CAJ | 학술논문

机器视觉中的缺陷检测是个难点。图像的滤波处理通常是为了消除噪声干扰，处理的原则是保留大部分有效信号，抑制少数的异常信号。在机器视觉的缺陷检测时，需要提取图像中的每一个异常信号，忽略大部分正常图像信号，与图像的滤波处理正好相反。在大量实践基础上，本文总结出了一套基于高斯滤波算法的缺陷检测方法，特别适合细微缺陷的检测。该方法已经投入应用，在应用过程中取得了良好的效果。
궤기시각중적결함검측시개난점。도상적려파처리통상시위료소제조성간우，처리적원칙시보류대부분유효신호，억제소수적이상신호。재궤기시각적결함검측시，수요제취도상중적매일개이상신호，홀략대부분정상도상신호，여도상적려파처리정호상반。재대량실천기출상，본문총결출료일투기우고사려파산법적결함검측방법，특별괄합세미결함적검측。해방법이경투입응용，재응용과정중취득료량호적효과。
As known the detection of the defects in the machine vision is difficult. The purpose of filtering the image is to eliminate the noise interference whose principles of treatment include retaining most of effective signals and suppressing a few abnormal signals. However , in the detection of the defects in the machine vision , one needs to extract every single abnormal signal in the image and ignore most of normal image signals , which runs counter to the filtering process. By trial and error, the paper comes up with a set of defect detection methods based on Gaussian filtering algorithm that especially suit minor defect detection. This method, already used, proves very effective.