The process of analyzing images from Minimally Invasive Procedures suffers from some problems related to the conditions of these images. Among these problems, it is possible to notice that the illumination issues are one of the most complicated to solve, as these surgeries are performed through small incisions in the human body. Therefore, the present project presents a model based on a deep neural approach that is able to improve low-light surgical images, making them more optimized for analysis processes performed by human surgeons or assistant robots.
Recommended citation: Carvalho, Caio Jordão; Souza, Antonio Carlos. (2019). “Deep Autoencoder Approach to Enhance Low-Light Images during Minimally Invasive Procedures.”