Abstract: To improve the precision of imaging three-dimensional reconstruction based on Shape from Shading, an improved illumination model was proposed. This model not only improved the diffusion reflectance component, but the influence of specular reflectance component with linear superposition was also considered. The variables of the model were dispersed by using finite difference method and the model was solved by using Jacobi Iteration. And then the three-dimensional reconstruction height was obtained. By taking the metallic ball image as example and reconstructing its three-dimensional image, the experimental results showed that the relative height error after using improved illumination model was dropped by 12.19 % compared to the traditional methods.

Key words: shape from shading, illumination model, three-dimensional reconstruction, Jacobi iteration

摘要： 针对基于灰度重构形状的表面三维重构方法中重构精度不高的问题,提出一种改进的光照模型。该模型不仅改进了物体表面的漫反射分量,而且考虑了物体表面镜面反射分量对表面重构所产生的影响,把漫反射分量和镜面反射分量线性叠加。采用有限差分法离散改进光照模型中的变量,再利用Jacobi迭代法求解改进光照模型，得到物体表面点的重构高度值。以金属半球图像为例，分别采用改进的光照模型和传统光照模型进行三维重构。通过结果分析表明，采用改进的光照模型提高了重构精度，重构后的高度误差降至12.19%。

关键词: 灰度重构形状, 光照模型, 三维重构, Jacobi迭代