摘要
为解决工业制造中齿轮缺陷检测难的问题,提出一种基于改进的YOLOv3网络的缺陷检测方法.首先构建齿轮缺陷图像数据集,包括图像采集与扩充和缺陷标注;其次采用密集连接网络(DenseNet)结构代替原有的网络结构,提高特征提取能力;最后增加网络预测尺度,提高对于小尺寸缺陷的检测能力.利用齿轮缺陷图像对该方法进行验证,发现所提方法的平均精确率均值比YOLOv3网络提高了3.87%,对齿轮缺失部分的精确率提高了5.7%.与YOLOv3网络相比,所提方法在齿轮缺陷检测上有一定的先进性和有效性.
In this study,we propose an improved YOLOv3 network detection method to solve the problem that gear defects are difficult to detect in industrial manufacturing.First,a gear defect image database is constructed by performing various activities,including image acquisition and expansion and defect labeling.Second,the feature extraction ability is improved using the DenseNet network structure instead of the original network structure.Finally,the small-size defect detection ability is improved by increasing the network prediction scale.When compared with the YOLOv3 network,the mean average precision and the missing-part precision of the gear increased by 3.87%and 5.7%,respectively,using the proposed method.This experiment demonstrates that the proposed method exhibits several advantages and that the gear defects can be effectively detected.
作者
张广世
葛广英
朱荣华
孙群
Zhang Guangshi;Ge Guangying;Zhu Ronghua;Sun Qun(College of Physics and Information Engineering,Liaocheng University,Liaocheng,Shandong 252059,China;College of Mechanical and Automotive Engineering,Liaocheng University,Liaocheng,Shandong 252059,China)
出处
《激光与光电子学进展》
CSCD
北大核心
2020年第12期145-153,共9页
Laser & Optoelectronics Progress
基金
中央引导地方科技发展专项资金。
关键词
图像处理
缺陷检测
特征提取
预测尺度
平均精确率
image processing
defect detection
feature extraction
prediction scale
average precision
