.png){kind=icon}
.png){kind=icon}
Return
A binocular vision-based detection system for coded targets
.png){kind=icon}
PRE
AI
DOI：10.1088/1361-6501/ae59a7.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To address the poor robustness of traditional coded targets under complex backgrounds, this paper proposes a high-capacity array ring-structured coded target, termed ArUco_CACT, which combines the coding advantages of ArUco markers with the geometric stability of ring structures and adopts a hierarchical coding scheme to enhance robustness. To further improve detection performance, a lightweight detection network, YOLO-CAGNet, is developed to achieve accurate recognition in noisy scenes, attaining a detection accuracy of 98.3%. In addition, an adaptive regions of interest extraction and distortion correction mechanism is designed to enable sub-pixel localization and stable decoding under wide viewing angles. Experimental results demonstrate that the proposed system exhibits strong anti-interference capability and real-time performance in various complex environments, achieving measurement accuracies of 0.020 mm (static) and 0.028 mm (dynamic) on a binocular platform. These findings verify its applicability in challenging conditions and provide a feasible solution for efficient coded-target recognition and spatial measurement in multi-vision sensor systems.
Keywords:
ArUco_CACT
YOLO-CAGNet
ring-structured coded target
binocular vision
sub-pixel localization
Journal
IF:
3.4
Papers:
2.6K
Citations:
2.3W