|
Putting pots together by matching their profiles. |
|
Written by Andrew Willis
|
|
Monday, 30 January 2006 |
If a surface of revolution such as an archaeological pot is broken into many fragments it is clear that given accurate measurements of all the pieces, one may reconstruct the broken object from its pieces. In a new paper, we describe how to reconstruct surfaces such as these using only their apparent outer contour when viewed from the side. One major benefit of this approach is that the surface may be reconstructed from pieces which may not share a matching boundary. All that is necessary to reconstruct the apparent contour is a set of fragments which, when matched, cover the entire apparent contour of the unknown pot shape. From the figure below, (a) and (c) are 3D meshes of 2 fragments from a single pot. Their estimated apparent outer contours are (b) and (d) respectively. The complete profile estimated by matching the apparent contours is shown in (e). Note that the curve in (e) may be revolved around the pot's axis of symmetry (also estimated) to regenerate the a-priori unknown pot outer surface. |
|
Last Updated ( Thursday, 09 November 2006 )
|
|
|
Written by Andrew Willis
|
|
Sunday, 29 January 2006 |
|
Work has started on image processing techniques for performing experiments using data available from the Mars rovers Spirit and Opportunity. A significant amount of the rover power is expended on communications, i.e., receiving instructions and sending data back to Earth. Automatic detection of interesting features can be used to perform semi-autonomous experimentation. Automatically performed experiments will reduce the amount of transmitted instructions to the spacecraft. Energy saved using the more efficient communication scheme may then be allocated to other mission critical tasks such as prolonging the life of the spacecraft or further experimentation, data collection, and transmissions.
|
|
Last Updated ( Monday, 30 January 2006 )
|
|
|
Written by Andrew Willis
|
|
Thursday, 10 November 2005 |
| The Charlotte Machine Vision Laboratory Woodward Hall Room 213 University of Charlotte 9201 University City Blvd. Charlotte, NC 28223 |
Welcome to the VisionLab, a research laboratory for investigating problems in computer vision, pattern recognition, medical and natural image processing and solving difficult geometric problems in high (>=3) dimensional spaces.
|
|
Last Updated ( Thursday, 20 September 2007 )
|
|
|
<< Start < Prev 1 2 3 4 5 Next > End >>
|
| Results 10 - 12 of 13 |
