v0.3.6

• New features

New open() function: Added a top-level open() function as an alias to the Geoimage constructor. This aligns with standard Python conventions (like rasterio or PIL). You can now load an image using:

New clip() function: A method to clip pixel values within a specified range.

• Bug fixes

Filtering: Fixed a minor bug that occurred during generic filtering when using large filter kernels.

v0.3.5

Fix some bugs in distribution files and metadata

v0.3.3

Fix minor bug in metadata

v0.3.2

Improvements:

• im2tiles: Output Folder Management (Breaking Change)

The default behavior of the im2tiles function has been changed. If the output folder already exists, the function will now delete by default any existing snippets within it before generating new ones.

To keep the existing snippets during execution, use the new argument: reset=False.

• astype: New In-Place Modification Option (inplace)

A new inplace parameter has been added to the astype function.

If inplace=True, the data type modification will be performed directly on the current object. By default, inplace=False, which returns a new instance with the modified data type.

v0.3.1

Bug fixes & improvement:

• Naming: Fixed an issue where band names could be reset when manipulating images that already had saved names.

• im2tiles: Now clears any existing snippets in a folder before saving new ones, preventing leftover files from previous runs.

v0.3.0

This new version introduces several new features and improvements focused on dimensionality reduction, visualization, and overall consistency across the library.

New features

• Dimensionality reduction: Added support for PCA, t-SNE, and LLE methods for feature extraction and visualization.

• Visualization: Added a zoom parameter to all visualization functions (.visu(), .colorcomp(), .plot_spectra()), allowing users to focus on specific regions of interest.

Improvements

• Crop function harmonization: The crop() function has been standardized.
  It now accepts either:

  • a pixel-based area definition: area=((deb_row, end_row), (deb_col, end_col)), or

  • geographic coordinates of the north-west (lat1, lon1) and south-east (lat2, lon2) corners.

• Consistent documentation: Updated comments and docstrings to ensure consistency with the predict() function and other ML-related methods.

Bug fixes

• Fixed a bug occurring when opening windowed versions of images that caused incorrect band naming.

0.2.2

Correct minor bug in plotting

0.2.1

• Added a new boolean test: image.isnan() to check for NaN values in an image.

• Renamed resampling() to resample().

  • Both functions remain available in this version, but resampling() is deprecated and will be removed in a future release.

• Renamed apply_ML_model() to predict().

  • Both functions remain available in this version, but apply_ML_model() is deprecated and will be removed in a future release.

0.2.0

This release introduces several new features (custom band names persistence, metadata-only loading, partial image reading, improved lat/lon visualization, and warnings for multi-band stacks) while remaining fully backward compatible.
The version has therefore been bumped from 0.1.4 to 0.2.0.

Here are the main changes

• User-defined band names

  • Band names set by the user via im.change_names are now automatically saved with im.save and reloaded with rastereasy.Geoimage.

• Metadata-only loading

  • You can now load only the metadata without reading the full image using meta_only=True: im = rastereasy.Geoimage('myimage.tif', meta_only=True)

• Partial image reading (window or area)

  • It is now possible to read a specific part of the image with the area parameter:

  • By indices: area=((start_row, end_row), (start_col, end_col))

  • By geographic coordinates: area=((lon1, lon2), (lat1, lat2)) with extent='latlon'

• Warning for multi-band images in files2stack

  • When using rastereasy.files2stack with images containing multiple bands, a warning is displayed to inform the user.

• Minor bugs in visualization for latitude/longitude coordinates has also been fixed.

0.1.4

Add useful functions for ML