Contribution to Color Point Evaluation in Indexing Context

S. Paris (France) and H. Chabbi-Drissi (Switzerland)


color models, gray model, local jets, points of interest, in dexing, matching.


This paper investigates color images in the indexing con text. However, no indexing process is proposed here, as points are the tokens used to characterize images. These points have to be extracted and characterized in the color space. The gaussian color model will be presented and compared to two typical works in this field: the first one deals with the standard RGB space which provides some easy tools for point extraction and characterization, while the second one, defined in the gray space, is used to study the contribution of color in point extraction and character ization. Our study gives the definition of these two theo ries and then presents experiments to validate or invalidate some of of their aspects.

Important Links:

Go Back