【摘要】 由明暗重构三维形状(SFS)是计算机视觉当中有关初级视觉的研究内容之一。由于该问题存在明显的不适定性,因而需要采用使问题正则化的形式,其中,将问题转变成变分形式的最小化方法是最早并得到广泛研究的方法类,但这类方法的整体表现并不好。论文对此类方法从其解决问题的思路上进行分析,找出了该类方法当中存在的一些问题,即在正则化问题的过程当中引入了无具体意义的约束条件,从而对该类算法所表现出的一些现象给出了较为合理的解释。更多还原

【Abstract】 Shape from Shading is a kind of typical ill-posed problem,which are widely concerned in the field of com-puter vision.It discusses how to derive the shape of the objects only through the shading information within a image.Many minimization methods have been discussed since the concept of SFS was proposed,but they usually perform badly and unsteadily.The analysis in this paper will disclose the reason of this kind phenomenon lies on the fact that,some meaningless constraints are involved during the regularizing process.更多还原