Optimization of Pointer Meter Recognition

In this study, a pointer meter recognition system (PMRS) is designed based on improvement Hough transform algorithm. PMRS offers three advantages in processing captured images: first, adaptive shooting distance, PMRS can adapt to any shooting distance—whether short or long—between the camera and the dial; second, adaptive shooting angle, PMRS can also adapt to any angle between camera and dial; the third, high computational efficiency, the experimental result showed that, PMRS has almost twice the computational efficiency than the general Hough transform algorithms based on multi-resolution. The above three advantages of PMRS are achieved through the following two methods. For the first, adopting the dynamic gain value of the voting mechanism to reduce invalid or redundant calculations in Hough transform which is based on multi resolution mechanism. The combination of variable gain value of the voting mechanism and multi resolution mechanism make PMRS adapt any shooting distance and high computational efficiency. The second, two theorems proposed in this paper, which make PMRS has the advantages of adaptive shooting angle. Theorem 1 proves that the left-right position relationship between any two points in an image has the characteristic of rotational invariance. Theorem 2 proves the consistency between the rotation angle of any vector in an image and the rotation angle of the image. So, Theorem 2 ensures the theoretical feasibility of the improved algorithm, and Theorem 1 provides implementation methods and principles for improved algorithms. Therefore, PMRS can recognize pointer meter without being affected by the angle of image capture.