A binocular camera calibration method and device
By acquiring and fusing binocular image features and using a pose regressor to predict the absolute pose of the camera, the problem of automated calibration of fixed large baseline binocular cameras is solved, achieving high-precision and low-cost calibration results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU POWER SUPPLY BUREAU GUANGDONG POWER GRID CO LTD
- Filing Date
- 2026-03-23
- Publication Date
- 2026-07-17
AI Technical Summary
There are challenges in performing high-precision, automated calibration of fixed, large-baseline binocular cameras in unattended environments. Existing technologies are costly and have low accuracy, and deep learning methods are limited in application in power scenarios with limited data.
The system acquires pairs of binocular images, performs feature extraction processing, determines the matching degree and geometric position information, fuses them into a high-dimensional feature vector, predicts the absolute pose of the camera through a pose regressor, calculates the relative pose information, and achieves automated calibration.
It enables high-precision automated calibration of large-baseline binocular cameras under unattended conditions, reducing labor costs, improving calibration efficiency and accuracy, and enhancing robustness in challenging scenarios.
Smart Images
Figure CN122415744A_ABST