Using the IPOL Journal for Online Reproducible Research in Remote Sensing

Miguel Colom (1) , Tristan Dagobert (2, 1) , Carlo de Franchis (1) , Rafael Grompone Von Gioi (1) , Charles Hessel (1) , Jean-Michel Morel (1)

1 CB - CB - Centre Borelli - UMR 9010
2 DGA

Reproducible research is needed to ensure that scientific results in the literature are reliable, unbiased, and verifiable by others. The journal Image Processing On Line (IPOL) publishes reproducible articles since 2010. This means publishing an algorithm by a literary description, a pseudocode, its source code, a series of test examples, an online facility allowing to test the code on this data and other data submitted by the user, and finally an experimental archive. In this work, we discuss how to publish and review reproducible research in the specific discipline of remote sensing. We put a special emphasis on the construction and proper documentation of public datasets. We show case studies of remote sensing articles publicly available in IPOL, which demonstrate the feasibility of reproducible research in this area. The methods and their application are explained, along with details on how the datasets were built and made available for evaluation, comparison, and scoring to eventually help establish a reliable state-of-the-art of the discipline. Finally, we give specific recommendations for authors and editors willing to publish reproducible research in remote sensing.

Mots clés: Algorithm comparison; datasets; remote sensing; reproducible research; image processing on line (IPOL).

Domaines: Sciences de l'information et de la communication; Traitement du signal et de l'image [eess.SP]

Miguel Colom, Tristan Dagobert, Carlo de Franchis, Rafael Grompone Von Gioi, Charles Hessel, et al.. Using the IPOL Journal for Online Reproducible Research in Remote Sensing. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13, pp.6384-6390. ⟨10.1109/JSTARS.2020.3032100⟩. ⟨hal-03135023⟩