Method for the recognition of raised characters, corresponding computer program and device

1. A method of character recognition implemented by an electronic device, wherein the method comprises:

recognizing, in an image representing an object comprising at least one raised character, called a basic image, at least one raised character of said basic image, wherein recognizing comprises:

a phase of processing at least one image, said phase comprising at least one implementation of a Phong reflection model and delivering at least one identification image, comprising generating by the Phong reflection model, preliminarily or on request, model images of raised characters in a plurality of pre-determined lighting conditions delivering a plurality of sets of reference images;

a phase of identifying characters as a function of the basic image and said at least one identification image, comprising:

comparing at least certain portions of the basic image with at least certain of the reference images, delivering a list of selected reference images, called a list of identification images, that correspond to the characters in the basic image; and

recognizing characters in the basic image using the list of identification images,

wherein said phase of processing said at least one image comprises building a normal map, representing normals of dots on at least one zone of the basic image delivering said identification image and wherein building comprises:

identifying a flat zone on said basic image;

determining a lighting condition from a light intensity of a pixel or a group of pixels on the flat zone;

computing normals of said dots of said at least one zone according to the determined lighting condition.

2. The method according to claim 1 , wherein the raised characters are situated on a bank card in said basic image.

3. The method according to claim 2 , wherein the method comprises a phase of preliminary processing of an image of the bank card captured by a camera, delivering said basic image.

4. The method according to claim 3 , wherein said phase of preliminary processing comprises:

detecting edges of the bank card, delivering an intermediate image;

selecting lines in the intermediate image;

computing intersections of the lines, delivering four corners of the bank card; and

converting the intermediate image, so that the four corners coincide with four corners of a rectangle, delivering said basic image.

5. The method according to claim 1 , wherein the determining and the computing use the following equation according to the Phong illumination model:

I

P

=

k

a

⁢

i

a

+

∑

m

⁢

(

k

d

⁡

(

L

^

m

·

N

^

)

⁢

i

m

,

d

+

k

s

⁡

(

R

^

m

·

V

^

)

α

⁢

i

m

,

s

)

where:

k a ∈[0, 1] is a constant related to the ambient component;

k d ∈[0, 1] is a constant related to the scattered component;

k s ∈[0,1] is a constant related to specular component;

α>>1 is a constant related to the brilliance of the material;

m is the number of light sources;

I p represents a total reflected intensity at a given point;

{circumflex over (L)} is a director vector of light;

{circumflex over (N)} is a director vector of the surface normal;

{circumflex over (R)} is a director vector of the direction in which light would be reflected on a mirror, {circumflex over (R)}=2({circumflex over (L)}·{circumflex over (N)}){circumflex over (N)}−{circumflex over (L)}.

6. The method according to claim 1 , wherein said phase of identifying said characters comprises:

sub-dividing said normal map, delivering a list of sections of said normal map, each section corresponding to a character in said at least one zone;

building a list of current vectors from said list of sections, each current vector representing a section;

comparing said current vectors with a list of model vectors preliminarily built on the basis of normal mappings of the known characters, each model vector corresponding to a known character;

determining a list of current characters corresponding to said list of current vectors.

7. The method according to claim 6 , wherein said current vector and said model vectors have m×n dimensions, each of said normal maps of the known characters having a size of m×n pixels.

8. The method according to claim 6 , wherein said at least one zone comprises a zone of a card number of a bank card and in that said list of current characters comprises said card number of said bank card.

9. The method according to claim 8 , wherein the method comprises verifying said card number of said bank card according to Luhn's formula.

10. An electronic device for recognizing raised characters in a basic image, wherein the device comprises:

a processing unit; and

a non-transitory computer-readable medium comprising program code instructions stored thereon, which when executed by the processing unit configure the processing unit to perform acts comprising:

a phase of processing at least one image, said phase comprising at least one implementation of a Phong reflection model and delivering at least one identification image, comprising generating by the Phong reflection model, preliminarily or on request, model images of raised characters in a plurality of pre-determined lighting conditions delivering a plurality of sets of reference images;

a phase of identifying characters as a function of the basic image and said at least one identification image, comprising:

comparing at least certain portions of the basic image with at least certain of the reference images, delivering a list of selected reference images, called a list of identification images, that correspond to the characters in the basic image; and

recognizing characters in the basic image using the list of identification images,

wherein said phase of processing said at least one image comprises building a normal map, representing normals of dots on at least one zone of the basic image delivering said identification image and wherein building comprises:

identifying a flat zone on said basic image;

determining a lighting condition from a light intensity of a pixel or a group of pixels on the flat zone;

computing normals of said dots of said at least one zone according to the determined lighting condition.

11. A non-transitory computer-readable medium comprising a computer program product stored thereon, comprising program code instructions to implement a method of recognizing, in an image representing an object comprising at least one raised character, called a basic image, at least one raised character of said basic image, when said program is executed on a computer, wherein the method comprises:

a phase of processing at least one image, said phase comprising at least one implementation of a Phong reflection model and delivering at least one identification image, comprising generating by the Phong reflection model, preliminarily or on request, model images of raised characters in a plurality of pre-determined lighting conditions delivering a plurality of sets of reference images;

a phase of identifying characters as a function of the basic image and said at least one identification image, comprising:

comparing at least certain portions of the basic image with at least certain of the reference images, delivering a list of selected reference images, called a list of identification images, that correspond to the characters in the basic image; and

recognizing characters in the basic image using the list of identification images,

wherein said phase of processing said at least one image comprises building a normal map, representing normals of dots on at least one zone of the basic image delivering said identification image and wherein building comprises:

identifying a flat zone on said basic image;

determining a lighting condition from a light intensity of a pixel or a group of pixels on the flat zone;

computing normals of said dots of said at least one zone according to the determined lighting condition.