| Autor: |
Juliusz L. Kulikowski, D. Wierzbicka, M. Przytulska |
| Rok vydání: |
2007 |
| Předmět: |
|
| Zdroj: |
World Congress on Medical Physics and Biomedical Engineering 2006 ISBN: 9783540368397 |
| DOI: |
10.1007/978-3-540-36841-0_575 |
| Popis: |
It is presented a method of texture recognition based on the analysis of multilevel morphological spectra of biomedical images under consideration. The n-th level morphological spectrum of the image is defined as a matrix whose elements have been obtained due to an rearrangement of the results of four standard algebraic operations performed on the components of the former, (n-1)-th level spectrum. The higher is the spectrum-level, the larger are the morphological structures being taken into consideration in texture analysis; the lowest (0-th) level spectrum corresponds to the bitmap of the original image. It is shown that the original image can be reconstructed on the basis of its morphological spectrum of any given level. The set of spectra can be represented in the form of a tree. Morphological spectra corresponding to selected branches of this tree are characteristic for certain classes of textures and, as such, they can be taken into account for effective discrimination and recognition of the given classes of textures. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
