Takes a set of classified raster layers with probabilities,
whose metadata is created by sits_cube,
and applies a Bayesian smoothing function.
Usage
sits_smooth(cube, ...)
# S3 method for class 'probs_cube'
sits_smooth(
cube,
...,
window_size = 9L,
neigh_fraction = 0.5,
smoothness = 20,
exclusion_mask = NULL,
memsize = 4L,
multicores = 2L,
output_dir,
version = "v1",
progress = TRUE
)
# S3 method for class 'probs_vector_cube'
sits_smooth(cube, ...)
# S3 method for class 'raster_cube'
sits_smooth(cube, ...)
# S3 method for class 'derived_cube'
sits_smooth(cube, ...)
# Default S3 method
sits_smooth(cube, ...)Arguments
- cube
Probability data cube.
- ...
Other parameters for specific functions.
- window_size
Size of the neighborhood (integer, min = 3, max = 21)
- neigh_fraction
Fraction of neighbors with high probabilities to be used in Bayesian inference. (numeric, min = 0.1, max = 1.0)
- smoothness
Estimated variance of logit of class probabilities (Bayesian smoothing parameter) (integer vector or scalar, min = 1, max = 200).
- exclusion_mask
Areas to be excluded from the classification process. It can be defined as a sf object or a shapefile.
- memsize
Memory available for classification in GB (integer, min = 1, max = 16384).
- multicores
Number of cores to be used for classification (integer, min = 1, max = 2048).
- output_dir
Valid directory for output file. (character vector of length 1).
- version
Version of the output (character vector of length 1).
- progress
Check progress bar?
Note
The main sits classification workflow has the following steps:
sits_cube: selects a ARD image collection from a cloud provider.sits_cube_copy: copies an ARD image collection from a cloud provider to a local directory for faster processing.sits_regularize: create a regular data cube from an ARD image collection.sits_apply: create new indices by combining bands of a regular data cube (optional).sits_get_data: extract time series from a regular data cube based on user-provided labelled samples.sits_train: train a machine learning model based on image time series.sits_classify: classify a data cube using a machine learning model and obtain a probability cube.sits_smooth: post-process a probability cube using a spatial smoother to remove outliers and increase spatial consistency.sits_label_classification: produce a classified map by selecting the label with the highest probability from a smoothed cube.
Machine learning algorithms rely on training samples that are derived from “pure” pixels, hand-picked by users to represent the desired output classes. Given the presence of mixed pixels in images regardless of resolution, and the considerable data variability within each class, these classifiers often produce results with misclassified pixels.
Post-processing the results of sits_classify
using sits_smooth reduces salt-and-pepper and border effects.
By minimizing noise, sits_smooth brings a significant gain
in the overall accuracy and interpretability of the final output.
References
Gilberto Camara, Renato Assunção, Alexandre Carvalho, Rolf Simões, Felipe Souza, Felipe Carlos, Anielli Souza, Ana Rorato, Ana Paula Dal’Asta, “Bayesian inference for post-processing of remote sensing image classification”. Remote Sensing, 16(23), 4572, 2024. doi:10.3390/rs16234572 .
Examples
if (sits_run_examples()) {
# create am xgboost model
xgb_model <- sits_train(samples_modis_ndvi, sits_xgboost())
# create a data cube from local files
data_dir <- system.file("extdata/raster/mod13q1", package = "sits")
cube <- sits_cube(
source = "BDC",
collection = "MOD13Q1-6.1",
data_dir = data_dir
)
# classify a data cube
probs_cube <- sits_classify(
data = cube, ml_model = xgb_model, output_dir = tempdir()
)
# plot the probability cube
plot(probs_cube)
# smooth the probability cube using Bayesian statistics
bayes_cube <- sits_smooth(probs_cube, output_dir = tempdir())
# plot the smoothed cube
plot(bayes_cube)
# label the probability cube
label_cube <- sits_label_classification(
bayes_cube,
output_dir = tempdir()
)
# plot the labelled cube
plot(label_cube)
}