基于图像识别的非入侵式液位测量方法

Method of Non-contact Liquid Level Detection Based on Image Recognition

针对现有的液位检测系统存在不能同时满足非入侵式测量的高适应性、高准确性和高稳定性等问题,本文采用图像传感器,融合高精度标定技术及图像边缘处理技术提出了非入侵式液面高度测量方法,组建了由图像采集及高度测量两部分组成的液位检测机制。首先构建高精、高稳定的图像传感器与微处理器接口;然后通过软件驱动,实现基于图像传感器的实时图像采集系统;最后通过进行高精度标定,以满足实际运用。实验结果表明,采用CMOS图像传感器获取图像,通过差分边缘检测算子检测边缘特征点的方法,经过标定后,整个系统测量总精度达到±0.3 mm,能够满足在石油化工领域液面高度测量精度及测量范围的要求。

Aiming at the problem of current liquid level detection which cannot meet the requirement of a non-contact measuring method with good adaptability, high accuracy and reliability, this paper proposes a liquid level measurement based on image edge detection by using COMS image sensor. This paper synthesizes the technology of precise calibration and image edge detection in a non-contact liquid level measurement, and constructs liquid level detection composed of image acquisition and the level detection. Firstly, the interfaces of CMOS image senor and S3C2440 processor are built; secondly, images by CMOS image sensor after designing driver software are acquired; thirdly, the system to supply the practical manipulation is calibrated. The experimental results indicate that after using CMOS image sensor to acquire images, detecting features of image edge by difference edge operator, and calibrating, this system can achieve the precision of ±0.3mm. It can meet the demands of detecting liquid level in the areas of chemical and petroleum industry.