Measurement of Gloss from Color Images

It is the goal of computer vision to automatically recover the three-dimensional shape of objects in the scene from images. Most current research in computer vision analyzes black-and-white images and assumes that the objects in the scene are matte. Brightness variation in the image is then attributed to variations of surface orientation of the objects and to material changes at object boundaries. However, real scenes generally contain glossy objects, as well as matte objects. Highlights on glossy objects provide additional brightness variations in the images and are commonly misinterpreted by current comput vision systems. Shafer has introduced a spectrally-based dichromatic reflection model that accounts for both diffuse and specular reflection. Along with the model, we describe a method that exploits the model to detect and remove highlights from color images. This approach thus provides a useful preprocessor for many areas of computer vision. We present the results of applying the technique to real images.     «

It is the goal of computer vision to automatically recover the three-dimensional shape of objects in the scene from images. Most current research in computer vision analyzes black-and-white images and assumes that the objects in the scene are matte. Brightness variation in the image is then attributed to variations of surface orientation of the objects and to material changes at object boundaries. However, real scenes generally contain glossy objects, as well as matte objects. Highlights on glos...     »