Color illumination models for image matching and indexing

Recent works have demonstrated that the direct use of grey levels for image matching and indexing allows one to build very powerful image recognition systems. The present paper attempts to enlarge these results to the case of color images. First, it presents a small abstract about photometry and cameras, which allows one to justify the choice of a color representation system. Then it presents, evaluates, and compares several illumination models, and discusses image normalization techniques. Finally, the paper presents a set of color invariants for image matching and indexing.

[1]  Charles A. Poynton,et al.  A technical introduction to digital video , 1996 .

[2]  R. Mohr,et al.  What accuracy for 3D measurements with cameras? , 1996, Proceedings of 13th International Conference on Pattern Recognition.

[3]  Gaurav Sharma,et al.  Digital color imaging , 1997, IEEE Trans. Image Process..

[4]  Cordelia Schmid,et al.  Local Grayvalue Invariants for Image Retrieval , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[5]  M. S. Drew,et al.  Color constancy - Generalized diagonal transforms suffice , 1994 .

[6]  Reinhard Koch,et al.  Matching of affinely invariant regions for visual servoing , 1999, Proceedings 1999 IEEE International Conference on Robotics and Automation (Cat. No.99CH36288C).

[7]  Charles Poynton,et al.  Frequently Asked Questions about Color , 1997 .

[8]  Brian V. Funt,et al.  Color Angular Indexing , 1996, ECCV.