Combining micro-computed tomography (CT) with a photographic approach could improve accuracy and specificity when digitizing natural objects, Japanese researchers have found.
Takashi Ijiri, PhD, and colleagues aren’t pioneers of the mixed approach to digitization—many studies have explored the feasibility of using CT and manual photography in digitizing objects like flowers and insects—but combining the two is a novel idea, the authors wrote in a PLOS ONE study.
“Natural objects, such as insects and plants, have been important targets of digitization because digital formats have various benefits,” Ijiri et al. wrote. “They are deterioration-free, space-efficient and highly accessible. In addition, it is possible to capture and store digital information that is invisible to the naked eye by using various devices, like highly detailed surface textures obtained with microscopy and internal structures captured with x-ray CT.”
Both avenues have seen success, the researchers said, but they also both have pitfalls. While the image-based approach has the ability to obtain 3D shapes and textures at the same time, the fact that it relies solely on photos taken from multiple viewpoints means the method is weak when it comes to reconstructing concave or occluded areas.
Almost oppositely, the CT-based approach reconstructs an organism by segmenting volume measurements obtained through x-ray CT scans but fails to image colors and has a hard time reconstructing surface textures.
Ijiri and co-authors said their technique combined both approaches—CT volume was used to generate a base shape, while photographs taken from a multitude of angles helped achieve an accurate surface texture.
“A key idea is camera position estimation,” the authors wrote. “Given a photograph and a 3D model obtained from CT volume, we estimate the camera position of the photograph relative to the 3D model. We then project the photograph onto the 3D model from the found camera position to obtain a texture.”
The researchers also devised techniques for combining a handful of textures obtained from multiple photographs and for synthesizing missing texture regions caused by occlusion. They proved their method successful in digitization tests of small natural organisms.
“We illustrated the feasibility of the presented technique by adopting it to flowers, insects and mushrooms to create 3D textured models of them,” Ijiri and colleagues said. “Our ongoing future work is to automate the process completely.”