【作者】 张军； 戴霞；

【Author】 Zhang Jun Dai Xia (Institute of computer and mathematics,XiHua University,Chengdu 610039)

【机构】 西华大学数学与计算机学院；

【摘要】 提出了一种基于"误差约减"思想的从明暗恢复形状方法,使从明暗恢复形状方法的精度进一步提高。第一步,利用现有从明暗恢复形状的方法得到曲面的三维信息;第二步,在一定光照模型下计算出恢复曲面的光照图像,再从该图像中恢复曲面形状;第三步,将两次计算得到的曲面三维信息进行比较以推断误差并进行修正。实验结果显示,该算法能减少从明暗恢复形状方法的模型误差,使其更加适应真实环境下的SFS问题。更多还原

【Abstract】 This paper proposes an error-reduction-based algorithm for shape from shading to improve the precision of SFS method.We carry out the algorithm in three steps:first,reconstruct the 3D information of the surface by the existent method;then,project the surface to the image under a certain illumination model and reconstruct the 3D information of the surface from that image;finally,these two reconstruction results are combined to reduce the reconstruction error and produce a superior surface.The experiment results show that the algorithm has a high efficiency to reduce the reconstruction error.更多还原