中文 | English

J. Hefei Univ. Tech. (Nat. Sci.)

合肥工业大学学报(自然科学版)

JOURNAL OF HEFEI UNIVERSITY OF TECHNOLOGY (NATURAL SCIENCE)

Competent Authorities: Ministry of Education of the People's Republic of China

Sponsor by: Hefei University of Technology

ISSN 1000-1646, CN 34-1001/N

Glue defect detection based on region growing and fusion feature SVM

Download Full Text

基于区域生长和融合特征 SVM 的涂胶缺陷检测

Journal Information

J. Hefei Univ. Tech. (Nat. Sci.), Sept.2023, Vol.46 No.9: 1171-1177

DOI: 10.3969/j.issn.1003-5060.2023.09.004

Authors

CHEN Suxin, WAN Shouxiang, LIU Wei

(School of Mechanical Engineering, Hefei University of Technology, Hefei 230009, China)

Abstract and Keywords

Abstract: Aiming at the problems of low accuracy and efficiency of defect detection in the traditional gluing process, this paper proposes a glue defect detection algorithm based on region growing and fusion feature support vector machine (SVM). Firstly, the image is subjected to preprocessing operations. Then, the gluing region is extracted using an improved fast refinement algorithm, and the skeleton feature after removing burrs is used as the initial growth seed. In order to improve the phenomenon of incomplete segmentation, the center pixel-weighted grayscale and the region-adaptive threshold growth criterion are used to segment the complete gluing region. Finally, based on the advantages of the improved edge gradient feature and the regional texture feature, the improved histogram of oriented gradient-multi radius block local binary pattern (IHOG-MBLBP) fusion feature is introduced into the SVM multi-classifier for training to realize the accurate detection of defects in the gluing region. Experiments verify that the designed defect detection algorithm can accurately extract the skeleton and segment the complete gluing region, and has high accuracy and efficiency in glue defect detection, which can meet the needs of industrial production.

Keywords: glue; skeleton extraction; region growing method; fusion feature; support vector machine (SVM); defect detection

作者信息

陈甦欣，万寿祥，刘伟

（合肥工业大学机械工程学院，安徽合肥230009）

摘要和关键词

摘要: 针对传统的涂胶工艺缺陷检测精度低、效率低的问题, 文章提出一种基于区域生长和融合特征支持向量机(support vector machine, SVM)的涂胶缺陷检测算法。首先对图像进行预处理操作; 然后通过改进的快速细化算法提取涂胶区域, 将去除毛刺后骨架特征作为起始生长种子, 为改善分割不完全现象, 采用中心像素加权灰度和区域自适应阈值生长准则分割出完整的涂胶区域; 最后结合改进边缘梯度特征和区域纹理特征的优点, 将改进的梯度方向直方图-多半径局部二值模式(improved histogram of oriented gradient-multi radius block local binary pattern, IHOG-MBLBP)融合特征送入SVM多分类器进行训练, 实现对涂胶区域缺陷的精确检测。经过实验验证, 所设计的缺陷检测算法能够精确地提取骨架并分割出完整的涂胶区域, 对于涂胶缺陷具有较高的检测精度和效率, 能够满足工业生产需求。

关键词: 涂胶；骨架提取；区域生长法；融合特征；支持向量机（SVM）；缺陷检测

Funding

国家产业技术基础公共服务平台资助项目（20190089921）

Previous Next

Back to Contents