An Adaptive Change Threshold Selection Method Based on Land Cover Posterior Probability and Spatial Neighborhood Information
Change threshold selection (CTS) plays an important role in land cover change detection. The traditional CTS methods are mainly based on the information contained in grayscale histogram distributions or pixel neighborhoods. However, land cover is highly spatially heterogeneous, and changes in differ...
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
IEEE
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/cfd69ca4061d40dd81549050a3c8e9b0 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Change threshold selection (CTS) plays an important role in land cover change detection. The traditional CTS methods are mainly based on the information contained in grayscale histogram distributions or pixel neighborhoods. However, land cover is highly spatially heterogeneous, and changes in different land cover types are characterized by different magnitudes. Unfortunately, few CTS studies have considered the effects of both land cover type and spatial heterogeneity on CTS, potentially leading to false alarms or missed alarms. To address this challenge, we propose an adaptive CTS method based on land cover posterior probability and spatial neighborhood information (LCSN). First, the posterior probability of the change magnitude in each land cover type is calculated according to a Bayesian criterion to integrate the land cover type information. Second, the posterior probability is calculated using a bilateral filtering method to construct the spatial surface based on the land cover type and spatial neighborhood information. Finally, the degree of difference between the spatial surface and the change magnitude map is taken as the final threshold. The proposed LCSN method is verified with Landsat 8-Operational Land Imager images and IKONOS images. The experimental results show that the LCSN method is effective in reducing the pseudo changes and identifying changes in land cover types with low grayscale values in the corresponding change magnitude maps. |
|---|