Global and Local Highlight Analysis in Color Images

In this paper we present two approaches to highlight removal in color images. The first approach is based on a global analysis of a single color image. Threedimensional color descriptors are transformed to a pair of two-dimensional descriptors. The second approach is based on a local analysis of three color images. The images are taken in analogy to photometric stereo tasks. For every surface point in the scene there exist three color vectors which are stored in the three images. The matte reflection component is calculated by analyzing exclusively these three vectors. Both approaches in this paper are based on the dichromatic reflection model for inhomogeneous dielectric materials.