at welding defects in the industrial process, based on the welding image generated by X-ray to detect the weld quality, use the MGLNS-Retinex algorithm and the improved region growth algorithm to detect the weld defect area, and innovatively pass the defect point cloud. For defect classification and identification, the defect point cloud can reduce the amount of data and improve the operation speed while achieving a high classification accuracy.KeywordsWeld defectsDefects detectionDefect identificationPointnet++