基于对称性特征的指针仪表识别

The pointer instrument recognition based on symmetry

指针式仪表在电力行业应用广泛,对指针式仪表读数自动识别是实现电网设备自动化和智能化巡检的基础。通过提取待识别指针图像中的对称性特征,在角度和距离空间中进行投票,得到候选指针对称轴,再结合指针边缘像素点的共线性和指针对称轴线段对应的像素值近似一致性等先验信息,对候选指针进行筛选,并根据指针对称轴的旋转角度计算得到仪表的最终读数。实验结果表明,该方法能够准确定位指针区域并识别仪表示数,且算法耗时很短,能够满足对指针式仪表实时巡检的需求。

The pointer instruments are widely used in the power system industry. The development of an automatic identification algorithm for pointer meter readings is the basis for automatic and intelligent inspection of electric power facilities. This paper extracts the features of the pointer images to identify the symmetry,and vote in the space of angle and distance to obtain the candidate symmetry axes of the pointer. Combining with the priori information of co-linearity of the pixels at the edges of the pointer and the corresponding consistency of the pixel values corresponding to the axis line of the pointer,it filters the candidate pointer,calculates the final reading of the meter based on the rotation angle of the symmetry axis of the pointer. Experimental results show that the method can identify the pointer area and the corresponding reading of the meter accurately in rather short time,satisfy the real-time inspection requirements for pointer meters.