基于图像识别的吸量管液位检测方法研究

Method of pipette liquid level detection based on image recognition

在传统的玻璃量器检定过程中,人工检测方法测量效率和准确度都比较低,不能满足计量院对检定精度和效率的要求,文中设计了一套基于图像识别的高精度吸量管自动检定系统。该系统由图像处理和硬件控制两部分组成,选用DSP和Windows系统作为硬、软件平台。对采集的液位图像用OpenCV进行灰度化、滤波、图像分割、液位识别,最终得到凹液面的纵坐标。通过计算凹液面与检定点的像素差来控制蠕动泵的通电时间,从而控制放水速度的快慢,实现闭环PID控制。实验结果表明该方法具有较高的鲁棒性和较快的测量速度,液位测量精度可达到0.1 mm。

In the measuring process of glass container,as the traditional manual detection method has low measuring efficiency and accuracy,and can′ t meet the accuracy and efficiency requirements of the metrology institute,a system of highprecision and fast reaction pipette automatic detection based on image recognition is designed in this paper. The system consists of image processing and hardware control. The digital signal processor(DSP)and Windows system are selected in this system as the hardware and software platform. The grayscale processing,filtering,image segmentation and liquid level extraction are performed for the collected liquid level image by OpenCV. The ordinate of the concave liquid surface in the image is obtained.The power-on time of the peristaltic pump is controlled by the calculated pixel difference between the concave liquid surface and the detected point to realize the speed control of the water discharge and control of the closed-loop PID. The experimental result shows that the method has a high robustness and faster measuring speed,and measurement accuracy of liquid level can reach0.1 mm.