[1]DENG Dongjin,GE Quanbo,DAI Yuewei.Visual measurement of unmanned ship attitude based on custom elliptical in a low-light environment[J].CAAI Transactions on Intelligent Systems,2025,20(2):486-494.[doi:10.11992/tis.202403018]
Copy
CAAI Transactions on Intelligent Systems[ISSN 1673-4785/CN 23-1538/TP] Volume:
20
Number of periods:
2025 2
Page number:
486-494
Column:
人工智能院长论坛
Public date:
2025-03-05
- Title:
-
Visual measurement of unmanned ship attitude based on custom elliptical in a low-light environment
- Author(s):
-
DENG Dongjin1; GE Quanbo2; 3; 4; DAI Yuewei1; 5
-
1. School of Electronics and Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, China;
2. School of Automation, Nanjing University of Information Science & Technology, Nanjing 210044, China;
3. Jiangsu P
-
- Keywords:
-
threshold segmentation; arc-to-chord ratio; feature mapping; ambiguity; eight-neighborhood tracking; posture vision; elliptical plane method; fast ellipse detection
- CLC:
-
TP30
- DOI:
-
10.11992/tis.202403018
- Abstract:
-
This study proposes an attitude estimation method for unmanned ships using fast ellipse detection to address inaccurate monocular visual attitude measurement caused by low image contrast and increased noise under low-light conditions. First, it employs an adaptive color threshold segmentation algorithm with color and contrast enhancement to optimize edge detection. Second, eight-neighborhood tracking combined with the arc aspect ratio method is designed to eliminate the pseudo-arc caused by noise in low-illumination images. The improved arc feature mapping technology is also used to further distinguish the real elliptical arc segment from the pseudo-arc segment generated by noise, which significantly reduces the computational burden of parameter fitting. Finally, a geometric constraint strategy is established to eradicate the ambiguity of elliptic attitude angle calculation using the parallelism of the elliptic plane normal vector and the rectangular normal vector, which improves the robustness of the algorithm in low-light environments. Experiments indicate that the proposed algorithm offers faster detection speed and higher accuracy for unmanned ship attitude estimation.