DDPM-SegFormer: Highly refined feature land use and land cover segmentation with a fused denoising diffusion probabilistic model and transformer.
Junfu FanZongwen ShiZhoupeng RenYuke ZhouMin JiPublished in: Int. J. Appl. Earth Obs. Geoinformation (2024)
Keyphrases
- land cover
- denoising
- probabilistic model
- remote sensing
- change detection
- multispectral
- remote sensing images
- land cover classification
- satellite images
- supervised classification
- remote sensing imagery
- nonlinear diffusion
- remote sensing data
- geographic information systems
- image segmentation
- image denoising
- image processing
- satellite data
- remotely sensed data
- level set
- multiscale
- climate change
- urban growth
- image analysis
- anisotropic diffusion
- edge detection
- image registration
- environmental variables
- data sets
- hyperspectral
- bayesian networks
- spatial data
- multiresolution
- learning algorithm
- image fusion
- image regions
- cellular automata
- moving objects
- feature vectors
- high resolution