Material properties include texture of a surface and some optical parameters of a scene such as reflectance and refractive index, and so on. Recovering material properties are referred to obtain these parameters from images and a geometric model of a scene. Some algorithms have been developed for opaque objects. Our work focuses on translucent objects. Compared with the traditional opaque objects, translucent objects have not only reflectance parameters, but also transparent parameters and refractive index as well. Due to the interaction of the parameters in the rendering, it's more difficult to recover them.

At Start, we investigate and present a new method for recovering material properties of crystal-like objects. First of all, the objects in the image are separated into two classes: the opaque objects and the translucent objects. For the translucent objects, by analyzing refractive phenomena, we propose a method to recover an approximation of the refractive index and transmission coefficients. First, obtain the refractive index with the minimum error of color variance by a certain step calculations and then iteratively recover the color parameters. Experiments show that our algorithms can be completed in few iterative steps, and recovery error is very small. With the recovered material properties, the scene can be rendered under novel lighting or viewing conditions, and it can also be rendered when new objects are added into the scene.

Further, we investigate how to recover materials of translucent objects. Translucent objects have not only reflective coefficient but also transmission coefficient and refractive index. Here a new algorithm is presented to recover all these parameters iteratively. According to the interrelation and interaction of the parameters, we identify the sequence of parameters in the recovery and design an effective algorithm to get the parameters of the translucent objects under complex environment illumination. Furthermore, we do some experiments to validate the feasibility.