融合SFS和主动视觉技术的未知物体重建方法

Research on reconstruction method for unknown objects through incorporating SFS algorithm and active vision technology

针对未知三维物体自动重建问题,提出一种结合明暗恢复形状的被动视觉和主动视觉技术的新颖三维重建方法。首先利用摄像机分别获取初始位置及旋转180°位置下的未知物体图像信息,采用改进的Tsai-SFS算法恢复物体表面粗糙轮廓,从而获取未知物体的最大尺寸信息。然后结合单目线激光主动视觉系统沿Z轴旋转方向和绕X轴翻转方向上的可视空间模型,预测出物体未知区域信息。最终依据下一视点的可见性判据,将规划过程分为物体侧表面重建规划和上表面重建规划。其中将能获取最大可视曲面面积和最佳表面重建精度的位置定义为下一最优视点位置。并经过实体模型的三维重建验证所提方法的可行性及有效性。

This paper presents a novel 3D reconstruction method through incorporating passive vision method based on shape from shading and active vision method based on visible space,which aims at the automatic measurement and reconstruction of unknown three dimensional objects.Firstly,the images of unknown object at initial position and the position rotating 180 degrees from the initial position are obtained using a CCD camera.To obtain the rough model of the unknown object,an improved Tsai-SFS algorithm is adopted to recover the object shape according to the prior obtained images.And then through incorporating the visible space models of the monocular line-laser active vision system along Z axis rotation and X axis flip directions,the unknown region of the object is predicted.Finally,according to the visibility of the next viewpoint,the view planning process is divided into an object surface reconstruction planning and an upper surface reconstruction planning.And the location that can obtain maximum visual surface area and optimal reconstruction precision is defined as the next best view point.Real object was reconstructed in experiment to validate the feasibility and efficiency of the proposed method.