Abstract:Recent advancements in optical metrology have enabled continuous documentation of dense 3-dimensional (3D) point clouds of construction projects, including cultural heritage preservation projects. These point clouds must then be further processed to generate semantic digital models, which is integral to the lifecycle management of heritage sites. For large scale and continuous digital documentation, processing of dense 3D point clouds is computationally cumbersome, and consequentially requires additional hardware for data management and analysis, increasing the time, cost, and complexity of the project. Fast and reliable solutions for generating the geometric digital models is, hence, eminently desirable. This manuscript presents an original approach to generate reliable semantic digital models of heritage hemispherical domes using only two images. New closed formulations were derived to establish the relationships between a sphere and its projected ellipse onto an image. These formulations were then utilized to create new methods for: (i) selecting the best pair of images from an image network; (ii) detecting ellipses corresponding to projection of spheres in images; (iii) matching of the detected ellipses between images; and (iv) generating the sphere's geometric digital models. The effectiveness of the proposed method was evaluated under both laboratory and real-world datasets. Laboratory experiments revealed that the proposed process using the best pair of images provided results as accurate as that achieved using eight randomly selected images, while improving computation time by a factor of 50. The results of the two real-world datasets showed that the digital model of a hemispherical dome was generated with 6.2mm accuracy, while improving the total computation time of current best practice by a factor of 7. Real-world experimentation also showed that the proposed method can provide metric scale definition for photogrammetric point clouds with 3mm accuracy using spherical targets. The results suggest that the proposed method was successful in automatically generating fast, and accurate geometric digital models of hemispherical domes.

Automatic Recognition and Digital Documentation of Cultural Heritage Hemispherical Domes using Images

Spherical Dome: Design, Digital Twin Manufacturing, and Testing of a Glubam Structure

FAST AND ACCURATE DOCUMENTATION OF ARCHITECTURAL HERITAGE WITH LOW-COST SPHERICAL PANORAMIC PHOTOGRAPHS FROM 360 CAMERAS

Use of Low-Cost Spherical Cameras for the Digitisation of Cultural Heritage Structures into 3D Point Clouds

Modeling and Visualization of Three Dimensional Objects Using Low-Cost Terrestrial Photogrammetry

3D Structure from 2D Dimensional Images Using Structure from Motion Algorithms

Multi-Sensor Image and Range-Based Techniques for the Geometric Documentation and the Photorealistic 3D Modeling of Complex Architectural Monuments

3D DIGITAL AND PHYSICAL RECONSTRUCTION OF A COLLAPSED DOME USING SFM TECHNIQUES FROM HISTORICAL IMAGES

DENSE MULTI-VIEW IMAGE MATCHING FOR DSM GENERATION FROM SATELLITE IMAGES

3D reconstruction of spherical images based on incremental structure from motion

Embedded Shape Matching in Photogrammetry Data for Modeling Making Knowledge

An automatic workflow for orientation of historical images with large radiometric and geometric differences

Modeling and Processing of Smart Point Clouds of Cultural Relics with Complex Geometries

Gaussian Heritage: 3D Digitization of Cultural Heritage with Integrated Object Segmentation

FEW SHOT PHOTOGRAMETRY: A COMPARISON BETWEEN NERF AND MVS-SFM FOR THE DOCUMENTATION OF CULTURAL HERITAGE

Multi-resolution 3D reconstruction of cultural landscape heritage based on cloud computing and hd image data

STATE OF THE ART IN DIGITAL SURFACE MODELLING FROM MULTI-VIEW HIGH-RESOLUTION SATELLITE IMAGES

Image-based metric heritage modeling in the near-infrared spectrum

AN IMAGE-BASED DEEP LEARNING WORKFLOW FOR 3D HERITAGE POINT CLOUD SEMANTIC SEGMENTATION

Modeling Method for Large-Scale Cultural Heritage Sites and Objects Using Real Geometric Data and Real Texture Data