To solve problems such as low reading precision,poor robustness,and lack of real-time automatic reading of point-er meters,this paper puts forward a method of identification and reading of pointer meters based on a lightweight model.First,the improved YOLOv5 is used to identify and locate the position of the instrument to be detected in the complex scene,and the instru-ment area is extracted.Then,the instrument image is corrected by feature point matching and perspective transformation.Finally,the M-LSD algorithm is used to extract the pointer and calculate the scale reading of the pointer by straight line fitting.The experi-mental results show that the maximum error of the proposed method is 0.05mA,and the proposed method has good robustness,ac-curacy,and real-time performance within the minimum scale range,and can meet the practical application in different complex scenarios.
关键词
指针式仪表/YOLOv5/识别定位/图像校正/直线拟合/仪表读数
Key words
Pointer instrument/YOLOv5/Identification and positioning/Image correction/Line fitting/Instrument reading