Research Article Open Access

Endoscopy video analysis algorithms and their independence of rotation, brightness, contrast, color and blur

Abstract

The article presents selected image analysis algorithms for endoscopy videos. Mathematical methods that are part of these algorithms are described, and authors’ claims about the characteristics of these algorithms, such as the independence of rotation, brightness, contrast, etc. are mentioned. Using the common test on the real endoscopic image database and a set of image transformations, the validity of these claims was checked and compared between algorithms. Many of the results seem to differ from the declaration of the authors, sometimes even strongly denying them. In addition, some algorithms were found extraordinary sensitive to blurring of the images, which indicates the possibility of using them for the detection of blurry frames, not just diseases.

Jan Cychnerski

To read the full article Download Full Article | Visit Full Article