基于特征提取的铝铸件X射线探伤图像缺陷评级技术
Defect Rating Technology for Aluminum Casting X-ray Flaw Detection Image Based on Feature Extraction
吴楚澔 1侯明君 1李蒙 2计效园 1房宇 2孙晓龙 1李沁阳 1董淏 1李宁 1周建新1
作者信息
- 1. 华中科技大学材料成形与模具技术全国重点实验室,武汉 430074
- 2. 新江科技(江苏)有限公司,河源 517001
- 折叠
摘要
针对铝铸件X射线探伤图像人工缺陷评级方法存在的评级结果不稳定、量化困难等问题,设计了一个基于特征提取算法的铸件缺陷评级系统.首先,通过分析特征参数与缺陷级别间的关系,选择面积作为气孔、夹杂、缩孔和缩松的评级指标,并对照国标图谱建立缺陷评级量化体系.然后,对X射线图像进行预处理、边缘检测、轮廓提取和量化计算,并依据所建立体系进行评级;最后开发出一个集成上述功能的自动缺陷评级系统.经验证,该系统与专家评片结果存在±1级的误差.该系统能在一定程度上保证评级品质,便于实现铸件的品质监测.
Abstract
Aiming at the problems of unstable rating results and difficulty in quantification existing in artificial defect rating technology for aluminum casting X-ray flaw detection images,a casting defect rating system based on feature extraction algorithm was designed.Firstly,the area was selected as the rating index of pores,inclusions,shrinkage holes and shrinkage by analyzing the relationship between the characteristic parameters and the defect level,and the defect rating quantitative system was established according to the national standard map.Then,the X-ray images were preprocessed,edge-detected,contoured and quantified,and rated according to the established system.Finally,an automatic defect rating system integrating the above functions was developed,which possesses an error of±1 level with the expert evaluation results after verification.In conclusion,the system can ensure the rating quality to a certain extent,which is convenient for the quality monitoring of castings.
关键词
铝铸件/X射线图像/缺陷评级/图像处理/特征提取Key words
Aluminum Castings/X-ray Images/Defect Rating/Image Processing/Feature Extraction引用本文复制引用
基金项目
国家重点研发计划(2020YFB1710100)
国家自然科学基金(51905188)
国家自然科学基金(52275337)
国家自然科学基金(52090042)
出版年
2024
NETL
NSTL
万方数据