一种陶瓷方形扁平封装外观缺陷检测方法

提出一种基于机器视觉的陶瓷方形扁平封装外观缺陷检测方法。对于封装外形尺寸较大而缺陷较细微的情形,将待检片分为多个区域与标准样片进行比对检测。首先通过Foerstner特征点检测法提取标准片图像的特征点,然后使用随机抽样一致性(RANSAC)图像匹配算法,将所有标准片图像拼接并融合生成一张标准片全幅面模板,再将待检片分区与标准片模板进行序贯比对,以提取可疑区域,最后利用支持向量机(SVM)分类器对可疑区域进行筛选分类。实验结果表明,这种方法不仅克服了传统视觉检测过程中视野范围与图像分辨率相互制约的矛盾,且对陶瓷方形扁平封装表面缺陷具有较高的检出率。

A Defects Detection Method for Ceramic Quad Flat Package Appearance

A machine-vision-based defect detection method for ceramic quad flat package appearance was proposed.For case of larger package size and smaller defects,the sample to be inspected was divided into several regions and compared with the standard sample.Firstly,the feature points of the standard image were extracted by Foerstner feature point detection method.Then all the standard images were stitched and merged by random sample consensus(RANSAC)image matching algorithm to generate a standard full-frame template.And then the sample to be inspected was compared with the standard template sequentially to extract the suspicious area.Finally the suspicious region was filtered and classified by using the support vector machine(SVM)classifier.The test results show that the proposed method can overcome the contradiction between the visual field range and the image resolution in the traditional visual inspection process,and has a high detection rate for the ceramic quad flat package surface defects.