计算机视觉中传统的物体三维形状恢复方法大都基于光源是无穷远点光源的假设,如由单幅图像恢复三维形状(shape from shading)、光度立体视觉(photometric stereo)方法等。实际环境中的光源往往不能满足这个假设,因此大大限制了这些方法的应用。本文提出了一种更加实用的球扩展光源模型,并推导出此光源下漫反射物体表面的反射图。此模型在三维坐标系中描述了球光源位置、亮度、漫反射物体表面反射度与图象灰度的关系,从而可方便地应用于计算机视觉和计算机图形学。基于此模型,本文由单幅图像方法恢复物体三维形状。实验结果表明,基于此模型的由单幅图像恢复三维物体形状方法可正确恢复物体三维形状。

It will be very useful if shape recovery can be performed under indoor lights which maybe spherical shape. As a first step towards this goal a spherical light source model is developed for doing shape recovery. This model is more practical and convenient than other extended light source model. Base on it, the shape from shading method can be used to recover surface shape from one image. Our method is verified by both real and simulated experiments. The results show that this light source model is convenient and method for shape recovering is accurate.