计算机视觉
眼底(子宫)
人工智能
直线(几何图形)
像素
计算机科学
分叉
数学
几何学
眼科
医学
物理
量子力学
非线性系统
作者
Yu Meng,Hao Meng,guo-Zhong liu
出处
期刊:2021 7th Annual International Conference on Network and Information Systems for Computers (ICNISC)
日期:2018-04-01
卷期号:: 140-143
被引量:2
标识
DOI:10.1109/icnisc.2018.00035
摘要
In order to obtain accurate 3D information of the retinal vascular structure, this paper presents a method of 3D reconstruction of retinal vascular structure based on binocular stereovision. According to the structural characteristics of blood vessels in the fundus, the blood vessel is simplified into a skeleton of single-pixel width. The bifurcation points and the line segments between these points are extracted and described. The bifurcation points are firstly matched by epipolar geometry and then the line segments between two bifurcation points are used to match the pixel points on the line segments. Finally, the matched bifurcation points and line segments were reconstructed by the parameter matrix of the camera in order to improve the spatial information of the blood vessels in the fundus, and to completely depict the 3D structure of the vascular veins in the fundus. Through the demonstration and experiments, this method can realize 3D reconstruction of the retinal vascular structure and the real-time monitoring of retinal position and posture to be examined, which can be used to correct the B-scan images of the OCT and provide an accurate 3D image for ophthalmology.
科研通智能强力驱动
Strongly Powered by AbleSci AI