Related Experiment Video
Updated: Aug 16, 2025
07:22
Quantitative Fundus Autofluorescence for the Evaluation of Retinal Diseases
Published on: March 11, 2016
11.5K
Luminosity rectified blind Richardson-Lucy deconvolution for single retinal image restoration
Shuhe Zhang1, Carroll A B Webers1, Tos T J M Berendschot1
1University Eye Clinic Maastricht, Maastricht University Medical Center +, P.O. Box 5800, Maastricht, AZ 6202, the Netherlands.
Computer Methods and Programs in Biomedicine
|December 23, 2022
Summary
This study introduces a new method to improve retinal image quality by simultaneously correcting uneven illumination and blurriness. The luminosity rectified Richardson-Lucy (LRRL) framework enhances diagnostic accuracy for ophthalmologists.
Area of Science:
- Ophthalmology
- Medical Imaging
- Image Processing
Background:
- Retinal images suffer from degradation due to poor illumination, blurriness from optical aberrations, and motion.
- Degraded retinal images hinder accurate diagnosis by ophthalmologists.
Purpose of the Study:
- To propose a novel framework for single retinal image restoration.
- To enhance the quality of retinal images for improved diagnostic effectiveness.
Main Methods:
- Developed a luminosity rectified Richardson-Lucy (LRRL) blind deconvolution framework.
- Established an image formation model using double-pass fundus reflection.
- Utilized a differentiable non-convex cost function for joint illumination correction and blind deconvolution.
- Employed gradient descent with Nesterov-accelerated adaptive momentum estimation for efficient optimization.
Main Results:
- Tested the LRRL framework on 1719 images from three public databases.
- Evaluated image quality using metrics such as definition, sharpness, entropy, and multiscale contrast.
- Demonstrated superior performance compared to existing state-of-the-art retinal image blind deconvolution methods.
Conclusions:
- The LRRL framework effectively corrects illumination issues and enhances retinal image clarity.
- Achieved superior restoration quality and implementation efficiency compared to other methods.
- The developed MATLAB code is publicly available on Github for broader use.
