Source
Which document and page does the region belong to?
BETASIGNATURE POSITIONING
Keep every signature region
tied to its document.
ChequeDB's beta provides positioning data alongside signature detection. Use the location of each candidate to highlight it on the page, connect it with the expected field and prepare a crop for comparison without losing the surrounding document.
Forms · contracts · mandates · delivery notes · cheques
Document → signature evidenceILLUSTRATION
THE FOUR QUESTIONS
Source drawing · 420 × 520 px
01 · x:36 y:378
w:154 h:7502 · x:230 y:378
w:154 h:75Illustrated boxes use an upper-left origin, in pixels, on the 420 × 520 source drawing. Keep the page and coordinate frame with each region.
Synthetic drawings and example workflow. No upload, model inference or live validation is performed here.
REGION PLUS CONTEXT
A useful position points back to the right page.
Localisation is the technical term for finding where a signature candidate sits in the document image. A bounding region describes that area. Its meaning depends on the page and image geometry it refers to.
Keep the source page available when using a crop. Nearby text, signature labels and overlapping stamps can change how a reviewer interprets the writing. Several candidates on one page should remain distinguishable rather than being treated as one combined signature.
Geometry
What image dimensions, units and orientation describe its location?
Region
Which detected candidate does this box or boundary identify?
SYNTHETIC COORDINATE EXAMPLE
Turn a location into a visible region.
Consider a fictional rendered page measuring 1,200 by 1,600 pixels. This example uses an upper-left origin and a rectangular region. It illustrates coordinate handling, not a ChequeDB API response.
Multiplying each fraction by the relevant image dimension gives the pixel value shown. The rectangle extends from x 300 to 660 and y 1,120 to 1,248. Its position is useful only with the same page dimensions and origin convention.
Document services use different geometry representations. Agree the beta coordinate convention before implementing overlays or crops; do not assume that another provider's field names or units apply.
| Region component | Pixels on this page | Fraction of page dimension |
|---|---|---|
| Left edge | 300 | 0.25 of page width |
| Top edge | 1,120 | 0.70 of page height |
| Width | 360 | 0.30 of page width |
| Height | 128 | 0.08 of page height |
IMAGE TRANSFORMS
Display the box on the image it describes.
If a page is resized for a preview, its pixel coordinates need the corresponding scale. If it is rotated or deskewed before detection, a region described on the processed image must be mapped to the image the reviewer sees.
Agree which image is authoritative for positioning: the submitted page, a corrected page or a rendered preview. Preserve enough transformation context to reproduce the highlight. Reusing coordinates on a different image can move the box onto nearby text or clip the writing.
COMPARISON INPUTS
Keep the strokes complete when preparing a crop.
A crop makes the candidate easier to inspect beside a reference. It should include the full useful writing, including an opening stroke or terminal flourish that extends beyond the printed signature line.
Cropping too tightly can create an apparent difference by cutting off a stroke. Cropping too widely can include unrelated notes or stamps. Evaluate crop completeness on your own layouts, and retain a clear relationship between the crop, its region and the original page.
INTEGRATION WORKSHEET
Agree the geometry before building the review screen.
Location supports a field-placement check; it does not identify the signer. On a customer-and-countersigner agreement, retain each region's context and use the agreed document mapping before associating it with an expected role.
| Question | What to settle in the beta integration |
|---|---|
| Which page? | Document identity and page numbering convention. |
| Which coordinates? | Origin, units, rectangle or boundary representation and source dimensions. |
| Which orientation? | The source image used for detection and any transformation needed for display. |
| Which candidate? | How regions remain distinct and relate to prepared crops. |
| Which expected field? | The layout context used to assess placement, with review for ambiguity. |
QUESTIONS & ANSWERS
Positioning: practical questions.
Is positioning the same as signature localisation?
Yes. Both describe where a signature candidate occurs in the document image. The location becomes useful when its page, dimensions, units and orientation are known.
Are positions returned in pixels or normalised coordinates?
Confirm the exact convention in the current beta contract. The example on this page shows how the two representations relate; it does not specify production response fields.
Can the same box be used after resizing or rotation?
Only with the appropriate mapping. Scaling changes pixel coordinates, and rotation changes the coordinate frame. Align the region with the displayed image rather than reusing values without their source context.
Does the region tell us which person signed?
It shows a candidate's location. A nearby field label or known layout can help your process associate that region with an expected role, but position alone does not establish identity.
CHEQUEDB SIGNATURES · BETA
Start with your documents.
Define the checks that matter.
Discuss your document types, signature references and validation workflow with the ChequeDB team.