X射线焊缝图像的缺陷检测与识别技术.pdfVIP

本文参考文献引用格式：刘辉。万文，熊震宇．x射线焊缝图像的缺陷检测与识别技术[J]．电焊机，2017，47(04)：89—93 嚣瓣缓撵缝厘缀鳓缺褊检测霸识剔攮恭 刘辉，万文，熊震宇 (南昌航空大学，江西南昌330063) 摘要：轩聍目前无损检测主要采用人工方式存在的主观不一致、检测效率低、操作复杂等问题，设谤了一龚 焊缝缺陷白番检测系统。提出基于Otsu双阈值分割的缺陷区域自动提取、图像的降噪和灰度增强的图像预 处理方法；通过SUSAN算法检测烽缝缺陷目标，并结合形态学孔洞填充算法修正缺陷目标；诗算焊缝缺陷 j餐j禄待灌寥数，并结合所设计的深度为4的二更树分类识别逻辑流程，实现了较好的焊缝缺陷的检测结暴。 关键词：焊缝缺陷；图像处理；特握寥数 中图分类号：TG409 文献标志码：A DOI：10．7512／j．issn．1001—2303．2017．04。19 Defectdetectionand ofX weld ray image recognitiontechnology LIUHui，WANWen，X10NG Zhenyu 330063，China) (NanehangHangkongUniversity，Nanchang atthe inthenondestructive artificial as methods，such Abstract：Aimingproblemsexisting testingusing subjective and setofautomatic forwelddefectswas method efficiencycomplicatedoperation，a inspectionsystem designed．Animagepreprocessing basedonOtsudoublethreshold forautomaticextractionofdefective noisereductionand enhan- segmentation area，image grayscale of with hole cementwas SUSAN fordetectionwelddefect combined filling to proposed；byalgorithm target，and morphology algorithm defect thecalculationofthe characteristicoftheweld combinedwiththe modify targets；through target parameters defects，and design oftwoofthe4 classification achieveabetterdetectionresultsofthewelddefects． depth binary,tree recognitionlogicprocess，to words：weld Key defect；imageprocessing；characteristicparameters U 刖舌