Our brain is able to reconstruct volumes and perceive the shape of an object by relying on multiple sets of clues:
- Motion parallax and perspective give a relative sense of the position of objects in the scene
- Focus and field of view
- Stereoscopic effect (slight view difference between two eyes)
- Experience and familiarity with the object fills the remaining gaps (by recognizing an object we expect it to have a given shape)
[Click on the image to see the animation]
This couple of stereo-images gives enough information for the brain to reconstruct the depth of the scene. (source: http://www.stereomaker.net/sample/ani/ani_e.htm)
As a side note, the fact that all these simultaneous information are used by the brain means that, beside holography, there is no possibility to display great 3D information without causing motion sickness or nausea to some people.
A 3D scanner using all these passive clues would be complex to build and probably not very accurate. So most of the current scanner rely on active methods.
Active scanners can use laser for direct distance measurements:
- Time of flight: these videos shows how to use laser to scan the Sphinx or drive autonomous cars in the DARPA challenge
- Or displacement in the case of some industrial sensors. These sensors can achieve amazing accuracy in micro-meters over a range of a few centimeters
These sensors are unfortunately very expensive and not accessible to the hobbyist, so, to build a home scanner we have to rely on triangulation. The principle is to send a coherent pattern on a object to extract its shape from the deformation of the shadow. A great introductory course is available here.
After examining all the 3D scanner option I’ve chosen to use a laser to project a line and cameras will capture the line deformation. The line sweep will be performed using two galvanometers used in the laser show business. This is the perfect occasion to go and explore the world of laserists…
Two galvos are used to deflect the laser beam on the X and Y axis