Abstract | ||
---|---|---|
The selection of suitable features is the most critical part of any classification process. In the area of textural classification, one way to determine if the features are suitable is to synthesize an image which has the given textural features. An algorithm for synthesizing textures that have a set of given cooccurrence matrices is presented, and it is shown that these synthetic images do indeed match their real counterparts very closely. The successful synthesis of textures motivates the use of co-occurrences as features for texture classification. A new algorithm for classifying textures based on co-occurrence feature vectors that are modelled as multinomial density functions is presented. |
Year | DOI | Venue |
---|---|---|
1995 | 10.1016/0097-8493(94)00119-J | COMPUTERS & GRAPHICS |
DocType | Volume | Issue |
Journal | 19 | 1 |
ISSN | Citations | PageRank |
0097-8493 | 11 | 2.30 |
References | Authors | |
6 | 1 |
Name | Order | Citations | PageRank |
---|---|---|---|
Gabriele Lohmann | 1 | 356 | 40.09 |