Incorporating edge information into best merge region-growing segmentation
| Autor: | Edoardo Pasolli, James C. Tilton |
|---|---|
| Přispěvatelé: | Tilton, J. C., Pasolli, E. |
| Rok vydání: | 2014 |
| Předmět: |
Segmentation-based object categorization
Computer science business.industry Scale-space segmentation Image processing Pattern recognition image edge detection Image segmentation Edge detection Image texture Region growing Artificial intelligence Range segmentation business image analysi image segmentation Feature detection (computer vision) |
| Zdroj: | IGARSS |
| DOI: | 10.1109/igarss.2014.6947591 |
| Popis: | We have previously developed a best merge region-growing approach that integrates nonadjacent region object aggregation with the neighboring region merge process usually employed in region growing segmentation approaches. This approach has been named HSeg, because it provides a hierarchical set of image segmentation results. Up to this point, HSeg considered only global region feature information in the region growing decision process. We present here three new versions of HSeg that include local edge information into the region growing decision process at different levels of rigor. We then compare the effectiveness and processing times of these new versions HSeg with each other and with the original version of HSeg. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|