POTATOES, ARTIFICIAL INTELLIGENCE AND OTHER AMENITIES: PLAYING WITH COLORS ON PANCHROMATIC AERIAL PHOTOGRAPHS

Abstract

Historical photographs, whether taken from the air or from the ground, are usually synonyms
of grayscale or sepia prints. From the very beginning of photography, during the first
half of 19th century, people were amazed by this new media that could record all aspects
of a scene with great detail. Soon though, everybody started wondering why would such
an impressive innovation fail to record colors? A process of trials and errors then started
(including the most successful and pioneer one, involving the use of potato starch, by Lumière
brothers) aiming to add colors to photographs, till the consolidation of new systems
(camera and film) capable to collect photographs directly in color. In the past, before and
during this innovative approach, native black and white photographs were painted in the
effort to give them life. Today, only few methods are available to convert a panchromatic
image into a color one, and they need a number of steps and further development to work
properly. The paper tries to present different methods to colorize native black and white
photographs, based on available automatic or interactive Artificial Intelligence (Machine
Learning or Deep Learning) algorithms, on revised remote sensing procedures and on
visual tricks, aiming at exploring the possible improvement in readability and interpretation
of photographed contests in the usual analytic process of photo-interpretation. At the
same time, colorized historical photographs hold different appeal in the general public
and have the potential to attract and involve non-experts in the archaeological/historical
reconstruction phases.