摘要
Visual method is now broadly used in reverse engineering for 3D reconstruction. The traditional computer vision methods are feature-based, i.e., they require that the objects must reveal features owing to geometry or textures. For textureless free-form surfaces, dense feature points are added artificially. In this paper, a new method is put forward combining computer vision with CAGD. The surface is subdivided into N-side Gregory patches using marked curves, and a stereo algorithm is used to reconstruct the curves. Then, the cross boundary tangent vector is computed through reflectance analysis. At last, the whole surface can be reconstructed by jointing these patches with 1Gcontinuity.
Visual method is now broadly used in reverse engineering for 3D reconstruction. The traditional computer vision methods are feature-based, i.e., they require that the objects must reveal features owing to geometry or textures. For textureless free-form surfaces, dense feature points are added artificially. In this paper, a new method is put forward combining computer vision with CAGD. The surface is subdivided into N-side Gregory patches using marked curves, and a stereo algorithm is used to reconstruct the curves. Then, the cross boundary tangent vector is computed through reflectance analysis. At last, the whole surface can be reconstructed by jointing these patches with 1Gcontinuity.
基金
the National Science Foundation(59975057)
