文档介绍：摘要
三维重建技术作为计算机视觉领域的一个重要研究方向,已在社会生产、
生活的各个方面显示出越来越不可忽视的地位,具有广泛的应用价值。从明暗
恢复形状(SFS)的方法,由于只需单幅图像的灰度信息就可恢复景物三维表面形
状,所以近年来在理论研究和实践应用中得到了不断的发展和完善,成为三维
重建技术的一个研究热点。
论文首先回顾了 SFS 研究所走过的历程,总结了近年来该领域的研究热点
问题。在理解 SFS 问题本质的基础上,系统地归纳总结了目前国内外现有的 SFS
算法,即最小化方法、演化方法、局部方法和线性化方法,并给出了算法的性
能分析与比较。传统算法研究 SFS 问题时,大多假设一幅图像中反照率值ρ为
单一值,这种假设使得 SFS 算法的应用有局限性。针对这一点,本文给出了一
种局部反照率值的估计方法,首先计算出图像每一点对应的反照率值,设计出
基于变化反照率值的 SFS 算法;然后通过对合成图像的三维形貌恢复实验检验
了 SFS 线性化方法对正确解的收敛性,对真实图像的实验说明本算法相对于传
统算法的改进性和有效性,扩大了 SFS 算法的应用范围。
SFS 线性化方法可应用于不同领域,如自然景物、文物及艺术品复原、人
脸识别及一些人所不能及的特殊领域。随着计算机图像处理技术的不断进步和
对 SFS 问题研究的进一步深入,SFS 算法将具有广泛的应用前景。

关键词:从明暗恢复形状朗伯体反射模型线性化方法反照率估计

I
ABSTRACT
As one important research direction puter vision, 3D reconstruction
plays more and more important role that can’t be ignored in the various fields of
production and life in society and has wide application value. Recently, duing to its
special feature (it can restore three dimension shape information from one image)
shape from shading(SFS) gets great improvement and rapid development in
methodology and application . SFS es a new research hotspot of 3D passive
reconstruction.
At the beginning, a full survey about SFS research work is given. The recent
hotspots of this field are summarized. Based on understanding SFS problem in
nature, a systematic summarization is given for all current SFS algorithms which are
minimization approaches, propagation approaches, local approaches, and linear
approaches in the world. And this paper gave the analysis parison of these
reconstruction algorithms. Most traditional algorithms assumed that the albedo of
object surface is a single value in an image, but this assumption made application of
traditional SFS algorithms restrictive. This paper presents a local albedo estimation
method putes the albedo for each pixel of an image and works out a SFS
algorithm with variable albedo. Then this paper presents two sets of experiments,
T