Weighted Map for Reflectance and Shading Separation Using a Single Image

In real world, a scene is composed by many characteristics. Intrinsic images represent these characteristics by two components, reflectance (the albedo of each point) and shading (the illumination of each point). Because reflectance images are invariant under different illumination conditions, they are more appropriate for some vision applications, such as recognition, detection. We develop the system to separate them from a single image. Firstly, a presented method, called Weighted-Map Method, is used to separate reflectance and shading. A weighted map is created by first transforming original color domain into new color domain and then extracting some useful property. Secondly, we build Markov Random Fields and use Belief Propagation to propagate local information in order to help us correct misclassifications from neighbors. According to our experimental results, our system can apply to not only real images but also synthesized images.