中文摘要：

针对仪表的智能识别方法在电力系统的工程应用,利用图像处理的方法对指针式仪表的智能读数进行研究;首先,针对摄像头多视角、多距离的安装问题,利用定向二进制描述符ORB （oriented FAST and rotated BRIEF）算法求解仪表模板图像与待测图像的透视变换矩阵,用于定位指针旋转区域;然后根据表盘灰度特征信息,提出基于圆周区域的累积直方图法（circle-based regional cumu-lative histogram,CRH）对指针进行定位,由指针偏转角度得到读数;实验结果表明,该方法对指针读数识别十分有效,达到了实用化要求,且具有实时性好和精度高的特点.

英文摘要：

For engineering application of the intelligent recognition of the meters in electric system, this paper studied an intelligent read- ing method based on imagf processing for automatic pointer instrument. First, for the setup problem of the camera with multiple viewpoints and scales, a binary descriptor ORB （Oriented FAST and Rotated BRIEF） is utilized for getting the perspective transformation matrix be- tween the template and the test images, and then locate the pointer rotation region. Second, according to the gray character information of the dial, a Circle--based Regional cumulative Histogram （CRH） method is proposed to finely locate the pointer, then the reading is obtained by the pointer deflection angle. The experimental results demonstrate that the method is of high efficiency and up to the engineering require- ment. Meanwhile, the method is real time and accuracy for the pointer reading recognition.