Method for identifying a sign on a deformed document

Patent metadata

Field	Value
Publication number	US-10025977-B2
Application number	US-201514878837-A
Country	US
Kind code	B2
Filing date	Oct 8, 2015
Priority date	Oct 10, 2014
Publication date	Jul 17, 2018
Grant date	Jul 17, 2018

Abstract

This invention relates to a method for identifying a sign on an image of a document that can be deformed comprising:
an acquisition (E 1 ) of said digital image of said document; a determination (E 2 ) in the acquired image of at least one candidate sign region using an image segmentation algorithm, for each candidate sign region, a calculation (E 3 ) of a signature comprising a piece of information concerning the location in the acquired image of said candidate sign region and a region descriptor concerning the local characteristics of the image in said region, an identification (E 4 ) of a sign on the image of the document using the calculated signatures comprising jointly a comparison (E 41 ) of the calculated signatures with reference signatures concerning sign regions of document models, said comparison being carried out according to a geometric deformation model of said document, and an estimation (E 42 ) according to said comparison of said geometric deformation model.

First claim

The invention claimed is:

1. A method for identifying at least one sign on at least one digital image of a document that can be deformed, said method being implemented by a processor able to be connected to a first storage device, said method comprising:
acquiring (E 1 ) said at least one digital image of said document;
determining (E 2 ) in the acquired digital image at least one sub-part of the acquired digital image, referred to as candidate sign region, using an image segmentation algorithm,
for each candidate sign region, calculating (E 3 ) a signature comprising location information in the acquired digital image, and a region descriptor concerning local characteristics of the acquired digital image in said candidate sign region,
identifying (E 4 ) at least one determined candidate sign region in the acquired digital image as a sign region,
said step of identifying comprising a matching of said determined candidate sign region with a reference sign region of a document model stored in the first storage device, said matching being carried out by comparing (E 41 ) the calculated signature of each candidate sign region with reference signatures corresponding to reference sign regions of document models,
the step of identifying also comprising an estimation (E 42 ) of a geometric deformation model of said document,
wherein said matching and said estimation are carried out concomitantly, each match between one of the determined candidate sign regions and one of the reference sign regions being carried out according to the geometric deformation model of the document, said geometric deformation model being estimated as a function of previously found matches.

2. The method of identifying as claimed in claim 1 , wherein for each candidate sign region:
said step of comparing (E 41 ) comprises:
comparing the signature calculated for said candidate sign region with said reference signatures stored in said first storage device,
matching the calculated signature with a reference signature according to said comparison and to the geometric deformation model of said document stored in a second storage device connected to said data processing device in such a way as to identify said candidate sign region,
and said step of estimating (E 42 ) comprises:
estimating a new geometric deformation model of said document from the current geometric deformation model and from said matching of the calculated signature with the reference signature,
storing in the second storage device said new geometric deformation model as the current deformation model.

3. The method of identifying according to claim 1 , wherein said document is a game ticket or an identity document.

4. The method of identifying according to claim 1 , wherein the sign to be identified is a geometric figure, a character, a group of characters or a graphical element.

5. The method of identifying according to claim 1 , wherein the sign to be identified is a pattern delimited by a closed contour.

6. The method of identifying according to claim 3 , wherein the sign to be identified is a box, a circle, a star or an alphanumeric character or a pattern specific to an issuing country of an identity document.

7. The method of identifying according to claim 3 , wherein the calculated signature of each candidate sign region further comprises scale information.

8. The method of identifying according to claim 3 , wherein the region descriptor relates to information on contour and/or content of said sign region.

9. The method of identifying according to claim 3 , wherein the geometric deformation model is determined using an inverse distance weighting algorithm or a splines interpolation algorithm.

10. A non-transitory computer-readable medium comprising code instructions for executing a method of identification as claimed in claim 3 , when said code instructions are executed by a processor.

11. A device for identifying at least one sign on at least one digital image of a document that can be deformed, said device comprising a processor able to be connected to a first storage device, the data processing device comprising:
a module for acquiring said at least one digital image of said document;
a module for determining in the acquired digital image of said document at least one sub-part of the acquired digital image, referred to as candidate sign region, using an image segmentation algorithm,
a module for calculating a signature for each candidate sign region comprising location information in the acquired digital image of and a region descriptor concerning local characteristics of the acquired digital image in said candidate sign region,
a module for identifying at least one sign region on said at least one digital image of said document among the determined candidate sign regions,
said module for identifying being configured to match a determined candidate sign region with a reference sign region of a document model stored in the first storage device, by operation of a module for comparing the calculated signature of each candidate sign region with reference signatures corresponding to reference sign regions of document models,
said module for identifying also comprising a module for estimating a geometric deformation model of said document,
wherein said matching and said estimation are carried out concomitantly, each match between one of the candidate sign regions and one of the reference sign regions being carried out according to a current geometric deformation model of the document, said current geometric deformation model being calculated as a function of previously found matches.

Assignees

• Morpho

Inventors

• Rouh Alain
• Beaudet Jean
• Rostaing Laurent

Classifications

• G06V30/414 Primary

  Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06V10/225

  based on a marking or identifier characterising the area · CPC title

• G06V10/20

  Image preprocessing · CPC title

• G06V10/247

  by affine transforms, e.g. correction due to perspective effects; Quadrilaterals, e.g. trapezoids · CPC title