| 引用本文: | 马欢欢,李云蓓,张元馨,等.计算机视觉在水处理技术中的应用：量筒液位智能识别[J].灌溉排水学报,2025,44(6):111-117. |
| MA Huanhuan,LI Yunbei,ZHANG Yuanxin,et al.计算机视觉在水处理技术中的应用：量筒液位智能识别[J].灌溉排水学报,2025,44(6):111-117. |
|
| 摘要: |
| 【目的】针对传统污泥比阻测定中的人工判读效率低、误差大及液面震荡问题，开发基于机器视觉的智能监测系统，通过提升精度、实时性与自动化水平，推动水处理检测向智能化方向发展。【方法】构建多模态图像处理与智能监测平台，基于OpenCV对量筒图像实施HSV空间转换、自适应ROI裁剪及形态学梯度强化液面边缘；采用改进型双边滤波抑制气泡与光斑噪声，结合Otsu算法实现动态阈值分割，实现液位精准识别。开发基于QT5框架的智能检测系统，集成多线程视频采集与卡尔曼滤波动态补偿，通过GUI界面实现液位可视化、远程联动及实时校正，提升动态液位监测精度与效率，增强系统鲁棒性。【结果】系统在标准条件下液位测量绝对误差精确到0.1 mL。引入偏离值补偿后，绝对误差精度稳定于0.01 mL，证实系统具备较高的鲁棒性与工程适用性。【结论】本研究建立的智能监测系统突破了传统动态液位测量局限，其模块化架构可拓展至医疗输液监测、工业过程控制等领域，具有显著的工程应用价值。 |
| 关键词: 图像处理；量筒液位；智能识别；OpenCV |
| DOI：10.13522/j.cnki.ggps.2023459 |
| 分类号: |
| 基金项目: |
|
| Computer vision-based detection of liquid levels within cylindrical tanks in wastewater treatment |
|
MA Huanhuan, LI Yunbei, ZHANG Yuanxin, HE Yu, SHI Ningning, FAN Anrui,
ZHOU Linhao, HUANG Biaobing, HONG Shuangxi, ZHANG Xiaoju
|
|
1. School of Environment, Henan Normal University, Xinxiang 453007, China; 2. School of Software, Henan Normal University, Xinxiang 453007, China; 3. College of Computer and Information Engineering, Henan Normal University, Xinxiang 453007, China
|
| Abstract: |
| 【Objective】Traditional methods for measuring specific resistance of sludge in wastewater treatment are limited by low efficiency, high errors and liquid surface oscillation. To address these issues, a computer vision-based intelligent monitoring system was developed to improves accuracy and automation of the measurements and overcome the limitations of conventional methods.【Method】A multi-modal image processing and intelligent monitoring platform was developed using OpenCV. Images of a graduated cylinder were converted to the HSV color images, followed by adaptive Region of Interest (ROI) cropping and morphological gradient operations to enhance the liquid surface edge detection. An improved bilateral filter was applied to remove the noise induced by bubbles and light reflections. The Otsu algorithm was employed for image segmentation, enabling precise recognition of liquid levels. The system was implemented using the Qt5 framework, incorporating multi-threaded video capture and dynamic compensation via a Kalman filter. Through a graphical user interface (GUI), the system provided liquid level visualization, remote linkage and real-time correction. This significantly improves accuracy and efficiency of dynamic liquid level monitoring and enhances system robustness.【Result】Experimental tests showed the absolute error of the system in liquid level measurement was less than 0.1 mL under standard conditions. Incorporating a deviation compensation can reduce the error to 0.01 mL, demonstrating the robustness and potential of the system for engineering application.【Conclusion】The intelligent detection system we developed overcomes the limitations of traditional dynamic liquid level measurement methods. Its modular architecture can be adapted for use in other fields, such as medical infusion monitoring and industrial process control. |
| Key words: image processing; cylinder liquid level; intelligent recognition; OpenCV |
