Computer Vision algorithms performance in architectural heritage multi-image based projects. General overview and operative evaluation: the North Tower of Buñol's Castle (Spain)

[EN] Multi-image based modeling has proven to be effective providing solutions for surveying and documenting cultural heritage, and in particular architectural heritage. In addition to the issues related with instruments and captation strategy, the operativity of these projects is supported by three bases: Computer Vision (C.V.) algorithms, analytical close-range photogrammetry, and theory of errors. In this work we propose an approach that examines the importance of the first, from two points of view. On one hand, we present a brief overview of its intervention in the different processing stages, both in photomodeling as in photograms stitching projects, thus reviewing the fundaments regarding the two classic branches of architectural photogrammetry. On the other, we present a review of the operational strategy with these algorithms, through a case study that evaluates the results of two software applications, advancing some methodological improvements.Cabanes Ginés, JL.; Bonafé, C. (2021). Computer Vision algorithms performance in architectural heritage multi-image based projects. General overview and operative evaluation: the North Tower of Buñol's Castle (Spain). SCIRES-IT. 11(2):125-138. https://doi.org/10.2423/i22394303v11n2p12512513811