鉴于传统的水位读数方法误差大，成本高，需要一种精确、实时、鲁棒的智能水位检测算法来高效读取水位，为此提出了一种基于计算机视觉的水位检测算法以满足实际需求。首先对拍摄到的图像进行预处理和边缘检测以找出水尺位置，并通过仿射变换对水尺进行矫正。通过两种策略在水尺区域找到水尺关键字的位置，即关键字处理。然后对边缘特征进行投影并检测出水面位置。最后根据关键字处理结果和边缘特征计算得到水面高度。大量实验和实地测试的结果表明：所提算法在基于计算机视觉的水位检测、水尺读数等领域具有理论和应用的双重价值。
In view of the problems of high uncertainty and high cost in existing water level meters, this paper presents a set of water level monitoring system based on computer vision to gain high accurate, real-time, robust intelligent water level monitoring. First, preprocessing and edge detection of captured images are carried out to find out the position of the water gauge to be read, and the calibration of the water gauge is carried out by using affine transformation algorithm. Second, the keyword in the water ruler area is positioned and processed by filtering method. Then the edge information is projected to find water surface. Finally, the height of water surface is calculated according to the results of keyword processing and edge information. Experimental verification and field deployment show that the complete water level detection scheme proposed in this paper has both theoretical and practical value in the field of water level detection and water gauge reading based on computer vision.
[1] 姜晓玉,花再军.基于图像处理的水位自动读取[J].电子设计工程, 2011, 19(23):23-25. Jiang X Y, Hua Z J. Automatic reading of water level based on image processing[J]. Electronic Design Engineering, 2011, 19(23):23-25.(in Chinese)
[2] 黄战华,熊浩伦,朱猛,等.嵌入式水尺图像检测系统与判读算法研究[J].光电工程, 2013, 40(4):1-7. Huang Z H, Xiong H L, Zhu M, et al. Research on embedded water gauge image detection system and interpretation algorithm[J]. Optoelectronic Engineering, 2013, 40(4):1-7.(in Chinese)
[3] Lopez-Molina C, De Baets B, Bustince H, et al. Multiscale edge detection based on Gaussian smoothing and edge tracking[J]. Knowledge-Based Systems, 2013, 44:101-111.
[4] 吴福朝,胡占义.线性确定无穷远平面的单应矩阵和摄象机自标定[J].自动化学报, 2002, 28(4):488-496. Wu F C, Hu Z Y. Homography matrix for linear determination of infinite plane and camera self-calibration[J]. Chinese Journal of Automation, 2002, 28(4):488-496.(in Chinese)
[5] Nalwa V S, Binford T O. On detecting edges[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, PAMI-8(6):699-714.
[6] Sobel M E. Asymptotic confidence intervals for indirect effects in structural equation models[J]. Sociological Methodology, 1982, 13:290-312.
[7] Mallat S, Hwang W L. Singularity detection and processing with wavelets[J]. IEEE Transactions on Information Theory, 1992, 38(2):617-643.
[8] Canny J. A computational approach to edge detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986(6):679-698.
[9] Hager G D, Belhumeur P N. Efficient region tracking with parametric models of geometry and illumination[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998, 20(10):1025-1039.
[10] 曾文锋,李树山,王江安.基于仿射变换模型的图像配准中的平移,旋转和缩放[J].红外与激光工程, 2001, 30(1):18-20. Zeng W F, Li S S, Wang J A. Translation, rotation and scaling in image registration based on affine transformation model[J]. Infrared and Laser Engineering, 2001, 30(1):18-20.(in Chinese)
[11] Huang G, Liu Z, Van Der Maaten L, et al. Densely connected convolutional networks[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2017:4700-4708.
[12] 张阳,杜文广.基于计算机视觉的嵌入式水位检控系统设计[J].制造业自动化, 2014, 36(9):121-123 Zhang Y, Du W G. Design of embedded water level inspection and control system based on computer vision[J]. Manufacturing Automation, 2014, 36(9):121-123.(in Chinese)
[13] He K, Sun J, Tang X. Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 33(12):2341-2353.
