摘要
针对磁瓦表面缺陷人工检测时存在效率低、稳定性差的问题,提出一种将小波变换应用于磁瓦表面缺陷检测的视觉检测方法。实验中对随机样品磁瓦和模板正品磁瓦进行相同的处理,对比两者处理过程中采集到的参数来判别实验磁瓦是否存在缺陷。先通过小波去噪处理图像,利用灰度直方图阈值提取图像并二值化,计算对比阈值提取区域匹配相似度;再用Canny边缘检测算法提取出阈值提取区域轮廓边缘,经过形态学操作,计算对比最大轮廓长度,最大轮廓面积和凸凹平均面积;最后通过设定提取特征的合格区间来判别磁瓦是否存在缺陷。实验结果显示,该算法对倒角不合格、偏磨、掉块和起层等缺陷的识别效果较好,可将磁瓦表面存在的显著缺陷准确地检测出来。
Aiming at the problems of low efficiency and poor stability in manual detection of magnetic tile surface defect,we propose a visual detection method which applies the wavelet transform to detecting the defects on magnetic tile surfaces.In experiment,we process the random sample magnetic tiles and the template quality tiles in the same way,and compare their parameters collected from the processes to determine whether the experimental magnetic tiles are defective.First,the method processes the image with wavelet denoising,and uses grey histogram threshold to extract the image and to binarise it,as well as calculates and compares the threshold values for extracting the matching similarity of the segmentation region; then it employs Canny edge detection algorithm to extract the edge contour of the region,through morphological operation,it calculates and compares the maximum length and the maximum area of the contour,as well as the average area of the convex and concave; finally,it determines whether a magnetic tile is defective by setting the qualified range for feature extraction.Experimental results show that the algorithm has good recognition effect on the defects of unqualified chamfering,uneven grinding,pit and layer,and the marked defects on the magnetic tile can be recognised accurately.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第11期210-213,274,共5页
Computer Applications and Software
基金
四川省科技支撑计划项目(2011CGZ0049)
关键词
磁瓦
小波变换
凸凹检测
图像处理
Magnetic tile
Wavelet transform
Concavo-convex detection
Image processing
