Skip to content

CHANGELOG.md

Latest commit

 

History

History
121 lines (101 loc) · 5.87 KB

CHANGELOG.md

File metadata and controls

121 lines (101 loc) · 5.87 KB
Raw

Changelog

All notable changes to MultiClean are documented here.

Unreleased

0.5.0 - 2026-08-14

Fixed

  • Edge smoothing no longer translates the output by one pixel down and to the right when smooth_edge_size is even. cv2.morphologyEx(MORPH_OPEN) applies a single anchor to both the erosion and the dilation, which is only correct when that anchor coincides with the structuring element's centre of symmetry — true for odd kernel sizes, false for even ones. The result was a shifted opening that could also add pixels to a class rather than only removing them. The erosion and dilation are now run separately with the dilation anchored at the reflection of the erosion's anchor, giving a true opening at every size.

    This changes output for even smooth_edge_size, including the default of 2. Output for odd values is bit-identical to previous releases. If you have cached results produced with an even smooth_edge_size, regenerate them or expect a one-pixel offset against new output.

  • The source distribution no longer ships the example notebooks, their sample rasters, or the README artwork. setuptools-scm hands the sdist every git-tracked file, so these were swept in automatically — 4.9 MB of a 7.2 MB tarball, none of it read by anything in the package. A MANIFEST.in now prunes them, taking the sdist to 3.9 MB. The notebooks remain on GitHub, where the README links them. Wheels were never affected and are unchanged.

0.4.0 - 2026-07-28

Changed

  • Switched the OpenCV runtime dependency from opencv-python to opencv-python-headless. MultiClean only uses morphologyEx, connectedComponentsWithStats and distanceTransformWithLabels — no GUI or video I/O — so the GUI build was pulling in shared libraries that are absent from slim container and CI images, where plain opencv-python fails at import with libGL.so.1: cannot open shared object file. Both distributions provide the same cv2 module, so if your environment needs the full opencv-python build for other work, install it explicitly; do not rely on both being present at once.
  • Raised the OpenCV floor to >=4.10. opencv-python builds before 4.10.0.84 are compiled against NumPy 1.x and fail at import when paired with NumPy 2.x, which the previous >=4.0 floor allowed a resolver to do.

Added

  • Declared numpy>=1.21 as an explicit runtime dependency. It was always imported directly and appears in the public type signatures, but was only installed transitively via OpenCV.

0.3.1 - 2026-07-21

Fixed

  • clean_array no longer raises KeyError when class_values names a class that does not occur in the input array. Requested classes absent from the array are now ignored, matching the behaviour that already applied when smooth_edge_size=0. This mainly affected tiled processing, where a single fixed class list is reused across tiles whose contents vary — the failure depended on the data, so it surfaced intermittently.

0.3.0 - 2026-05-02

Changed

  • Performance. clean_array is substantially faster on multi-class inputs. On a 15669×18633 / 147-class land-use raster, end-to-end runtime dropped from ~85 s to ~40 s. On the 8011×7901 / 4-class Landsat cloud-and-shadow example, runtime dropped from ~2.5 s to ~1.1 s. Wins came from:
    • Replacing the float32 smoothed-labels buffer with a uint8/uint16 class-code array (selected automatically based on class count). The per-class equality scan is 2-4× cheaper in memory bandwidth.
    • Combining per-class small-island masks in flight instead of accumulating all K of them first.
    • Filling invalid pixels in place rather than allocating a copy.
    • Replacing scipy.ndimage.distance_transform_edt with cv2.distanceTransformWithLabels for the nearest-valid fill (~3.4× faster on the fill stage). Both algorithms produce mathematically equivalent output (the same minimum L2 distance); they differ only in which equidistant source pixel wins a tie.
  • dtype preservation. The output now strictly matches the input dtype. Previously the pipeline routed everything through float32 internally, which silently downcast float64 inputs and rounded int32 values larger than 2²⁴ (and int64 values larger than 2⁵³).

Fixed

  • All-NaN float input with fill_nan=True now deterministically returns an all-NaN array. The previous code relied on whatever value np.empty happened to leave in the sentinel slot.
  • Large integer class values (int32 > 2²⁴, int64 > 2⁵³) are now preserved bit-exactly, instead of being silently rounded by the internal float32 round-trip.

Removed

  • Dropped the scipy runtime dependency. cv2 (already a runtime dependency) now handles the distance-transform fill.

0.2.0 - 2025-09-03

Added

  • fill_nan option on clean_array: when True, NaN values in float input arrays are filled from the nearest valid pixel rather than preserved as nodata.

0.1.0 - 2025-09-02

Added

  • Initial public release.
  • clean_array API for morphological cleaning of multi-class 2D arrays: per-class edge smoothing (morphological opening), per-class small-island removal (connected components), and gap filling using nearest-valid via Euclidean distance transform.
  • Documentation: README, two example notebooks (land use, cloud shadow), and a Google Colab tutorial notebook.