OrientationJ Test Images¶

16 grayscale 2D images for testing and benchmarking orientation analysis: seven synthetic images — four of them with an analytic ground-truth orientation — eight real images, and one artificial fiber phantom.

The folder is organized as:

images/ — the 16 source images (TIFF);
masks/ — a binary mask per image (uint8, 255 = meaningful structures);
results/ — this montage, the raw feature maps of a selection of images (orientation, coherency, energy as float32 TIFF, regenerate any with make_maps.py), and one overview panel per image: original, mask, orientation, coherency, energy, color survey, orientation distribution inside the mask, and vector field — computed with the OrientationJ Python port at the plugin defaults (cubic-spline gradient, σ = 1);
code/ — synthetic_images_2D.ipynb generates the synthetic images, masks_2D.ipynb generates the masks, make_montage.py renders the montage above, and make_maps.py the downloadable feature maps in results/maps/; the patterns themselves are drawn by synthetic_images_2D.py, which the notebook imports. The overview panels come from make_gallery.py in the Python port.

Click a panel for full size.

synthetic_chirp_1024¶

Download: image · orientation · coherency · energy

1024 × 1024, float32, values in [0, 1] — synthetic: radial chirp with a frequency sweep; the tangential ground-truth orientation is known at every pixel.

synthetic_rings_dither_512¶

Download: image

512 × 512, float32, values in [0, 1] — synthetic: thin concentric rings with a small dither (σ = 10⁻⁴) that avoids degenerate-tensor spikes; exact tangential truth.

synthetic_wave_512¶

Download: image

512 × 512, float32, values in [0, 1] — synthetic: fringes at exactly +60° and −30°, two scales at once.

synthetic_nematic_512¶

Download: image

512 × 512, float32, values in [0, 1] — synthetic: nematic-like texture with a dominant direction.

synthetic_filaments_512¶

Download: image

512 × 512, float32, values in [0, 1] — synthetic: random curved filaments.

synthetic_spiral_512¶

Download: image

512 × 512, float32, values in [0, 1] — synthetic: spiral.

synthetic_noise_512¶

Download: image

512 × 512, float32, values in [0, 1] — synthetic: isotropic noise; any measured anisotropy is bias.

collagen¶

Download: image · orientation · coherency · energy

512 × 512, uint8, values in [2, 255] — real: collagen fibers.

cell_aemisegger¶

Download: image · orientation · coherency · energy

728 × 728, uint8, values in [0, 252] — real: fluorescence cell, actin fibers.

dendrochronology¶

Download: image

512 × 512, uint8, values in [0, 255] — real: wood section, growth rings.

fibronectin_arafat_plos2025¶

Download: image

1024 × 1024, uint8, values in [0, 255] — real: fibronectin network (Arafat et al., PLOS ONE 2025).

fibrous_tissues_fibero2024¶

Download: image

2048 × 2048, uint16, values in [87, 4094] — real: fibrous tissue (FiberO, 2024).

fiji_directionality_montage¶

Download: image

1536 × 1536, uint16, values in [3089, 65535] — real: montage from the Fiji Directionality documentation.

nanofiber_fiji_diameterj¶

Download: image

512 × 512, uint8, values in [0, 255] — real: SEM nanofibers (DiameterJ sample).

polymer_slice_quanfima2018¶

Download: image

600 × 600, float32, values in [−0.0029, 0.0030] — real: polymer slice (quanfima, 2018).

z_artificial_fibers¶

Download: image

683 × 512, uint8, values in [0, 225] — artificial: fiber phantom (drawn, not generated by the synthetic-image notebook).

External datasets with ground truth¶

For validation beyond a smoke test, published datasets are also useful:

Hotaling et al. (2015), Data in Brief — synthetic fiber images drawn at specified angles, SEM images of steel wire with known diameters, electrospun nanofiber SEM images, their segmentations, and reference orientation histograms. doi:10.1016/j.dib.2015.07.012
Arafat et al. (2025), PLOS ONE — four extracellular-matrix datasets with manually delineated ground truth, used to benchmark seven fiber-tracing algorithms. doi:10.1371/journal.pone.0320006