Name
Abstract | ||
|---|---|---|
In many applications, it is desirable to reduce the number of color and texture features to a sufficient minimum. Computational complexity and storage cost are some of the obvious reasons for this requirement. A related reason is that although two features may carry good discriminatory information when treated separately, there is little gain if they are combined together in a feature vector, because of a possible high mutual correlation between them. Thus, complexity increases without much gain. The major task of this paper is to address the problem of selecting the most important features for a given textured color image so as to maintain the optimal number of color and texture characteristics.The basic approach explored in this paper is based on the discrete Karhunen-Love Transform (KLT). The reason behind the selection of KLT-based texture and color features is that an appropriately chosen transform can exploit and remove information redundancies, which usually exist in a wide range of color and texture scenes obtained by measuring devices. If the transform chosen is subjected to appropriate constraints, maximum information can be preserved in the output with reduced dimension, and without nuances that do not exist in the original input samples. This is crucial if any subsequent feature-analyzing function is to be able to produce meaningful results. Another reason is that KLT seems closely related to early visual pathway of primates. |
Year | Venue | Keywords |
|---|---|---|
2000 | PICS 2000: IMAGE PROCESSING, IMAGE QUALITY, IMAGE CAPTURE, SYSTEMS CONFERENCE, PROCEEDINGS | karhunen loeve transform |
Field | DocType | Citations |
Pattern recognition,Karhunen–Loève theorem,Computer science,Artificial intelligence,Feature generation | Conference | 0 |
PageRank | References | Authors |
0.34 | 0 | 2 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Mehmet Celenk | 1 | 170 | 37.28 |
| Ivan Chang | 2 | 1 | 1.30 |