AbstractCultural heritage objects can be recorded by a variety of 3D techniques. Accessible consumer imaging systems make it increasingly easy to capture image sets that can be processed to produce 3D reconstructions by heritage users who are not necessarily skilled in the process. The availability, flexibility, and automation that make image-based 3D reconstruction an accessible process increase the variability in the output models presenting new challenges in maintaining consistent, fit for purpose, 3D outputs. Conservation applications for image-based 3D reconstruction therefore require scientific rigour to establish reliable and accurate models, particularly for monitoring condition and measuring
change over time. Widening access requires better, more accessible, investigative methods to improve image-based 3D reconstruction and thereby ensure consistent high-quality. Reconstructions are dependent on the geometric and photographic qualities of the input images and the methodology used in the reconstruction workflow.
This thesis identified and investigated two less well explored aspects for improving imagebased 3D reconstruction through laboratory experimentation, both focused on the qualities of the input imagery. The first concerned the quantifiable scientific use of a modified camera combined with wavelength selection, while the second addressed a gap in the state of the art, seeking to quantify depth of field for 3D reconstruction with an associated automatic process of sharpness-based masking. The thesis demonstrated that a modified consumer digital camera can be paired with wavelength selection to improve image-based 3D reconstruction, demonstrating that reflected IR images can increase local image contrast and reduce wavelength specific specular reflections of a particular museum object to improve the consistency of the resulting 3D reconstruction. The imaging study and proposed wavelength selection workflow were informed by camera characterisation. The second approach experimentally quantified aspects of depth of field for 3D reconstruction of small to medium sized museum objects to inform imaging setup and acquisition combined with sharpnessbased masking as part of an image pre-processing step to optimise the processing performance for 3D reconstruction.
Working at the intersection of collections photography, conservation documentation, and scientific imaging, this thesis provides methods of imaging processes that enable heritage specialists to quantify their imaging systems in order to produce accurate and reliable scientific 3D reconstructions of small to medium sized heritage objects. Developed processes are demonstrated throughout the thesis by iterative recording and 3D reconstruction of test and museum objects. The outcome is the optimisation of an imagebased 3D reconstruction methodology to provide accurate and reliable scientific recording for conservation documentation while considering the accessibility to a wide range of users.
|Date of Award||Mar 2020|
|Supervisor||Roger Evans (Supervisor)|