Wavelength Selection Using a Modified Camera to Improve Image-Based 3D Reconstruction of Heritage Objects

ABSTRACT Tools for image-based 3D-reconstruction are commonly used for cultural heritage applications; however, wider usage has increased variability in the quality of output 3D models. Geometric variations between 3D models acquired with differing methods make metric conservation applications such as condition monitoring and measuring change over time challenging. This article presents an investigation of wavelength selection using a modified off-the-shelf DSLR camera and bandpass filters to improve input image quality in a 3D-reconstruction study of a wooden sculpture of a coyote and turtle from the Smithsonian American Art Museum. The sculpture has a large crack of concern to conservators, but its curved, dark shiny surface challenges image-based dimensional monitoring. Selecting infrared wavelengths rather than the visible light for 3D reconstruction input images reduced specular surface reflections and improved image contrast resulting in improved recording of the 3D shape. 3D-reconstructions using infrared radiation produce better reconstructions than those using visible light. In this case reconstructed surface discrepancies between visible light are ∼0.6 mm whilst those using infrared are ∼0.3 mm. Results suggest that reflected infrared images are more forgiving and flexible for recording 3D data over time for dark, shiny wooden surfaces and thus improve the reliability and comparability of image-based 3D-reconstruction.