Name
Abstract | ||
|---|---|---|
We present a new image-based algorithm for surface reconstruction of moving garment from multiple calibrated video cameras. Using a color-coded cloth texture, we reliably match circular features be- tween different camera views. As surface model we use an a priori known triangle mesh. By identifying the mesh vertices with tex- ture elements we obtain a coherent parameterization of the surface over time without further processing. Missing data points result- ing from self-shadowing are plausibly interpolated by minimizing a thin-plate functional. The deforming geometry can be used for different graphics applications, e.g. for realistic retexturing. We show results for real garments demonstrating the accuracy of the recovered e xible shape. Overview We consider cloth motion capture as an important validation for and alternative to simulation approaches. Real-time simulations re- quire simplications which should be justied by real-world data. In many cases the forces which drive cloth models are hard to model (e.g. aerodynamic or friction forces from body contact). To close the gap between model and experiment, an automatic measurement technique is needed. Earlier work (Pritchard and Heidrich 2003) uses a calibrated stereo camera pair and obtains the surface parameterization by feature matching. Another approach (Guskov et al. 2003) uses color-coded quad markers for real-time acquisition of non-rigid surfaces, includ- ing cloth. Our method introduces a new color code for this purpose which is more general and can be used for real garments. In contrast to earlier methods, it can also cope with fast motion. Our approach requires a custom-printed cloth pattern. We have cho- sen M-arrays (Morano et al. 1998), a color code which encodes each point in the pattern by its spatial neighborhood. Each 3 3 neigh- borhood of a point is unique and can be used for point identica- tion. Additionally, a triangle mesh for the garment is constructed as input for the acquisition algorithm. Based on photographs of the cloth panels, we design corresponding triangle meshes and sew them together. The scene is recorded with eight synchronized video cameras. Im- age feature recognition uses color classication in HSV color space and edge detection for the ellipse shapes. For matching the features to the cloth pattern we use a novel region grow approach where seed points are identied by a 3 3 window in the pattern that are grown using a search technique which adapts to the pattern lattice structure. Using geometric camera calibration and 2D image fea- ture positions, the 3D vertex coordinates are reconstructed by tri- angulation. The tted mesh is further processed by interpolative hole lling. A detailed description of the algorithms can be found in (Scholz et al. 2005). |
Year | DOI | Venue |
|---|---|---|
2005 | 10.1145/1187112.1187157 | international conference on computer graphics and interactive techniques |
Keywords | Field | DocType |
surface model,different graphics application,different camera view,coherent parameterization,texture element,triangle mesh,mesh vertex,garment motion,surface reconstruction,circular feature,cloth texture,missing data,feature recognition,edge detection,real time,motion capture,hsv color space,camera calibration,region growing | Computer vision,Motion capture,Computer graphics (images),Computer science,Artificial intelligence | Journal |
Volume | Issue | Citations |
24 | 3 | 31 |
PageRank | References | Authors |
2.52 | 5 | 5 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Volker Scholz | 1 | 43 | 4.07 |
| Timo Stich | 2 | 123 | 8.56 |
| marcus magnor | 3 | 383 | 28.88 |
| Michael Keckeisen | 4 | 157 | 11.48 |
| Markus Wacker | 5 | 123 | 9.33 |