Retina 3D Perception Reconstruction Algorithm Based on Visual Light Field Image

Lipeng Si

doi:10.1109/ACCESS.2020.3033116

Retina 3D Perception Reconstruction Algorithm Based on Visual Light Field Image

2020-01-01

4

OA

AI

DOI：10.1109/ACCESS.2020.3033116

原文链接

查看原文

摘要

En 中文

As an important part of the eye, the retina is very important to the human body. At present, the diagnosis and treatment of patients with retinal damage or detached blindness in clinical medicine rely on medical imaging technology and ophthalmological equipment. However, the three-dimensional image of the retina constructed by the existing imaging equipment has major defects, which is not conducive to the medical staff's treatment of the lesion, careful observation and precise judgment. Therefore, this article proposes a three-dimensional perceptual reconstruction of the retina based on visual light field images. Computer vision lighting technology is used to create four-dimensional spaces for retinal images, and to enable this, four-dimensional light data is connected to the three-dimensional retinal reconstruction. The resulting retinal image not only contains three-dimensional information, but also has four-dimensional optical shapes, providing physicians with well-thought-out and accurate images for diagnosing and treating diseases. This article first collected a large amount of data through the literature survey method, introduced in detail the related theories and algorithm processes of computer vision, light field imaging technology and 3D reconstruction technology, laying a sufficient theoretical foundation for the 3D perception reconstruction of the retina. Combining the light field imaging technology with 3D reconstruction technology, a 3D reconstruction based on visual light field images is proposed, taking 10 retinal images from the ophthalmology department of the Provincial Hospital as an example, the 3D perception of the retina based on visual light field images is carried out. Finally, a comprehensive comparison and analysis of several algorithms in the process of retina 3D perception reconstruction found that the logarithmic transformation method is better than the histogram equalization method in the enhanced image contrast algorithm; compared with the traditional median filter denoising algorithm, the adaptive weighted median filter denoising method proposed in this article has better effect; in the retinal image registration algorithm, the Powell algorithm handles the details of the image more in place, preserves the edge of the image intact, and the image quality is better; and the inter-layer interpolation algorithm. In this article, the image obtained by the adaptive optimization interpolation algorithm proposed is better than the image obtained by the nearest neighbor interpolation and linear interpolation algorithm.

Keyword:

Light fields

Retina

Imaging

Three-dimensional displays

Image reconstruction

Visualization

Computer vision

Visual light field image

light field imaging technology

retinal image

3D perception reconstruction

Retina 3D Perception Reconstruction Algorithm Based on Visual Light Field Image

摘要

AI总结

期刊

机构

引用论文