To apply the science of photogrammetry to close-range photography for capturing 3D models.
About the Project
Photogrammetry is the science of making measurements of an object using a number of 2D images which can include reconstructing the object in 3D. This is typically used with aerial photography to produce 3D maps, most recognisably Google Maps, as well as for the gaming industry to digitise structures.
The science is also applicable to close-range photography for capturing 3D models – an area that SCITEK has been exploring and investigating for a number of projects. Various photogrammetry packages are available, but typically follow the same workflow when reconstructing a scene:
- Images are first matched and aligned, a process known as aerotriangulation. This requires the same feature to be found in at least three images.
- Depth maps and dense clouds can then be created before a mesh is reconstructed.
- Images can also be used to project a high quality ‘texture’ onto the model resulting in high colour resolution.
The below shows an example reconstruction of a shell using the photogrammetry technique:
SCITEK is exploring the potential of the technology for capturing 3D information for a high number of parts automatically. In an era of big data, this could unlock huge potential in manufacture or damage analysis, from providing engineers with high quality models of designed parts to feature analysis of damaged parts. Time can then be focused on designing and improving components.
It also releases capabilities for feeding used or damaged parts into simulations so their behaviour can be better understood. Deeper knowledge of how component variations behave can allow parts which otherwise would have been scrapped to remain in use, leading to a more profitable and environmentally-friendly future.
Video by Bentley Institute