The key issues such as vision sensor, sampling planning are studied for reverse engineering. For the developed sensor, the low sensitivity to the incident angle and the high linearity between the output and the input ...The key issues such as vision sensor, sampling planning are studied for reverse engineering. For the developed sensor, the low sensitivity to the incident angle and the high linearity between the output and the input are carried out. During digitizing, the adaptive sampling planning method based on the curvature of surface and the slope of tangent line of measured point is proposed. After integrating the vision sensor, CNC milling and relevant software, the reverse engineering system based on computer vision is developed. The experiment shows that this system is reliable and efficient.展开更多
A new method is put forward combining computer vision with computer aidedgeometric design (CAGD) to resolve the problem of free-form surface reconstruction. The surface isfirst subdivided into N-sided Gregory patches,...A new method is put forward combining computer vision with computer aidedgeometric design (CAGD) to resolve the problem of free-form surface reconstruction. The surface isfirst subdivided into N-sided Gregory patches, and a stereo algorithm is used to reconstruct theboundary curves. Then, the cross boundary tangent vectors are computed through reflectance analysis.At last, the whole surface can be reconstructed jointing these patches with G^1 continuity(tangentcontinuity). Examples on synthetic images are given.展开更多
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 ...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.展开更多
文摘The key issues such as vision sensor, sampling planning are studied for reverse engineering. For the developed sensor, the low sensitivity to the incident angle and the high linearity between the output and the input are carried out. During digitizing, the adaptive sampling planning method based on the curvature of surface and the slope of tangent line of measured point is proposed. After integrating the vision sensor, CNC milling and relevant software, the reverse engineering system based on computer vision is developed. The experiment shows that this system is reliable and efficient.
基金National Natural Science Foundation of China(No.59975057).
文摘A new method is put forward combining computer vision with computer aidedgeometric design (CAGD) to resolve the problem of free-form surface reconstruction. The surface isfirst subdivided into N-sided Gregory patches, and a stereo algorithm is used to reconstruct theboundary curves. Then, the cross boundary tangent vectors are computed through reflectance analysis.At last, the whole surface can be reconstructed jointing these patches with G^1 continuity(tangentcontinuity). Examples on synthetic images are given.
文摘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.
