计算机科学
多光谱图像
人工智能
计算机视觉
单眼
帧(网络)
图像分辨率
帧速率
作者
Jules Karangwa,Linghua Kong,Tongfei You,Dingrong Yi,Jishi Zheng,Jiang Wei,Mao-Hsiung Hung,Fanru Meng
标识
DOI:10.1109/rcae51546.2020.9294392
摘要
Presently, there exist few 3D imaging methods that can simultaneously acquire both 3-dimensional spatial information and material information of the environment by a single shot and single camera. Moreover, they cannot meet the need for high efficiency with low energy consumption and compact shape at the same time. This paper proposes a multispectral 3D sensing system for single frame monocular equipment that performs synchronal acquisition of both visual image and depth information of the scene. The system uses the snapshot narrow-band multispectral longitudinal dispersion enhanced optical imaging system to obtain multiple spectral band images with different defocusing degrees and carries the depth recovery by comparing and analyzing the defocusing amount of each spectral band and the differential signal formed between spectral bands. This system provides a compact real-time 3D vision system with low energy consumption, high robustness, and strong endurance. It is suitable for unmanned aerial vehicles, aerospace, robots, and other automobiles equipment to navigate in both indoor and outdoor non-structured environments.
科研通智能强力驱动
Strongly Powered by AbleSci AI