摘要
在Pentland的线性化SFS模型基础上,提出一种基于松弛迭代方法的线性SFS算法,给出并证明该算法收敛的弱必要条件,并对实际遥感地形图像进行三维恢复的实验。通过与Pentland、TsaiShah以及Ulich等方法的比较,说明本文方法具有如下特点:(1)在边界条件未知的情况下,通过松弛迭代方法确实可以在一定程度上求得相应于图像的物体表面高度;(2)松弛因子在一定程度上控制并反映了表面高度的粗糙程度,并在一定意义上滤除了噪声。
In this paper, a new relaxation algorithm based on Pentland＇ s linear SFS model is presented, for which a weak neces- sary condition is also proposed and proved. Some experiments on shape recovery from a single terrain model image, by the meth- ods of Pentland, Tsai-Shah, Ulich and ours respectively, show that： （1）the relaxation based algorithm could really obtain the corresponding surface of the image to some extent without the boundary conditions; （2）the relaxation factor could control and re- fleet the roughness of the surface to some extent and might remove the noises in a meaning way.
出处
《计算机与现代化》
2013年第1期40-44,共5页
Computer and Modernization
