Reflectance reconstruction of objects with spatially-varying materials

This paper presents a method to reconstruct the surface reflectance of objects from images. For objects with spatially-varying materials, we assume that each point can be represented as a weighted sum of several basic materials. First, a cluster algorithm is applied to automatically compute the number of basic materials. Then, the weight map and parameters of the reflectance model are iteratively optimized. In this step, diffuse and specular albedo is separately estimated to exhibit more reliable results. Finally, using our output parameters the object could be reconstructed under arbitrary illumination/viewing conditions. Besides those objects composed with distinct materials, this method is also useful to reconstruct objects mixed with complex materials, such as alloy and marble.