|
There were a couple of things that caused the reconstructions to have problems:
|
|
If the object had any specular reflections, they would appear as spikes in the
depth image. To fix this, maybe it would be possible to automatically detect
specular reflections in the image, and remove or ignore them prior to
computing the normals.
|
|
If the object had jots of color variation on its surface, it tended to add depth
errors to the reconstruction. In some cases, it appears that color variation can
be difficult to distinguish from albedo variation. Examples of this are the
eyes and whiskers of the cat, which have more surface definition in the reconstruction
than in the original images.
|
| Click here to download original images
|
