Inline quality control for rapid archival workflows

Object-Level Target Checker

Automated color validation for NYPL's cultural heritage imaging studios, Digital Imaging Services (DIS), bridging high-level archiving standards with practical production needs.

The Object-Level Target Checker in the Wizard app is an automation-grade software module engineered to handle background image verification for high-throughput cultural heritage and digitization studios. Built directly into the Wizard automation ecosystem, the Target Checker processes entire folders of captures to verify technical performance without forcing technicians to manually sample pixels in external software.

By combining profile-agnostic computer vision with standard 8-bit imaging language, it bridges strict administrative oversight with frictionless operator execution.

Golden Thread color target with sampled chip areas.

AOI Inspector Preview

The checker maps representative chip areas, the central physical gap, and neutral witness regions directly onto the target. The interactive inspector mirrors the generated QC view: move over the image to read RGB, luminance, and browser-computed CIELAB values.

Open Interactive Inspector

Reference LAB Map

Chip values are arranged in the same left run, center gap, and right run used by the AOI overlay. Chip 10 is the white luminance reference. Chip 21 is the shadow luminance reference.

01Dark BrownL 38.76
a 13.81
b 14.69
02PeachL 65.15
a 19.21
b 17.92
03Blue-GreyL 49.61
a -4.20
b -21.33
04Olive GreenL 43.54
a -12.89
b 22.66
05LilacL 55.52
a 8.78
b -24.31
06SeafoamL 70.42
a -32.39
b -0.48
07OrangeL 63.13
a 35.43
b 57.84
08Deep BlueL 40.08
a 10.25
b -44.77
09RedL 51.75
a 47.36
b 16.93
10White L* RefL 95.34
a -0.90
b 1.90
11Grey 0.09L 92.09
a -0.92
b 1.46
12Grey 0.16L 86.92
a -1.12
b 0.97
13Grey 0.22L 82.37
a -1.12
b 0.56
14Grey 0.36L 72.17
a -1.05
b -0.04
15Grey 0.52L 62.32
a -1.10
b -0.01
16Grey 0.75L 49.61
a -1.29
b -0.10
17Grey 0.98L 38.89
a -0.23
b -0.48
18Grey 1.24L 28.60
a -1.09
b 0.07
19Grey 1.64L 17.97
a 0.04
b 0.09
20Grey 1.96L 9.50
a 0.45
b 0.25
21Shadow L* RefL 4.33
a 0.32
b -0.47
22PurpleL 30.32
a 22.13
b -19.02
23LimeL 72.50
a -22.92
b 56.08
24GoldL 72.10
a 19.51
b 67.85
25NavyL 29.51
a 13.42
b -47.69
26GreenL 55.60
a -38.46
b 32.19
27CrimsonL 43.38
a 50.74
b 29.13
28Bright YellowL 82.02
a 3.28
b 78.75
29MagentaL 52.85
a 49.90
b -12.86
30CyanL 50.86
a -27.78
b -27.68
Archival object capture with object-level color target below the image.

Photograph of the photograph courtesy of NYPL's Digital Collections.

01

Dual-Track Evaluation Architecture

Natively performs all colorimetric calculations in profile-agnostic CIELAB space to guarantee absolute metric consistency across sRGB, AdobeRGB, or ProPhoto captures. Simultaneously, it maps and reports operator results back to traditional 8-bit RGB counts from 0-255, matching studio standards.

02

Gap-Aware Mid-Gray Balance Verification

Explicitly engineered to accommodate the physical layout of the Golden Thread ISO Object-Level Target by Don Williams. The module isolates and samples gray patches on both sides of the target center, monitoring and flagging image color neutrality anomalies.

03

Intelligent Upright Stabilization and OCR Witnessing

Utilizes rapid grayscale brightness mapping and localized Apple Vision Framework OCR to detect target orientation natively. If a target is fed into the system rotated or upside down, the script stabilizes the coordinates, computes horizontal physical grid alignment, and executes structural sampling.

04

Hardened Vector Processing for Apple Silicon

Optimized for high-precision workflows on modern architectures. Explicit float32 casting and rigorous range clipping strip away sub-black or super-white digital noise pixel artifacts to support rapid processing without math-multiplication anomalies.

Technical Specifications and Metrics

Evaluation Metric Engine User Scale Default Tolerance
White Point CIELAB Lightness L* 8-bit RGB counts +/- 3.0 RGB counts
Neutrality Euclidean chroma distance Direct LAB chroma delta Custom, <= 1.0 max delta
Black Point CIELAB Lightness L* 8-bit RGB counts +/- 3.0 RGB counts, auto-skip at 0

Studio Workflow

  1. Ingest and stabilize: As folders of images are uploaded to the Target Checker tool, the engine first checks orientation thresholds and normalizes bit depth.
  2. Isolate and sample: Bounding boxes are generated across a 37.0 total unit horizontal scale to account for side spacers and the target center.
  3. Audit and report: Validation results are output directly into a scannable HTML quality control manifest, showing green passes for optimal images and clear error tags such as W:245 or N(L:1.4) for ease of correction.