Compare the signatures
Assess visible resemblance, reference variation and additional comparison signals.
Explore comparison workflowsSIGNATURE COMPARISON SOFTWARE
Signature comparison
& forgery indicators.
Compare questioned and reference signatures with forensic signature analysis software. Examine captured writing dynamics and extract specimens from documents.
Compare visible features, genuine capture data and extracted specimens with task-specific measurements and explanations.
VISUAL COMPARISONIllustrative signatures
Questioned sampleReference sample
See the overall form.
Review proportions, direction and visible stroke arrangement across the samples.
No score or verdict generated
Select a feature or play the illustration.
7 task-specific workflowsEach with a defined purpose and scope
Images, PDFs or capture dataInput requirements vary by workflow
Structured result outputsMeasurements support interpretation
01 / THE QUESTION DETERMINES THE METHOD
What do you need
to examine?
Start with the evidence you have. A scanned signature, a genuine digitizer capture and a full document require different analyses.
Examine the movement
Read speed, pressure, timing and pen lifts from genuine capture data.
Explore writing dynamicsExtract the specimens
Find candidate signatures, initials, handwritten names and company chops for review.
Explore document extractionHOW THE ANALYSIS WORKS
Make the comparison visible.
EVIDENCE / EXPLAINEDGuided demonstration
QUESTIONED / A KNOWN / B
ContourConnectionsSlantStroke width
Signatures / IN PRACTICE
See the shape. Compare the detail.
A questioned signature and known references
01 / Prepare the references
Separate questioned material from independently sourced known signatures. Retain natural variation across the reference set.
RECORDQuestioned specimen · known references
02 / Examine visible structure
Compare overall form and regional stroke detail. Explicit measurements can describe contours, connections and loops, slant and stroke width.
RECORDGlobal form · local features
03 / Compare complementary signals
Keep learned visual comparison and equal-weight shape measurements distinct. A separate visual review describes observed agreements and differences.
RECORDComparison values · descriptive review
04 / Document the result
Supported fusion combines scores into a calibrated likelihood ratio. Preserve its reference and calibration context alongside the separate comparison views.
RECORDSimilarity · distance · evidential weight
Demonstration material · annotations explain the workflow; no case result is calculated.
02 / CAPABILITIES, EXPLAINED
Signature analysis.
A clear output.
See what each workflow examines, what it needs and how to interpret the result. Expand a workflow for inputs, features and limitations.
01 / VISUAL COMPARISON
Visual signature similarity
Compare a questioned signature with known references, including signatures supplied in PDFs.
What you receive: a descriptive similarity score built from four equally weighted feature groups.
Inputs, features & limits
Required evidence
One questioned signature and 1–20 independently sourced known signatures.
- Measure contour shape, connected structures and loops, slant and stroke width.
- Give the four feature groups equal weight in the descriptive comparison.
- Keep each measurement group and the reference context visible alongside the overall similarity.
No genuine/forged verdict. Further validation is pending; visual resemblance is not proof of authorship.
JPEG · PNG · PDF
02 / REFERENCE VARIATION
Multiple-reference comparison
Compare the questioned specimen with variation across authenticated known signatures.
What you receive: a calibrated assessment and plain-language visual review.
Inputs, features & limits
Required evidence
One questioned signature and at least two known signatures.
- Encode visible signature characteristics as comparable numerical features.
- Compare the questioned specimen against supplied reference signatures.
- Use a separate visual review to describe observed similarities and differences.
- Keep the measured comparison and the visual explanation distinguishable.
The supplied unseen-scan notes reported about one in five genuine samples rejected. This is a specific test result, not a universal error rate.
PNG · JPEG · GIF · BMP · WEBP
03 / WRITING DYNAMICS
Movement, timing & pressure
Examine how a signature was written using actual position, time, pressure and pen-state records. Requires a genuine digitizer capture.
What you receive: writing measurements and side-by-side capture comparison.
Inputs, features & limits
Required evidence
A genuine digitizer capture; provide a second capture for questioned-versus-known comparison.
- Measure writing speed, pressure and pen movement.
- Inspect timing variation and real pen lifts.
- Validate capture signals before analysis.
A photo or scan cannot provide genuine speed or pressure data and is rejected for this workflow. No automatic same/different-writer score is produced.
SVC DIGITIZER CAPTURE
04 / ADDITIONAL SIGNAL
Independent distance comparison
Examine signature features at multiple scales alongside a plain-language review.
What you receive: raw distance and observable comparison points.
Inputs, features & limits
Required evidence
One questioned signature and at least two known signatures.
- Encode visible signature characteristics as comparable numerical features.
- Compare the questioned specimen against supplied reference signatures.
