Releases: clij/clij2
Releases · clij/clij2
2.3.0.6
New: Deployment to sonatype nexus
2.2.1.0
Bugfix / backwards compatibility breaking change:
- connectedComponentsLabeling Diamond yielded Box results (thanks to pdd2110 for reporting!)
2.2.0.25
New features
- CLIJ2 now checks if extensions are installed correctly and warns via stdout if not. Thanks to Laurent Thomas for suggesting!
Bug fixes:
- Default output image type of standard-deviation and sum projections is 32-bit float now. Thanks to Christian Tischer for reporting this bug.
- 2D images were pasted in the wrong slice when the destination image is 3D. Thanks to Nik Cordes for reporting this bug!
2.2.0.19
Bugfixes
- statisticsOfLabelledPixels and statisticsOfBackgroundAndLabelledPixels had wrong results for Bounding_box_x, Bounding_box_width, Bounding_box_y, Bounding_box_height, Bounding_box_z, Bounding_box_depth and minimum_intensity. (Thanks to Jim Rowe for reporting this bug)
- maximum_images of 2D images with different pixel type crashed
2.2.0.14
Bugfixes
- neighbors_of_neighbors ignored some neighbors
- centroids_of_labels and centroids_of_background_and_labels created an empty row in the results table
- exclude_labels_with_values_out_of_range and exclude_labels_with_values_within_range also support processing parametric maps now
- output of voronoi_labeling should be of type float per default
2.2.0.12
Bug fixes
- Corrected binary true/false determination in touch_matrix_to_mesh. This bug fix is only relavant for meshes with negative values on deges.
- Corrected average distance to touching neighbors determination; the background is not taken into account anymore.
- Paste2D crashed when auto-generating output images
2.2.0.11
Bugfixes
- Connected Components Labeling may have missed a single individual binary pixel in the top left of the image (first pixel in the array)
- Added a workaround for Intel Kabylake GPUs on Linux. See troubleshooting section if you use this combination of OS/GPU.
- NotEqual was broken in 3D
- Equal was broken in 3D
2.2.0.9
Bugfixes:
- SumOfAllPixels delivered wrong results in case of very large images (Thanks to Lennard Voortman for reporting)
- Workaround in AverageDistanceOfTouchingNeighbors: If a label doesn't touch neighbors, this distance = 0. This bugfix resolves some downstream issues and might be reverted later on (Thanks to LPUoO for reporting.)
2.2.0.7
New Features
- Added numpy-style documentation for parameters (Thanks to Juan Nunez-Iglesias for suggesting)
- Changed some parameters (names) to match snake_case
Bugfixes
- StandardDeviationZProjection didn't work on POCL (CPU-OpenCL, thanks to Talley Lambert for setting up the CI which found that bug)
- LabelSpots may have had wrong results when reusing/overwriting old buffers
- NonZeroOverwrite and OnlyZeroOverwrite functions didn't expose the correct API to Java/Jython/Groovy #23 (Thanks to Nicolas Chiaruttini for reporting)
2.2.0.1
Bug fix
- Change the pre-sign of bottom hat filter results (Thanks Nik Cordes for reporting)