Understanding the internal structure of museum objects is crucial for determining their origin, condition, and composition. Computed tomography (CT) scanning, a technique that computationally reconstructs a 3D image from multiple radiographs, offers a powerful means to achieve this. However, the cost and specialized nature of CT equipment, coupled with the challenges of transporting valuable artifacts, limit its accessibility within the museum setting. Most museums rely on 2D X-ray imaging, which lacks the depth information provided by CT. This paper addresses this limitation by proposing a method to generate 3D CT images using readily available 2D X-ray equipment and advanced image processing algorithms. This approach reduces the cost barrier and logistical hurdles associated with traditional CT scanning, thereby democratizing access to this crucial non-destructive analysis technique across a wide range of museums and cultural heritage institutions. The potential impact extends to a greater understanding of artifact creation, materials used, and historical contexts, leading to enhanced preservation and research capabilities.

Existing literature highlights the value of CT imaging in cultural heritage research, demonstrating its application in revealing manufacturing processes, assessing object condition, and determining provenance. Previous studies have explored expanding the possibilities of CT imaging through advanced image processing techniques, such as unfolding unopened documents and combining CT data with other 3D imaging modalities. While high-resolution commercial micro-CT systems and synchrotron facilities exist, their cost and limitations in accommodating diverse object sizes restrict widespread use. Museums have attempted to overcome these challenges with various solutions including portable CT setups and custom-built acquisition systems. However, these solutions often represent significant investments, and do not address the potential for leveraging already-available 2D X-ray systems.

This research introduces a novel marker-based approach for generating 3D CT reconstructions from standard 2D radiography data. This method eliminates the need for high-precision, computer-controlled hardware typically found in dedicated CT scanners. Instead, it relies on small metal markers placed around the object to computationally determine the necessary geometric system parameters during image acquisition. The workflow begins with the placement of markers in foam surrounding the object, which is then positioned on a rotation stage. A series of radiographs are acquired across a full circular range, capturing the markers within the field of view. Subsequent computational steps involve marker detection and labeling to extract marker trajectories, followed by the derivation of precise system parameters using these trajectories. Preprocessing steps include flat- and dark-field correction and removal of the markers via inpainting. Finally, a 3D reconstruction algorithm is applied, generating the CT image. This procedure was implemented using existing X-ray equipment at the British Museum, the J. Paul Getty Museum, and the Rijksmuseum, demonstrating the versatility of the method. A small wooden block served as a test object for comparison across different facilities and a high-resolution micro-CT system. A 19th-century plaster model ('Python Killing a Gnu') was used as a case study at the J. Paul Getty Museum to showcase the application of this method to a complex cultural heritage object.

The research successfully demonstrated that accurate 3D CT reconstructions could be achieved using only basic 2D radiography equipment. The marker-based approach effectively compensates for the lack of precise hardware calibration, enabling the generation of CT images from systems not initially designed for this purpose. The results obtained from the wooden block test object at the three museum facilities and the micro-CT facility showed that while resolution was not as high as that achieved by a dedicated CT scanner, the method successfully revealed internal features (tree rings and saw cuts). The analysis of the 'Python Killing a Gnu' plaster model yielded critical insights not observable with 2D radiography. The 3D reconstruction revealed details about the object's construction, including the presence of a hidden original rectangular base, multiple layers of plaster, the use of wax filler material, and the precise positioning of the internal metal armature. These findings confirmed hypotheses about the object's history and modification and provided new information about the artist's techniques. The successful implementation at the Getty Museum and Rijksmuseum, facilities previously only used for 2D radiography, highlighted the wide applicability and significant potential of the method. The research code and data are publicly available, promoting further development and application of this technique.

This research successfully addresses the limitations of accessing CT scanning for museum objects by demonstrating a cost-effective and widely accessible method. The findings show the potential to significantly increase the use of 3D CT scanning in cultural heritage research by leveraging existing infrastructure. The marker-based approach avoids the need for expensive and complex CT systems, reducing the financial burden and logistical difficulties associated with off-site scanning. This is particularly beneficial for museums with limited resources or for objects too fragile for transportation. The method's versatility, proven through successful implementation in diverse museum settings, emphasizes its potential for widespread adoption within the cultural heritage community. The increased accessibility of 3D CT imaging opens up new possibilities for non-destructive analysis, offering researchers valuable insights into object construction, materials, and history. The detailed visualizations allow for a more comprehensive understanding of cultural heritage artifacts than traditional 2D radiography, supporting conservation efforts and enhancing research capabilities across diverse museum collections.

This study presents a significant advancement in the accessibility of 3D CT scanning for cultural heritage objects. By utilizing a marker-based approach with readily available 2D radiography equipment, the research successfully demonstrates a cost-effective and widely applicable method. The results obtained from multiple museum settings and a complex case study object clearly illustrate the potential of this technique to enhance research capabilities across the cultural heritage sector. Future research could focus on expanding the method to include horizontal and vertical tiling for larger objects and further automating the computational workflow for enhanced user-friendliness.

The resolution of the resulting 3D CT images is influenced by the existing hardware, primarily the focal spot size of the X-ray tube and the distances between the source, object, and detector. These factors vary across different systems, impacting image quality. Currently, the method determines relative rather than absolute scale, requiring additional measurements for quantitative analysis. The field of view is constrained by the detector size, limiting the scannable object dimensions. The present method assumes a circular trajectory, although it could potentially be adapted for more complex acquisition geometries.