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OA

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.