Evaluación y análisis de algoritmos para el tratamiento de imágenes de campos de luz

Abstract

Fundus photograph is essential for the early detection of pathologies as well as the monitoring of their evolution. Macular edema in particular is one of the most delicate ophthalmic diseases. This paper presents some methodologies for the computational treatment of images captured with a plenoptic camera fitted to a slit lamp. The slit lamp is low cost equipment, available in every ophthalmic consulting room. Strategies for decoding color light fields in 5D matrix are presented. Public access test images are used while others have been obtained during the investigation. From a single decoded picture it is possible to view the scene in different perspectives and to refocus at different depths. The results of applying various algorithms are presented and their performance is discussed in terms of processing time. The multiperspective view and focal stack generated confirm that it is feasible to use the light fields of the ocular fundus as a strategy to add 3D information to retinal images, increasing its diagnostic value. Future work will involve deepening the study of algorithms, its optimization and combination, and the processing of focal stacks to generate depth maps of the ocular fundus.