陰影情報およびB-Spline曲面近似による立体形状の認識 : 第2報, ステレオ視による立体形状の認識

概要

論文の詳細を見る
In our previous study, we applied B-Spline surface method to estimate the local surface of the shading and used the steepest descent method to minimize the difference between the B-Spline surface and the image shading. These methods helped reduce the effects of the noise and also recognized the shape more accurately than by the conventional methods. However, we detected some problems in the applied methods, such as distortion in the result, presence of invisible occluded region, etc. Therefore, it is necessary to improve the accuracy in practical USe. In the present study, we observed the object from various directions by adopting a stereo vision for 3D recognition in the process. This newly proposed method was able to recognize the shape from image shading, and the method itself was also easy to achieve by selecting proper weights of two views.
1998-08-25