Image forensics that shows its working. Recompression tests, noise residuals and structure searches highlight the regions that behave differently, and every overlay links back to the measurement behind it.
[ 01 / Paradigm shift ]
Target vs Reality
Current
A Verdict Without Evidence
A tool returns a percentage with no explanation. The examiner cannot say what was measured, so the result is hard to rely on or defend.
Target
Regions, Not Guesses
Each sensor marks the region it measured and the reason. The examiner reviews the overlay, accepts or rejects it, and the decision is recorded.
[ 02 / Method ]
Built for scrutiny
01Preserve
The original image and its metadata are hashed and sealed.
02Probe
Recompression and noise tests are run on working copies.
03Search
Structure sensors look for repeated or foreign regions.
04Annotate
Overlays and written notes are attached to each finding.
Image Analysis Chain
Control planeAnalysis plane
01Metadata Read
EXIF · XMP · THUMBS
Reads embedded metadata, thumbnails and software tags and checks them against the pixels.
02Recompression Test
ELA · JPEG GHOSTS
Re-saves the image at known qualities and maps regions that respond differently.
03Noise Residual
SENSOR PATTERN
Extracts the noise residual and checks whether it is consistent across the frame.
04Structure Search
COPY-MOVE · SPLICE
Searches for repeated regions and edges that do not belong to the surrounding scene.
05Annotated Report
HEATMAPS · NOTES
Produces overlays and a written finding for each region the sensors flagged.
Control plane / CUSTODY LOG
Case Ledger
Keeps the original and each derived overlay linked by hash.
Hover or tap a stage to inspect
[ 03 / Core systems ]
Core sensors
01 / RECOMPRESSION01 / 04
ERROR LEVELANALYSIS
Edited regions can respond differently when the image is re-saved at a known quality.
ELA · JPEG GHOSTS
Stress the image and watch what moves.
MEASURES
Per-region response to controlled recompression at several qualities.
REPORTS
A heat map of regions that respond unlike their surroundings.
LIMITS
Previously compressed images show uneven results, so the sensor reports a reliability note.
02 / NOISE02 / 04
NOISERESIDUAL
Natural images carry a consistent noise pattern. Composites often break it.
RESIDUAL · BLOCK STATISTICS
Look beneath the picture.
MEASURES
Noise level and structure across blocks of the image.
REPORTS
Blocks whose noise statistics differ from the image as a whole.
LIMITS
Filters and resizing change noise, and the report records when they are detected.
03 / STRUCTURE03 / 04
COPY-MOVE& SPLICE
Duplicated and pasted regions leave matching or mismatched structure.
FEATURE MATCHING · EDGE CHECKS
Find what repeats and what does not belong.
MEASURES
Matching feature clusters inside the image, and edge continuity at region boundaries.
REPORTS
Linked pairs of regions and boundaries that look cut and placed.
LIMITS
Repeating patterns such as tiles and leaves create false pairs, so each match is reviewed by an examiner.
04 / PROVENANCE04 / 04
METADATACROSS-CHECK
Embedded fields are compared with what the pixels show.
EXIF · THUMBNAIL · SOFTWARE TAGS
Compare claims with content.
MEASURES
Camera tags, timestamps, embedded thumbnails and editing-software markers.
REPORTS
Differences between the embedded thumbnail and the main image, and inconsistent fields.
LIMITS
Metadata can be removed or forged, so it informs the examiner but does not decide the outcome.
[ 04 / Reference patterns ]
Where it applies
CASEWORK
Photographic evidence in a file
Overlays and notes that an examiner can add to a case bundle.
FINANCIAL
ID photos and proof documents
Screening of submitted images for editing before they reach a decision engine.
CLAIMS
Damage and receipt photos
Detection of reused or altered images across submissions.
[ 05 / Delivery & deployment ]
From evidence to report
Run it as a standalone workstation tool or as a service behind your own case system. Reports include the overlay images and the measurements.
Image forensics that shows its working. Recompression tests, noise residuals and structure searches highlight the regions that behave differently, and every overlay links back to the measurement behind it.
What changes compared with the current approach?+
A tool returns a percentage with no explanation. The examiner cannot say what was measured, so the result is hard to rely on or defend. Each sensor marks the region it measured and the reason. The examiner reviews the overlay, accepts or rejects it, and the decision is recorded.
Can it run on-premise or air-gapped?+
Yes. Analysis runs on your own hardware, and the platform can be installed with no connectivity. Reports are exported as signed PDF and machine-readable JSON.
How do I start?+
Request a briefing by form, WhatsApp (+60 17-773 7302) or email (kumar@spearcompute.com). We scope a pilot on samples you provide.