We show that x-ray computer tomography algorithms can be applied with minimal alteration to the three-dimensional reconstruction of visible sources. Diffraction and opacity affect visible systems more severely than x-ray systems. For camera-based tomography, diffraction can be neglected for objects within the depth of field. We show that, for convex objects, opacity has the effect of windowing the angular observation range and thus blurring the reconstruction. For concave objects, opacity leads to nonlinearity in the transformation from object to reconstruction and may cause multiple objects to map to the same reconstruction. In x-ray tomography, the contribution of an object point to a line integral is independent of the orientation of the line. In optical tomography, however, a Lambertian assumption may be more realistic. We derive an expression for the blur function (the patch response) for a Lambertian source. We present experimental results showing cone-beam reconstruction of an incoherently illuminated opaque object.
© 2001 Optical Society of America
[Optical Society of America ]
(100.6890) Image processing : Three-dimensional image processing
(100.6950) Image processing : Tomographic image processing
(110.6880) Imaging systems : Three-dimensional image acquisition
(150.6910) Machine vision : Three-dimensional sensing
Daniel L. Marks, Ronald Stack, Andrew J. Johnson, David J. Brady, and David C. Munson, "Cone-Beam Tomography with a Digital Camera," Appl. Opt. 40, 1795-1805 (2001)