How Useful are Colour Invariants for Image Retrieval?

The images captured by a digital camera do not depend only on the characteris- tics of the objects in the scene but also on the colour of the scene illumination. To account for this confounding influence various colour invariants have been introduced whose aim is to provide descriptors that do not change with a change in light. In this paper we evaluate the usefulness of these invariants for general purpose content-based image retrieval. The results show that the application of invariants may lead to a severe degradation in retrieval performance.

[1]  Christos Faloutsos,et al.  QBIC project: querying images by content, using color, texture, and shape , 1993, Electronic Imaging.

[2]  M. H. Brill,et al.  Heuristic analysis of von Kries color constancy. , 1986, Journal of the Optical Society of America. A, Optics and image science.

[3]  Bernt Schiele,et al.  Comprehensive Colour Image Normalization , 1998, ECCV.

[4]  Arnold W. M. Smeulders,et al.  Color Based Object Recognition , 1997, ICIAP.

[5]  Brian V. Funt,et al.  Color Angle Invariants for Object Recognition , 1995, CIC.

[6]  Brian V. Funt,et al.  Color Constant Color Indexing , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

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

[8]  G. Buchsbaum A spatial processor model for object colour perception , 1980 .

[9]  Michael J. Swain,et al.  Color indexing , 1991, International Journal of Computer Vision.