figure-duplicate-audit

name	figure-duplicate-audit
description	Audit scientific figures for duplicated, reused, transformed, or uninformative image panels. Use for 图片查重, 论文图像重复, PDF figure 审核, when the user attaches a PDF, asks to review selected PDF pages, or tags a directory containing manuscript input images. For PDFs, extract large embedded figure images before splitting them into panels; for directories, preserve originals and split every composite image directly. Produces a reviewed panel manifest, all-pairs candidate table, visual evidence, coverage accounting, and a cautious integrity report.
compatibility	Requires Python 3.10+, Pillow, NumPy, pypdf, and pypdfium2. OpenCV with SIFT support is required for the full feature-matching pass.

Figure duplicate audit

Audit at the smallest meaningful experimental-image unit. Hashes and feature matches find candidates; they do not establish misconduct or even duplication on their own.

Inputs and workspace

Accept either one PDF or one directory of manuscript images. Resolve tagged or attached paths before running anything. Ask for a page range only when the user did not specify one and scanning the whole PDF would materially change scope.

Create a new analysis directory such as analysis/figure-audit-YYYYMMDD-HHMM. Never modify source images, overwrite a prior audit, or silently omit an unreadable file.

Locate this skill's scripts/audit_figures.py from the resource paths returned by use_skill. If imports fail, load local-env-setup, create a project-local environment, and install the packages named in compatibility. Do not continue with the hash-only fallback when the user requested a strict or exhaustive review.

1. Prepare sources

For a PDF:

python audit_figures.py prepare --input PAPER.pdf --output AUDIT_DIR --pages "1-40,49-54"

The script extracts qualifying embedded raster images first. It renders a page only when no large embedded image is available and the page looks like a figure page, or when --render-fallback all is explicitly used. Review sources.json, skipped.json, and sources-contact-sheet.png; confirm that every requested figure is represented. A page render still contains captions and page furniture, so crop the figure before panel splitting.

For a directory:

python audit_figures.py prepare --input FIGURE_DIR --output AUDIT_DIR

The script recursively inventories supported images, normalizes EXIF orientation into audit copies, and records hashes and original paths. It does not alter the directory.

2. Verify panel boundaries

prepare writes conservative panel proposals to panels.json. They are only proposals. View every source at full resolution and edit the manifest until:

• every data-bearing photograph, microscopy field, histology tile, plate, wound, gel/blot region, or other experimental image has its own box;
• repeated grids are split into individual experimental units, with stable labels such as Fig2-D-r1-c2 rather than anonymous indices;
• labels, legends, scale bars, and axes are not mistaken for independent data panels;
• adjacent boxes do not overlap accidentally;
• expected derivatives share a derivation_group (for example raw channels and merge, overview and inset, or known longitudinal views);
• kind records the modality when known (microscopy, histology, western-blot, gel, plate, wound, ivis, chart, or schematic).

Run:

python audit_figures.py materialize --workspace AUDIT_DIR

Inspect panels-contact-sheet.png immediately. Fix bad crops and rerun. Do not scan until manifest-warnings.json has no unexplained out-of-bounds, duplicate-ID, or overlapping-box warning. Preserve parent/context crops when a tighter data-only crop is needed for matching.

3. Run all-pairs screening

python audit_figures.py scan --workspace AUDIT_DIR --features required

The scan combines exact pixel hashes, perceptual hashes, normalized correlation, and SIFT + RANSAC geometry. It writes candidates.csv, candidates.json, quality-flags.csv, and scan-summary.json. Review every candidate, not only the first page of the table. Re-scan after any crop change.

Automatic scores are triage signals. Repeated labels, axes, membrane grids, plate rims, scale bars, and regular tissue texture often produce false matches. Conversely, different crops, contrast changes, rotation, mirroring, or recompression can hide a duplicate from hashes and global correlation.

4. Confirm or exclude candidates

Generate evidence for selected pairs or the highest-ranked unresolved pairs:

python audit_figures.py evidence --workspace AUDIT_DIR --pair PANEL_A,PANEL_B
python audit_figures.py evidence --workspace AUDIT_DIR --top 20

Inspect the full panels, data-only crops, match-line view, registered red/green overlay, and metrics together. For circular plates or other strong borders, repeat with a tighter interior crop. For blots, compare both whole blot context and protein-by-lane crops. For microscopy, distinguish same-field channel derivation from cross-condition reuse. Consult references/review-protocol.md for modality-specific checks and verdicts.

Never call a pair confirmed from an inlier count or NCC alone. Confirmation requires geometrically consistent correspondence across independent random details in the data region, a plausible transform, visual agreement after registration, and review of the experimental relationship. Record strong negative controls from visually similar nonmatching panels when possible.

5. Review uninformative images

Treat automated quality flags as prompts. Mark a panel uninformative only for a specific reason such as blank/placeholder content, corruption, unreadably low resolution, a caption mismatch, or unrelated residual artwork. A negative result, schematic, control, or visually sparse field is not "useless" merely because it contains little signal.

6. Report

The final report must include:

1. exact input, page scope, figure/source count, and unreadable or skipped files;
2. number of reviewed sources, panels, and all-pairs comparisons;
3. methods and whether the SIFT pass actually ran;
4. a table of confirmed duplicate, high-confidence concern, needs raw data, expected derivative/longitudinal view, and excluded false positive;
5. panel IDs, source/page, bounding boxes, metrics, and evidence paths for every reported concern;
6. separately listed quality/uninformative findings;
7. limitations, especially uncertain panel boundaries or unsplit lanes.

Use neutral language: the audit identifies image reuse or similarity, not intent. Do not claim the review is exhaustive unless coverage accounting shows that every in-scope source and experimental-image unit was inspected.