Pointer meter recognition algorithm based on two-dimensional convolution and calculus accumulation

. Aiming at the problem that the outdoor pointer instrument dial is easily obscured by environmental factors, an instrument reading recognition method based on two-dimensional convolution and calculus accumulation is proposed. A weighted multi-feature matching algorithm is proposed to improve the success rate of template matching and accurately obtain the center coordinates of the rotating shaft. The dial image is transformed by polar coordinate transformation and calculus accumulation. Finally, the maximum value is the position of the target pointer, which is automatically read in the two-axis coordinate system. Experiments show that the error compared with the actual values is <0.2%, with good robustness under harsh conditions such as too strong and too dark light intensity and occlusion on the dial.