利用光源投影坐标系处理SFS中的自阴影问题

对于由明暗恢复形状(SFS)问题来说，自阴影的存在是影响其重构效果的因素之一，应该加以特殊的处理。为了使自阴影的处理不延伸破坏其他区域，并能保护整体的重构形状，提出了一套从阴影分割、坐标系变换到数据拟合的自阴影处理方法。该方法使用了根据图像平面以及光源方向定义的光源投影坐标系，以降低自阴影处理问题的复杂性，从而使得可以通过逐行恢复图像内跨各自阴影数据区段高度变化量的方式来实现其基本处理要求，并可采用抛物线拟合的方法来完成重构的阴影区形状与其他部分的光滑连接。实验表明，该方法能有效地改善SFS的重构形状。

Solving Self-shadow Problem of Shape from Shading in Light Source Projected System

To the problem of shape from shading, the existence of self shadowed regions in the image is one of the factors which could cause the deformation to the final 3D reconstructed surface and it should be coped with in its special way. In the paper, a set of method is presented, which is composed of shadow region segmentation, coordinate system transmission, and the parabolic data interpolation. The method is based on the principle that the distortion of self shadow should be enclosed within its region, in order not to destroy the global reconstructed shape. The light source projected system is introduced, which is defined on the image plane and the direction of light source. This can greatly simplify the problem, and makes it possible to analyze the self shadowed data sections in the image vow by vow. And the primary goal could be achieved by recovering the depth gaps across the self shadowed data sections. Meanwhile, the height incremental values at each shadowed pixel could be estimated by parabolic interpolation, and makes the recovered shadowed shapes connect with other regions smoothly. It has demonstrated that, this proposed method could improve the reconstructed 3D shapes from shading efficiently.