Word wheel gas meter error detection method based on YOLO positioning and double template matching

By employing YOLO positioning and dual-template matching on gas meters, the problems of low detection efficiency and misjudgment in gas meter digit error detection are solved, achieving stable and high-precision detection on edge devices and meeting the high standards required for automated verification.

CN122391819APending Publication Date: 2026-07-14HANGZHOU DIANZI UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HANGZHOU DIANZI UNIV
Filing Date
2026-04-20
Publication Date
2026-07-14

AI Technical Summary

Technical Problem

Existing technologies for detecting errors in gas meter dials suffer from problems such as low detection efficiency, susceptibility to errors, insufficient computing power, and difficulty in maintaining stable accuracy under conditions of lighting changes and mechanical vibration interference. In particular, deep learning methods have high computational complexity and are prone to misjudgment due to single-template matching.

Method used

The method of YOLO positioning and dual template matching is adopted. The character wheel region is located on the edge computing device by using a lightweight YOLO model. Combined with dual template matching and fallback confirmation mechanism, character wheel error detection is performed.

Benefits of technology

It achieves stable positioning and high-precision detection of the character wheel area on edge devices with limited resources, significantly improving detection accuracy and real-time performance, overcoming the shortcomings of existing technologies, and meeting the high standards required for automated verification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122391819A_ABST
    Figure CN122391819A_ABST
Patent Text Reader

Abstract

The method comprises the following steps: constructing a data set and training a feature YOLO model; deploying the model to an edge computing device, using the device to locate the scale bar area of the word wheel gas meter, and adaptively extracting the word wheel jump detection area as the anchor point; after the word wheel starts to rotate, images are extracted from the word wheel jump detection area at a preset time interval to construct a first template and a second template; during real-time detection of the word wheel, the first similarity between the current frame image and the first template and the second similarity between the current frame image and the second template are calculated; a back-falling confirmation mechanism is used to determine the maximum second similarity in the current detection period, and according to the size relationship between the maximum second similarity and the direct determination threshold and the backtracking determination threshold, it is determined whether the word wheel has completed a complete rotation; the actual time for the word wheel to complete the selected number of turns of the gas meter detection system is recorded, and the word error of the word wheel gas meter is calculated.
Need to check novelty before this filing date? Find Prior Art