Abstract: This paper psoposes a method of reconstructing curved surface from two images. Two images are photographed from the frontal and lateral direction of the curved object. Then the silhouettes of the two images are obtained by user-interaction. And the smoothest 3D surface which projects exactly to these silhouettes can be recovered by minimizing the linearly constrained quadratic object function. In addition, surface creases are realized by adding a parameter to the quadratic object function. The promising reconstructing results validate the effectiveness of this method.

Key words: 3D reconstruction, curved surface reconstruction, image based modeling

摘要： 提出了一种利用曲面物体的两幅图像进行曲面重建的方法。从物体的正面和侧面拍摄两幅图像，而后利用用户交互获得这两幅图像中的侧影轮廓线。并通过最小化以这些侧影轮廓线作为约束二次目标函数，重建出最光滑的三维曲面。另外，在二次目标函数中增加一个系数，实现了曲面折痕。利用真实图像进行实验，验证了该方法的可行性和正确性。

关键词: 三维重建, 曲面重建, 基于图像的建模

CLC Number: 

• TP391