USE OF INVARIANTS FOR RECOGNITION OF 3-DIMENSIONAL COLOR TEXTURES

We introduce an image model for color texture based on spatial correla tion functions defined both within and between color bands. We show th at three-dimensional geometric transformations of a surface in the sce ne produce corresponding transformations in these correlation function s. From this analysis we derive invariants of color correlation functi ons that can be computed efficiently and that can be used for geometry -invariant recognition. We show experimentally that these invariants a re effective for recognition in situations in which neither color dist ributions nor gray-scale texture is sufficient. After recognition the estimated color correlation functions can be used to recover the three -dimensional location and orientation of surfaces. The color-texture i nvariants can be used in the recognition of three-dimensional objects, the segmentation of images of three-dimensional scenes, and the searc hing of image data bases.