CAJ | 학술논문

针对基于机器视觉的吹塑产品外形尺寸在线检测系统易受环境的影响而导致检测精度及速度无法保证的问题，提出了一种基于改进形态学的亚像素边缘提取方法。首先在已获得感兴趣区域的基础上，采用形态学梯度算子和Canny算法确定物体边缘点的像素级位置；然后以该位置的邻域像素灰度值作为判断的补充信息，采用高斯拟合方法求得边缘点的亚像素边缘轮廓，进而得到外观尺寸。试验结果表明：该提取方法具有抗噪声能力强、计算速度快、定位精度高等优点，既能够获得连续光滑的亚像素轮廓，还能够很好地保留原图像的信息，且每次检测周期约为0．2s。
침대기우궤기시각적취소산품외형척촌재선검측계통역수배경적영향이도치검측정도급속도무법보증적문제，제출료일충기우개진형태학적아상소변연제취방법。수선재이획득감흥취구역적기출상，채용형태학제도산자화Canny산법학정물체변연점적상소급위치；연후이해위치적린역상소회도치작위판단적보충신식，채용고사의합방법구득변연점적아상소변연륜곽，진이득도외관척촌。시험결과표명：해제취방법구유항조성능력강、계산속도쾌、정위정도고등우점，기능구획득련속광활적아상소륜곽，환능구흔호지보류원도상적신식，차매차검측주기약위0．2s。
Because of the issue that the dimensions of blow molding products obtained from machine vision inspection system are susceptible to environments making it is difficult to guarantee the detection accuracy and detection speed ,a sub‐pixel edge detec‐tion technology based on improved morphological methods is proposed .Firstly ,the pixel‐level position of object’s edge points is determined by the morphological gradient operator and Canny algorithm .Using the pixel gray value as the supplementary infor‐mation for judging in the neighborhood of the position ,the Gaussian fitting method is then adopted to obtain sub‐pixel edge pro‐file of edge points and accordingly to get the size of appearance .The results show that the algorithm has merits of strong anti‐noise capability ,high computing speed ,high precision positioning and so on .It can not only obtain a continuous smooth sub‐pixel contour ,but also preserve original image information ,and each test cycle is about 0 .2 s .