一种含有镜面反射由明暗恢复形状的新算法  

New Algorithm of Shape-from-Shading with Specular Reflectance

作　　者：杨磊[1] 韩九强[1] 

机构地区：[1]西安交通大学电子与信息工程学院,西安710049

出　　处：《西安交通大学学报》2007年第4期435-438,共4页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(60502021);教育部高等学校博士点专项科研基金资助项目(20050698025)

摘　　要：提出了一种基于混合反射模型的由明暗恢复形状的离散算法.采用有限差分近似微分运算,将一阶非线性微分方程所描述的反射图方程转化为关于未知表面高度的代数方程,再由反射图方程和图像梯度信息构造目标方程,进而用Newton迭代算法求出该方程的数值解,得到表面三维形状.所提出的新算法具有快速准确的特点,在恢复物体的细节和边缘时,比同类算法更准确.合成半球图像的实验结果表明,与三角形面元算法相比,新算法恢复高度的平均相对误差和CPU运行时间分别减少了37.2%和20.22%.A discrete shape-from-shading （SFS） algorithm based on hybrid reflectance model is proposed. The finite difference is used to approximate differential operation; the reflectance map equation described by the first order nonlinear differential equation is transformed into an algebra- ic equation about the unknown surface heights, and then the objective equation is constructed by the reflectance map equation and gradient information of image. Moreover, the Newton iterative algorithm is utilized to obtain the numerical solution and 3D shape of the surface. The proposed SFS algorithm is fast and effective and can recover more edges and details of objects than other same kind of methods. The experimental results of the synthesized hemisphere image show that the relative height error and CPU time of the proposed SFS are dropped by 37.2% and 20. 22% respectively compared with the triangular-facet-based algorithm.

关 键 词：三维形状恢复 混合反射模型 反射图方程 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]