成都大学学报（自然科学版）2026，Vol.45Issue(1)：66-74,9.DOI:10.3969/j.issn.1004-5422.2026.01.010
基于双线性插值和归一化相似度匹配的冲压件缺陷检测方法研究
Research on Defect Detection Method for Stamped Parts Based on Bilinear Interpolation and Normalized Similarity Matching
摘要
Abstract
In response to the problems of unstable detection accuracy and high time consumption in tradi-tional quality inspection methods for stamped parts,this study proposed a surface defect detection method for automotive stamped parts based on bilinear interpolation and normalized similarity matching.Firstly,according to the structural characteristics of automotive stamped parts,the collected original images were pre-processed through filtering,and the bilinear interpolation algorithm was employed to resize the ima-ges,optimizing their size parameters.Secondly,image edge detection and affine transformation techniques were applied to correct the tilt of the images of the stamped parts.Finally,the mean-removed normalized similarity matching algorithm was utilized to perform matching calculations on the collected the images of stamped parts.To validate the effectiveness and practicality of the proposed algorithm,a simulation envi-ronment was constructed by using MATLAB software.And experiments including image tilt correction and grayscale-difference-resistant matching were designed.The experimental results demonstrated that,the proposed method and the traditional normalized correlation matching algorithm based on grayscale values both achieved a rate of over 94％in detection accuracy,acceptable product matching rate,and defective product matching rate.Additionally,the matching time for a single sample was reduced to 3.53 seconds,and the detection time was reduced to only 30％of that of traditional normalized correlation matching al-gorithm based on grayscale values,significantly improving the defect detection efficiency.关键词
汽车冲压件/双线性插值/归一化相似度/缺陷检测Key words
automotive stamped parts/bilinear interpolation/normalized similarity/defect detection分类
交通工程引用本文复制引用
滕浩,吴昊荣,吕秋荣,孙付春,罗瑶,刘唯..基于双线性插值和归一化相似度匹配的冲压件缺陷检测方法研究[J].成都大学学报（自然科学版）,2026,45(1):66-74,9.基金项目
四川省大学生创新训练计划项目(S202411079012X) （S202411079012X）
