Search

To locate and read the complex pointer meter dial for the images with uneven illumination, blurred dial, and tilted dial, this paper firstly proposes an improved BRISK‐FREAK algorithm for dial position. Then, combined with the Line Segment Detector (LSD) algorithm, an automatic identification method for complex pointer meter is proposed. The dial of a large number of SF6 complex pressure pointer meter images are located and the results reveal that the proposed improved BRISK‐FREAK algorithm has good adaptability under strong interference. The computational speed of the proposed algorithm is 33% and 17% higher than the Scale‐Invariant Feature Transform (SIFT) algorithm and the Binary Robust Invariant Scalable Keypoints (BRISK) algorithm respectively. The positioning success rate of the proposed algorithm is 40%, 64%, and 32% higher than that of the SIFT, Oriented FAST and Rotated BRIEF (ORB) and BRISK algorithms respectively. The reading success rate of the proposed method is 94.5%, which is 19.5%, 39.9% and 14.8% higher than that of the methods based on the ORB, SIFT and BRISK algorithms respectively. It is particularly suitable for application in the actual substations to realize the identification of complex pointer meters. [ABSTRACT FROM AUTHOR]

Abstract To locate and read the complex pointer meter dial for the images with uneven illumination, blurred dial, and tilted dial, this paper firstly proposes an improved BRISK‐FREAK algorithm for dial position. Then, combined with the Line Segment Detector (LSD) algorithm, an automatic identification method for complex pointer meter is proposed. The dial of a large number of SF6 complex pressure pointer meter images are located and the results reveal that the proposed improved BRISK‐FREAK algorithm has good adaptability under strong interference. The computational speed of the proposed algorithm is 33% and 17% higher than the Scale‐Invariant Feature Transform (SIFT) algorithm and the Binary Robust Invariant Scalable Keypoints (BRISK) algorithm respectively. The positioning success rate of the proposed algorithm is 40%, 64%, and 32% higher than that of the SIFT, Oriented FAST and Rotated BRIEF (ORB) and BRISK algorithms respectively. The reading success rate of the proposed method is 94.5%, which is 19.5%, 39.9% and 14.8% higher than that of the methods based on the ORB, SIFT and BRISK algorithms respectively. It is particularly suitable for application in the actual substations to realize the identification of complex pointer meters.