A Pointer Meter Recognition Method Based on Image Processing
Xianwei Han, Yimin Zhang, Wei Gao · Proceedings of the 2020 4th International Conference on Electronic Information Technology and Computer Engineering · 2020
Aiming at the problems of large error in manual reading of pointer meter and limited application environment, this paper proposes a pointer-based instrument recognition method based on image processing. Firstly, the acquired instrument image is preprocessed, and the pointer area is extracted by using image processing methods such as edge detection and contour tracking; Secondly, the LSD algorithm is used to extract the straight line segments of the graduated scale, and the center and radius of the dial are determined by fitting the circle at the midpoints of these straight line segments; Thirdly, the central tick mark is used to extract the main tick marks in the tick loop band, and the accuracy of the main tick mark extraction is improved by checking the relationship of the angle difference between adjacent tick marks; Then, the improved Hough transform method is used to detect the pointer line, the KNN algorithm performs character recognition, and the adjacent characters satisfying the preset distance and angle relationship are combined and stored to complete the matching of the main tick marks and values; Finally, a distance method is used to automatically read the pointer meter. The experimental results show that the method can get the meter reading better under different illumination and background conditions, and the error is smaller than the manual reading. The algorithm is stable and reliable.