中文摘要：

指针式仪表的自动读数是近年来的热门研究内容，但一直存在由成像畸变导致的误差较大的问题。本文提出了一种基于计算机视觉的指针式仪表示值精确判读方法，在对指针式仪表图像中的指针与表盘长刻度进行提取并对其倾斜角度进行检测的基础上，依据针孔成像模型，分析了摄像机内部畸变参数对两直线间夹角成像失真的影响，建立了指针与零刻度间夹角、长刻度与零刻度间夹角的失真校正算法，采用牛顿插值法对指针与零刻度间夹角和仪表示值间的关系进行函数逼近，提高了基于计算机视觉的指针式仪表智能判读的准确度。对0．5级伏安表的判读实验证明了方法的有效性与正确性。

英文摘要：

Automatic reading of analog meter has been a research hotspot recently, but the reading error caused by imaging distortion always exists. In this paper, an intelligent precise reading method for analog meter is presented on the basis of computer vision. A method for extracting the positions of the needle and the long scale lines in the image and detecting their included angles is introduced. Based on pin-hole imaging model, the effect of the camera skew factors on the imaging distortion of the included angle of two lines is analyzed. A correction algorithm for the imaging distortion of the included angles of the needle and zero scale, as well zero scale and the long scale is proposed. Newton interpolation method is used for the function approximation of the relationship between the analog meter reading and the included angle of the needle and zero scale, which improves the accuracy of analog meter reading based on computer vision. Experiment results for volt-ampere meter with accuracy of 0.5% prove the validity and correctness of the proposed method.