Abstract
In this paper the application of computer graphics and computer vision for texture extraction from infrared image sequences is described. These techniques normally are used in computer graphics to project a virtual scene onto the image plane. For the matching between the images and the given 3D model a strategy is presented based on the estimation of planes in the image sequence using homography. Textures are extracted using an algorithm based on the principles of ray casting to generate partial textures for every visible surface in every single image of the sequence. The textures generated from different images of the sequence belonging to the same façade are combined. By combining the intersection points of several images generated during the ray casting, a dense structure of points can be used to texture and analyse even big building complexes.
| Original language | English |
|---|---|
| Pages (from-to) | 65-70 |
| Number of pages | 6 |
| Journal | International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives |
| Volume | 36 |
| State | Published - 2007 |
| Event | Joint Conference of ISPRS Working Groups I/2, III/2, III/4, III/5, IV/3 on Photogrammetric Image Analysis, PIA 2007 - Munich, Germany Duration: 19 Sep 2007 → 21 Sep 2007 |
Keywords
- 3D building model
- Close range
- Image analysis
- Image sequences
- Terrestrial
- Texture extraction
- Thermal infrared
- Urban