Evaluating a Slam-Based Mobile Mapping System: a Methodological Comparison for 3D Heritage Scene Real-Time Reconstruction

摘要

Surveying techniques have increased their productivity, improving the amount and the quality of Cultural Heritage data for documentation purposes. The reason of this improvement is twofold: on one hand, CH related researches gained more and more attention given the tremendous importance they represent for the whole mankind. On the other, their knowledge and accurate documentation is urgent. Even if there are several acquisition techniques for each scale of representation, there is not still the winning solution for every condition; the consequence is that most of the surveying projects related to large and complex sites integrate multiple sensors and techniques [1]. Towards this direction, producers are introducing in the market sophisticated tools which integrates multiple sensors in an ”all in one solution” that in the literature are defined as combination of methods. One of them, recently introduced, is the SLAM (Simultaneously Localization and Mapping) technology, that has introduced a new system to acquire millions of 3D points in a short time and with good accuracy also in indoor location. In this light, the work that will be described in the article deals with the comparison of dense point clouds obtained with different acquisition techniques (namely TLS and Mobile Mapping), using as a case study a Roman Amphitheater, since, given its complexity, represents a challenging benchmark for such study. For a more detailed documentation of the site and to cope with the lack of imagery information from these latter techniques, a drone flight was performed. The main contributions of the article are: i) testing in a real case scenario a novel tool which integrates different sensors, assessing its suitability for archaeological settings; ii) validating its accuracy w.r.t. other well established techniques; iii) performing the complete mapping of an important archaeological sites, combining terrestrial and aerial acquisition techniques.