航天器
计算机科学
特征提取
人工智能
计算机视觉
目标检测
自适应光学
会合
霍夫变换
航程(航空)
算法
光学
模式识别(心理学)
物理
工程类
航空航天工程
图像(数学)
作者
Qi Liu,Xiyu Tang,Ju Huo
出处
期刊:Applied Optics
[Optica Publishing Group]
日期:2024-01-19
卷期号:63 (5): 1364-1364
被引量:1
摘要
In order to fulfill the requirements for various operations in space, such as rendezvous, docking, and capturing, there is a pressing need to achieve ultraclose-range spacecraft pose measurement. This paper addresses the challenges of pose measurement under low-light conditions at ultraclose range by introducing a stereovision solution based on target detection and adaptive circle extraction. Initially, an improved target detection algorithm is employed to expedite feature object detection. Subsequently, an adaptive circle extraction algorithm is developed through analysis of camera imaging to surmount challenges related to feature extraction and potential feature loss in the space environment. This approach facilitates swift and accurate measurement of spacecraft at ultraclose range. The results showcase a 66.36% reduction in parameter count for the enhanced target detection algorithm compared with the prevalent YOLOv7_tiny algorithm. Additionally, the adaptive circle extraction algorithm demonstrates an 11.4% increase in cooperative target feature extraction precision compared with existing methods while maintaining requisite detection speed. Simulation experiments indicate that the real-time position measurement error for spacecraft at ultraclose range is less than 0.18 mm, and angle measurement error is less than 0.05°. This presents a viable visual solution for spacecraft pose measurement at ultraclose range in low-light environments.
科研通智能强力驱动
Strongly Powered by AbleSci AI