An Intelligent Recognition Method For Pointer Meter Indicating Value Based On Scale-lines Feature

The pointer meter is often used to monitor the equipment parameters and running state,which is widely used in many production and life fields.Most pointer meter do not contain a communication interface and the measurement data needs to be obtained manually.In order to improve the reading efficiency of pointer meter and improve the reliability of data reading,intelligent identification methods for pointer meter have been put forward one after another.The existing indicating value recognition methods mainly rely on difference image,area contour centroid method,Hough line detection and other methods to extract the center of the dial and identify the pointer.These methods are easy to be affected by light conditions and camera angle,so they have a great impact on subsequent indicating value recognition.Based on the deep analysis of the pointer meter structure and image features,this paper put forward a method of indicating value recognition based on the scale line features.This paper mainly focuses on the accurate positioning of the center of pointer rotation,the reliable detection of the center line of pointer,the method of digital character extraction and recognition.Firstly,the region growth method is adopted to segment the dial region of the pointer meter,and the pointer and scale-lines are extracted by the eccentricity operator.According to the scale-line centripetalism,Hough transformation is used to fit the center line of pointer and scale lines,and then the common intersection points of these lines are calculated as the rotation center of the pointer.The center of rotation is used as a constraint,further detect the center line of pointer and the direction of pointer.Then,for the two kinds of pointer meters with scale line full circle distribution and partial distribution,the corresponding indicating value recognition methods are proposed.Based on the detected results of the pointer rotation center,the polar coordinate transformation of the pointer meter image is carried out to convert the scale lines with circular distribution into vertical lines.The distance between adjacent scale lines was calculated by using two-dimensional discrete Fourier transform,and the missing scale lines were interpolated.For the pointer meter containing only part of scale lines,the positions of zero scale-line and maximum scale-line are detected by the histogram of scale lines angle,and the indicating value of pointer meter is further identified by the angle between the center line of pointer and the zero scale line.Finally,based on the simulation image and two different types of pointer meterimages,the reading recognition experiment is completed.The results show that the proposed method has good robustness under uneven illumination and oblique angle of view,and can accurately extract the feature of pointer meter to achieve accurate indicating value recognition.