Abstract
Recovering the shape of the surface from its grayscale image is a classical problem in image processing, called "shape-from-shading" (SFS). One of main components of any SFS technique is the deformable surface model. In this work we present the deformable surface model, which is based on concept of active contours, or "snakes". The surface is first divided into the sequence of its cross-sections, or profiles. A profile is defined as an interpolation between equally spaced snake elements. The grayscale image of a profile is generated using ray tracing and compared to the corresponding fragment of an input image. The difference is included into the energy functional of the snake as an external energy term. The snake elements are iteratively migrated, thus deforming the profile, towards the energy functional minimization. "Segments" and "Leveling" subroutines are introduced to optimize the process and overcome trapping in local minimums.
Original language | English (US) |
---|---|
Pages (from-to) | 218-222 |
Number of pages | 5 |
Journal | Proceedings of the IEEE International Conference on Systems, Man and Cybernetics |
Volume | 1 |
State | Published - 2002 |
Event | 2002 IEEE International Conference on Systems, Man and Cybernetics - Yasmine Hammamet, Tunisia Duration: Oct 6 2002 → Oct 9 2002 |
ASJC Scopus subject areas
- Control and Systems Engineering
- Hardware and Architecture
Keywords
- Deformable Surface
- Profile
- Ray tracing
- Snake
- Surface recovery