摘要
加油机人工检定过程中需要人工读取液位、温度等数据,再进行计算和记录,整个过程繁琐且耗时,且受环境温度变化、光线强弱等因素会影响读数的准确性。文章提出一种基于机器视觉的液位图像识别方法。该方法将图像转化为二值线性灰度图,运用水平方向直方图分析技术提取液面位置,再通过液位定位策略确定具体位置,最后借助OCR技术精准读取液位数据。通过现场实验,自动检测的液面数据与目测数据的相对误差控制在0.5%以内,实验结果表明图像识别自动化检测加油机液位方法具有很好的鲁棒性和准确性。
In the manual calibration of gasoline dispensers,operators must manually read liquid level,temperature,and other parameters,followed by calculation and documentation.This process is labor-intensive,time-consuming,and susceptible to environmental factors such as temperature fluctuations and lighting conditions,which can compromise measurement accuracy.This paper proposes an image recognition method based on machine vision for liquid level detection.The approach involves converting the image into a binary grayscale image,followed by horizontal histogram analysis to extract the liquid surface position.A positioning strategy is then applied to determine the exact liquid level,and OCR technology is employed to accurately read the level value.Field experiments demonstrate that the relative error between automatically detected and visually observed liquid level data is kept within 0.5%.The results indicate that the proposed image recognition method offers high accuracy and strong robustness in the automated detection of gasoline dispenser liquid levels.
作者
郑国永
Zheng Guoyong(Sinopec Marketing Zhejiang Company,Hangzhou 310016,China)
出处
《办公自动化》
2025年第16期27-29,共3页
Office Informatization
关键词
加油机
液面检测
图像识别
机器视觉
gasoline dispenser
liquid level detection
image recognition
machine vision
