用于电能表分拣装置的数字图像处理方法研究

Research on the digital image processing method applied to smart electricity meter sorting device

针对智能电能表分拣装置图像识别需求,研究了电表液晶显示数字识别方法。在图像预处理阶段,应用直方图均衡化方法增强灰度图像,使用开操作增强前景后,再用Otsu算法做二值化处理。在液晶显示区域定位阶段,应用Canny算法检测边缘,提出使用了一种双向卷积结合连通域分析的滤波方法缩小目标区域,最后使用Hough变换结合几何形状特征实现精确定位。在数字分割与识别阶段,应用水平、垂直投影法分割数字,使用交叉点连线斜率组合特征弥补了原有数字结构特征的不足,最终应用结构特征法实现了数字准确识别。结果表明:提出的数字图像处理方法可满足电能表分拣装置应用要求。

Aiming at the demand of image recognition of smart electricity meter sorting device,some studies about recognition method for digital number shown on the LCD display of electricity meters were conducted in this paper.At the image preprocessing stage,the histogram equalization method was used to enhance the gray image,the open operation was employed to enhance the foreground,and then Otsu algorithm was adopted to realize binary processing.In the localization stage of LCD region,Canny algorithm was used to detect the edge firstly,then a filtering method of bidirectional convolution combined with connected domain analysis was proposed to reduce the target region.Finally,Hough transform combined with geometric features was used to achieve accurate localization.In the digital number segmentation and recognition stage,firstly,the horizontal and vertical projection methods were applied to segment the number,and the combination feature of intersection line slope was introduced to make up for the deficiency of the original structure feature of digital number,and the structure feature method was implemented to realize the accurate recognition of the digital number in the end.The test results showed that the proposed digital image processing method could meet the application requirements of the smart electricity meter sorting devices.