摘要
针对汽车线束检测项种类结构繁多、位置角度多变,以及人工检测效率低、漏检率高等问题,提出基于图像的汽车线束结构检测方案。通过匹配数据库汽车线束字符自动切换线束检测算法,结合各种图像检测传统算法和深度学习检测算法,实现不同线束各检测项的普适检测。应用到汽车线束生产检测中,平均误报率为2.38%,漏报率为0,证明了该检测方案能准确高效地完成不同汽车线束的检测。
An image-based automobile harness structure detection scheme is proposed to solve the problems such as various types and structures of automobile harness detection items,variable position angles,and low efficiency and high leakage rate of manual detection.The harness detection algorithm is automatically switched by matching the automotive harness characters in the database.Combining various traditional image detection algorithms and deep learning detection algorithms,universal detection of different harness detection items is achieved.In the actual application to automobile harness production detection,the average false alarm rate is 2.38%,and the missed alarm rate is 0.It proves that the detection scheme can accurately and efficiently complete the detection requirements of different automobile harness.
作者
黄新康
袁嫣红
Huang Xinkang;Yuan Yanhong(College of Mechanical Engineering,Zhejiang University of Technology,Hangzhou,Zhejiang 310018,China;Xinchang Research Institute of Zhejiang University of Technology)
出处
《计算机时代》
2023年第6期129-133,共5页
Computer Era
基金
浙江省科技计划项目(2022C01202)。
关键词
汽车线束
图像检测
传统检测算法
深度学习检测算法
人机界面
automobile wiring harness
image detection
traditional detection algorithm
deep learning detection algorithm
interface