- Use a separate visual review to describe observed similarities and differences.
- Keep the measured comparison and the visual explanation distinguishable.
Raw distance is not a calibrated probability or a verdict. Multiple outputs may share inputs or failure modes and must not be treated as automatically independent evidence.
PNG · JPEG · GIF · BMP · WEBP
05 / COMBINED ASSESSMENT
Combined signature evidence
Bring three comparison signals together while accounting for variation among known references.
What you receive: one calibrated likelihood ratio.
Inputs, features & limits
Required evidence
One questioned signature and at least two known signatures. A single reference produces an inconclusive result.
- Combine comparison outputs into one evidential measure.
- Account for variation in the reference set.
- Keep the hypotheses and calibration context explicit.
A likelihood ratio is not an authenticity probability. The supplied unseen-scan notes reported about one in seven genuine samples rejected; consider the calibration population and test scope.
PNG · JPEG · GIF · BMP · WEBP
06 / DOCUMENT SCREENING
Signature & mark extraction
Locate candidate signatures, initials, handwritten names and company chops on a document. Extracted regions remain candidate specimens.
What you receive: candidate crops and a linked custody record.
Inputs, features & limits
Required evidence
A document or image.
- Screen ink regions for candidate signatures, initials, handwritten names and company chops.
- Rank the candidate regions for selection and extraction.
- Retain each extracted specimen with its source-page relationship and recorded processing steps.
Candidate screening and extraction coverage depends on page quality and mark type. The replacement page-level detector is still in development.
PNG · JPEG · TIFF · BMP · WEBP · PDF
07 / PRACTICE WORKFLOW
Experimental signature verification
Explore reference variation and simulated skilled forgeries in a training context. Practice only; not enabled for casework.
What you receive: a practice score using a fixed decision cut-off.
Inputs, features & limits
Required evidence
One questioned signature and at least two known signatures.
Compares questioned-to-reference differences with the variation among references.
Practice only; not enabled for casework. Supplied unseen-writer notes reported roughly one in three skilled forgeries still accepted. Do not use the score as a forensic conclusion.
IMAGE INPUTS · PRACTICE SPECIMENS
03 / TRADITIONAL + ASSISTED EXAMINATION
Less repetitive work.
More room for scrutiny.
Measure writing features, compare reference variation and organise the resulting outputs in one workflow.
| Examination task | Traditional examination | With Admissible assistance |
|---|---|---|
| Specimen preparation | Manually identify regions, organise references and document source relationships. | Candidate extraction and linked records can support preparation; every region still needs review. |
| Visual comparison | Inspection of form, stroke detail and natural variation. | Add descriptive similarity, multiple-reference assessment and plain-language observations. |
| Writing dynamics | Examine genuine capture records and interpret their movement and pressure data. | Quantify captured speed, pressure, timing and pen lifts for side-by-side review. |
| Time & cost | Preparation, comparison, reasoning and reporting require working time. | May reduce repetitive preparation and measurement work. Setup, checks, review and service fees also count. |
| Accuracy | Depends on source quality, references, reference suitability and method suitability. | Adds measurable signals but introduces validation, calibration and input-suitability requirements. No guaranteed accuracy uplift. |
| Result documentation | Document observed similarities, differences and reference variation in the report. | Add linked scores, comparison observations and method parameters to the report. |
Traditional examination may already use digital tools. This comparison describes workflow roles, not a claim that manual practice is inferior or that assisted outputs are automatically correct.
TIME & COST / YOUR OWN ESTIMATES
Compare the full effort.
Enter the total working hours you expect for each approach, including preparation, checking, interpretation and reporting. Include additional assisted-workflow fees.
What about accuracy?
Time saved is not evidence of better accuracy. Assess each workflow’s tested population, failure modes and suitability for the actual questioned material.
Planning scenarioIllustrative, editable
Enter non-negative values within the displayed input limits.
Estimated time reduction3 hours
Estimated cost reductionUS$350
Traditional: US$1,200 · Assisted: US$850
Example assumptions only. These are not measured platform savings, quoted fees or a performance guarantee.
04 / A TRACEABLE PROCESS
From a questioned mark.
To a documented result.
Preserve the source and make the reasoning traceable. Choose the analysis only after confirming the examination question and input requirements.
Preserve & identify
Retain the source, document provenance and select independently sourced known references.
Choose the signal
Match the method to visual images, genuine captures or document extraction.
Review the result
Inspect the source alongside outputs. Consider conflicts, limitations and alternative explanations.
Explain the finding
Document observations, measurements, source links and processing history in the result record.
START WITH THE EVIDENCE YOU HAVE
Find the workflow that fits.
Images, PDFs and genuine writing captures answer different questions.