AstroBotany calibration · design proposal

ABCD v2: a bigger calibration card whose colours mean something to plant and soil scientists

· Munsell-defined colour chips · self-identifying ArUco corner markers (IDs )

ABCD v2 card
Richard Barker · for discussion with consultantsdr-richard-barker.github.io/ABCD_v2
Why a v2

v1 was too small, and its colours didn't map to anything we measure

v1, v1.5 and v2 to scale
Each card scaled from its own printed ruler2
Card anatomy

What is on the card

Annotated card
Geometry from card/abcd_v2_chips.json3
Colour system

Every chip is a Munsell notation; the printed colour is computed, then measured

1 · NotationSoil-chart chips, Plant Tissue book hues, calibration hues.
2 · RenotationMunsell's published data → CIE xyY.
3 · Print + screenLab D50 for print and measuring; sRGB for the file.
4 · Gamut gateMust print (ΔE2000 ≤ 3) and fit sRGB, else chroma steps down.
5 · MeasureSpectrophotometer readings per batch replace the design values.

chips reduced by the gate:

largest remaining ΔE2000 between design and printable colour

±

/255 per channel vs an independent Munsell viewer, chips

The scanned soil-chart PDF supplied notations and names only; its pixel colours depend on the scanner and were not used.

results/generator_log.txt · results/werth_crosscheck.txt4
Rows A + neutral ramp

Calibration: anchors, a hue circle and a 12-step grey ramp

Gives the colour-correction fit something to grip on in every hue direction, plus white, black and N5 (≈ 18 % grey) for exposure and white balance.

Calibration hues: highest chroma that prints5
Rows B–C

Plant tissue: leaf health, pigments, senescence and roots

Every hue is a page of the Munsell Plant Tissue Color Book. Value/chroma are design targets: we'd like consultants to confirm or swap them for actual book chips.

Hue list: Munsell Plant Tissue Color Book (vendor listing)6
Rows D–E

Soil: chips taken from the Munsell soil-chart pages, with their printed names

Hues 10R to 2.5Y. The low-chroma greys span grey-regolith darkness. No chip claims to match a specific simulant or Apollo sample; targeting simulants needs published Munsell values for them.

Source: Munsell Soil-Color Charts (soils.uga.edu scan)7
Corner markers

Standard ArUco markers: the card identifies itself

  • v1 already used OpenCV DICT_4X4_50 IDs .
  • v2 uses IDs , the best-separated block of four: at least 5 bits from any v1 marker, 8 from their own rotations.
  • Any ArUco reader (OpenCV, arucogen, our app) knows which card, which corner, and which way up.
  • Verified with OpenCV's own detector on both cards.
results/aruco_ids.txt8
Does the app find it?

Detection on synthetic photos, before and after marker-ID checking

results/detection_before.txt · results/detection_after.txt9
Printing, measuring + one v1 finding

From design values to measured references

  1. Print matte if possible.
  2. Measure ≥ 3 stickers per batch, every chip, with a spectrophotometer (D50 / 2°).
  3. Flag chips that vary by more than ΔE2000 2 between stickers.
  4. Measured Lab values replace design values in the app.
  5. Each print run gets its own batch number on the card.
v1 scale finding. Against its own printed ruler, v1's markers are not 4.3 cm apart in both directions as the app assumes:

To confirm on real photos before changing the app.

results/v1_geometry.txt10
Decisions needed

Open questions for consultants

Plant shades. Right tissue classes and shades? Swap for real Plant Tissue book chips?
Soil targets. Should chips match specific regolith simulants? Which, with what published values?
Format. Sticker or laminated card? Is 14 × 9 cm right for the imaging setups we support?
Markers + numbering. IDs 0–3 for v2; how should future versions and batches be numbered?
v1 scale. Confirm the 4.3 cm discrepancy on real v1 photos, then correct the app.
Measurement. Who measures printed batches, and with which instrument?

Comment on the GitHub issues: github.com/dr-richard-barker/ABCD_v2/issues

Site: dr-richard-barker.github.io/ABCD_v211