摘要 为解决指针式仪表示数读取中识别精度低和算法读取速度慢的问题，提出一种基于戴明回归和感兴趣区域（region of interest，ROI）细化的指针式仪表读数技术.给出了仪表示数读取的算法流程：首先选择ROI，采用基于颜色通道的剪影法和二值化形态学操作进行图像预处理；接着运用图像帧差法消除指针的抖动；然后利用ROI细化算法对待识别仪表的指针进行细化；再使用戴明回归法拟合出仪表指针所在直线的方程和斜率；最后根据指针斜率利用角度法计算仪表的实时示数.通过3组试验，测试了该方法的可行性和防抖动能力，比较了戴明回归拟合直线与霍夫直线检测拟合直线的检测精度，还比较了ROI细化算法与全局细化算法的计算速度.结果表明该方法检测的平均误差比霍夫直线检测减小了3785%，每张图像的平均计算时间比全局细化算法减少了192717 s，同时具有防抖动能力.
Abstract：To solve the problems of low recognition accuracy and slow reading speed of the algorithm in reading the indicator of pointer instrument, a reading technology of pointer instrument based on Deming regression and region of interest (ROI) thinning was proposed. The algorithm flow of reading the indicator of the instrument was given. ROI was selected, and the image preprocessing was carried out by the color channelbased silhouette method and binary morphology operation. The image frame difference method was used to eliminate pointer jitter, and ROI thinning algorithm was used to refine the pointer of instrument to be identified. Deming regression method was used to fit the equation and slope of the straight line where the pointer of instrument was located. According to the pointer slope, the angle method was used to calculate the realtime reading of the instrument. Through three sets of experiments, the feasibility and antijitter ability of the method were tested, and the detection accuracies of the fitting line by Deming regression and the fitting line by Hough line detection were compared. The calculation speeds of the ROI thinning algorithm and the global thinning algorithm were also compared. The results show that the average error of the proposed method is 37.85% less than that of Hough line detection, and the average calculation time of each image is 192717 s less than that of the global thinning algorithm. At the same time, the proposed method has the ability of antijitter.
DING Y S. Fast image feature region detection based on scale invariant feature transformation[J]. Journal of Jilin University (Science Edition), 2020,58(6):1461-1466.（in Chinese）
［2］
YUAN F. A method of correcting the pointer reading of deflection pointer instrument[C]∥Proceedings of the 2017 Chinese Automation Congress.\[S.l.\]: IEEE, doi：10.1109/CAC.2017.8243764.
［3］
ZHANG T T, ZHANG S M, WANG P. Automatic recognition of instrument dial readings with multiple reference points[C]∥Proceedings of the 2017 Chinese Automation Congress.\[S.l.\]: IEEE, doi：10.1109/CAC.2017.8244118.
［4］
ZHU J T, HUANG W H, CHEN W, et al. An ellipse fitting with PSO for automatic reading recognition of pointer instruments[C]∥Proceedings of the 2nd International Conference on Intelligent Autonomous Systems.\[S.l.\]: IEEE, doi：10.1109/ICoIAS.2019.00014.
LI J X. Binary image target contour recognition algorithm based on deep learning[J]. Journal of Jilin University (Science Edition), 2020,58(5):1189-1194.（in Chinese）
［6］
ZUO L, HE P L, ZHANG C H, et al. A robust approach to reading recognition of pointer meters based on improved maskRCNN[J]. Neurocomputing, 2020,388:90-101.
［7］
CUI Y, LUO W, FAN Q, et al. Design and implementation of pointertype multi meters intelligent recognition device based on ARM platform[J]. Journal of Physics: Conference Series,doi: 10.1088/17426596/960/1/012036.
［8］
LAI H W, KANG Q, PAN L, et al. A novel scale recognition method for pointer meters adapted to different types and shapes[C]∥Proceedings of the 2019 IEEE 15th International Conference on Automation Science and Engineering. Piscataway： IEEE Computer Society, doi：10.1109/COASE.2019.8843107.
DU J, WEI H L, FAN S J, et al. Automatic recognition algorithm of pointer pressure gauge based on HOUGH transform[J]. Machine Tool and Hydraulics, 2020, 48(11):70-75.（in Chinese）
[10]
ZHANG Y Q, DING M L, FU W Y F, et al. Reading recognition of pointer meter based on pattern recognition and dynamic threepoints on a line[C]∥Proceedings of the 9th International Conference on Machine Vision.\[S.l.\]: SPIE, doi: 10.1117/12.2268429.
[11]
TIAN E L, ZHANG H L, HANAFIAH M M. A pointer location algorithm for computer vision based automatic reading recognition of pointer gauges[J]. Open Physics, 2019, 17:86-92.
ZHANG Z F, WANG F Q, TIAN E L, et al. Machinevision based reading recognition for a pointer gauge[J]. Control Engineering of China, 2020, 27(3):581-586. （in Chinese）
MO W X, PEI L Q, HUANG Q D, et al. Development of automatic verification system for high precision pointer instrument based on template matching and table searching[J]. Electrical Measurement and Instrumentation, 2017, 54 (12): 100-105.（in Chinese）
LU J K, WAN Y F, CAO L H. Research on remote reading system of gas meter based on image recognition[J]. Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition), 2018,30(5):627-632.（in Chinese）
[15]
LI P F, WU Q X, XIAO Y, et al. An efficient algorithm for recognition of pointer scale value on dashboard[C]∥Proceedings of the 2017 10th International Congress on Image and Signal Processing,BioMedical Engineering and Informatics.\[S.l.\]:IEEE, doi: 10.1109/CISPBMEI.2017.8301900.
