Name
Abstract | ||
|---|---|---|
Content-based image retrieval primar- ily used color distributions as descrip- tors of the image content; researches have since focused on the use of var- ious color representation spaces, color and illumination invariance, color quan- tization and color matching. In order to overcome the many limitations of the description by a first-order distribution, several higher-order distributions have been introduced since (like autoconel- ogram or color coherence vectors). Al- though they can perform better, their computational complexity is prohibitive and they require parameter setting. We propose to upgrade the first order color distribution (color histogram) by em- bedding for each color additional infor- mation about its perceptual or statisti- cal relevance. Such information is ob- tained by using local activity measures such as the Laplacian, the entropy and others. Histograms computed on win- dows and combined by different ways of accumulation improve the informa- tion on geometric repartition of colors. We prove that the new color distribu- tion family is compact, robust and easy to compute and provides a superior re- trieval performance, independent with respect to the color representation. |
Year | Venue | Keywords |
|---|---|---|
2001 | EUSFLAT Conf. | first order,computational complexity,higher order,image retrieval,color histogram |
Field | DocType | Citations |
Computer vision,HSL and HSV,Color histogram,Pattern recognition,Image texture,Computer science,Color depth,Color balance,Artificial intelligence,Histogram equalization,Color normalization,Color image | Conference | 2 |
PageRank | References | Authors |
0.51 | 5 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Nozha Boujemaa | 1 | 1231 | 96.30 |
| Sabri Boughorbel | 2 | 127 | 15.32 |
| Constantin Vertan | 3 | 162 | 25.25 |