基于Gabor变换与区域生长的瓷砖表面缺陷视觉检测方法研究  

Research on Visual Detection Method of Tile Surface Defects Based on Gabor Transformation and Region Growing

作　　者：李光亚 邹国锋[1] 傅桂霞[1] LI Guang-ya;ZOU Guo-feng;FU Gui-xia(College of Electrical and Electronic Engineering,Shandong University of Technology,Zibo 255049)

机构地区：[1]山东理工大学电气与电子工程学院

出　　处：《现代计算机》2019年第24期37-42,共6页Modern Computer

基　　金：淄博市张店区校城融合发展项目(No.118228);大学生创新训练计划(No.118060)

摘　　要：针对瓷砖工业生产中表面缺陷检测严重依赖人工的难题和已有视觉检测技术的不足,提出基于Gabor变换与区域生长的瓷砖表面缺陷检测方法。首先采用中值滤波和局部直方图均衡化实现图像去噪和对比度增强,基于Gabor变换函数在变换域进一步对瓷砖表面图像进行处理,较好地消除光照不均和表面反光影响。然后基于自动区域生长算法实现瓷砖表面缺陷区域准确分割。最后采用双向积分投影实现缺陷区域标注。针对瓷砖表面的裂缝、孔洞、麻面、色差缺陷的实验证明所提方法的有效性和可行性。The detection of tile surface defects relies heavily on manual work and the existing visual detection technology has many disadvantages.Proposes a tile surface defect detection method based on Gabor transformation and region growing.First,the median filtering and local his.togram equalization are adopted to realize image denoising and contrast enhancement.Based on Gabor transform function,the tile surface image is further processed in the transform domain to better eliminate the influence of uneven illumination and surface reflection.Then,based on the automatic region growing algorithm,the defect areas of tile surface are accurately segmented.Finally,bidirectional integral projection is used to mark defect areas.The effectiveness and feasibility of the proposed method are proved by experiments on cracks,holes,pockmarks and chromatic aberration on the tile surface.

关 键 词：局部直方图均衡化 GABOR变换 区域生长 种子点选取 瓷砖表面缺陷检测 

分 类 号：TQ174.66[化学工程—陶瓷工业] TP391.41[化学工程—硅酸盐工业]