计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (15): 191-198.DOI: 10.3778/j.issn.1002-8331.1907-0130

• 图形图像处理 • 上一篇    下一篇

多特征融合的瓷砖表面缺陷检测算法研究

李军华,权小霞,汪宇玲   

  1. 1.南昌航空大学 江西省图像处理与模式识别重点实验室,南昌 330063
    2.东华理工大学 江西省放射性地学大数据技术工程实验室,南昌 330013
  • 出版日期:2020-08-01 发布日期:2020-07-30

Research on Defect Detection Algorithm of Ceramic Tile Surface with Multi-feature Fusion

LI Junhua, QUAN Xiaoxia, WANG Yuling   

  1. 1.College of Jiangxi Key Laboratory on Image Processing and Pattern Recognition, Nanchang Hangkong University, Nanchang 330063, China
    2.College of Jiangxi Engineering Laboratory on Radioactive Geoscience and Big Data Technology, East China University of Technology, Nanchang 330013, China
  • Online:2020-08-01 Published:2020-07-30

摘要:

鉴于单一特征在瓷砖种类较多的情况下,存在对瓷砖表面缺陷内容表达不明显,导致复杂瓷砖识别率较低。针对这个问题,在词袋模型(BoF)框架的基础上,提出一种有效的多特征融合算法用于瓷砖缺陷检测。该算法采用改进后的SIFT和颜色矩融合特征作为瓷砖图像的区域特征描述;根据每种特征对瓷砖被分类的准确率大小,给提取到的两种区域特征分配各自的权重系数实现特征的加权融合;形成综合特征向量送入SVM分类器达到瓷砖缺陷分类的目的。通过不同类型的瓷砖样本进行实验表明,该算法识别率高,对复杂瓷砖能实现较好的分类。

关键词: 瓷砖, 特征提取, 特征融合, 缺陷分类, 词袋模型(BoF)

Abstract:

In view of the single feature in the case of more kinds of ceramic tiles, there is no obvious expression of the surface defects of ceramic tiles, which leads to the low recognition rate of complex ceramic tiles. To address this issue, this paper proposes an effective multi-feature fusion algorithm for tile defect detection, which based on the framework of the Bag of Features(BoF). The proposed algorithm adopts the improved SIFT and color moment fusion features as the regional feature description of the ceramic tile image. According to the accuracy of the each feature classifies tiles, each of the extracted two regional features is assigned a respective weight coefficient to achieve weighted fusion of features. A comprehensive feature vector are put into SVM to achieve the purpose of classifying tile defects. The experiments on different types of tile samples demonstrate that the proposed algorithm has high recognition and can realize good classification of ceramic tiles.

Key words: ceramic tile, feature extraction, feature fusion, defect classification, Bag of Features(BoF)