Method and device for processing color information of a pattern having cluttered and useful information characters

• US 7,583,836 B2
• Filed: 03/10/2004
• Issued: 09/01/2009
• Est. Priority Date: 03/17/2003
• Status: Active Grant

• Current Assignee: Ip Ot Sub Ulc
• Original Assignee: Captaris, Inc. (Open Text Corporation)
• Inventors: Hund, Martin
• Primary Examiner(s): Couso; Jose L
• Application Number:
  US10/549,063
• Publication Number:
  US 20060210149A1
• Time in Patent Office: 2,001 Days
• Field of Search: 382162-167, 382/173, 382/176, 382/181, 358/3.21, 358515-520, 358462-464
• US Class Current: 382/162
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/162 Quantising the image signal

Abstract

In a method for processing of color information of a document provided with noise characters and usable characters, a digital image of the document is generated comprised of image points. At least one image region of a first type is selected corresponding to a region of the document that contains noise characters but no usable characters. A brightness value range based on brightness value ranges of image points of the first type is established. For saturation values of image points of the first type whose brightness values lie in the established brightness value range, a frequency with which these saturation values occur in the image region of the first type is determined. Based on these saturation values whose determined frequency exceeds a predetermined value, a saturation value range containing these saturation values is established. For color tone values of image points of the first type whose brightness values lie in the established brightness value range, a frequency with which these color tone values occur in the image region of the first type is determined. Based on color tone values whose determined frequency exceeds a predetermined value, a color tone value range containing these color tone values is established. Colors of those image points whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range are associated with a first color class of colors corresponding to the noise characters. Colors of image points of the first type that are not associated with the first color class are associated with a second color class of colors corresponding to a character-less document. At least one image region of a second type is selected known to correspond to a region that contains neither noise nor usable characters. Colors of image points of the second type whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range are excluded from the first color class and are associated with the second color class.

Claims

1. A method for processing of color information of a document provided with noise characters and usable characters, the method performed by a computer system with a processor and memory, the method comprising the steps of:

generating a digital image of the document comprised of image points whose color is respectively established by a brightness value, a saturation value, and a color tone value;

selecting at least one image region of a first type which is known to correspond to a region of the document that contains noise characters but no usable characters;

establishing a brightness value range based on brightness value ranges of image points of the image region of the first type, said brightness value range being assumed to contain brightness values of the image points corresponding to the noise characters, wherein establishing the brightness value range is performed by the processor executing instructions stored in the memory;

for saturation values of those image points of the image region of the first type whose brightness values lie in the established brightness value range, respectively determining a frequency with which these saturation values occur in the image region of the first type, wherein determining the frequency with which these saturation values occur is performed by the processor executing instructions stored in the memory;

based on those saturation values whose determined frequency exceeds a predetermined value, establishing a saturation value range containing these saturation values, wherein establishing the saturation value range is performed by the processor executing instructions stored in the memory;

for color tone values of those image points of the image region of the first type whose brightness values lie in the established brightness value range, respectively determining a frequency with which these color tone values occur in the image region of the first type, wherein determining the frequency of color tone values is performed by the processor executing instructions stored in the memory;

based on those color tone values whose determined frequency exceeds a predetermined value, establishing a color tone value range containing these color tone values, wherein establishing the color tone value range is performed by the processor executing instructions stored in the memory;

associating the colors of those image points of the image region of the first type whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range with a first color class of colors corresponding to the noise characters, wherein associating the colors with the first color class is performed by the processor executing instructions stored in the memory;

associating the colors of those image points of the image region of the first type that are not associated with the first color class with a second color class of colors corresponding to a character-less document, wherein associating the colors with the second color class is performed by the processor executing instructions stored in the memory;

selecting at least one image region of a second type which is known to correspond to a region of the document that contains neither noise characters nor usable characters; and

excluding the colors of those image points of the image region of the second type whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range from the first color class and associating said colors with the second color class, wherein excluding the colors from the first color class and associating said colors with the second color class is performed by the processor executing instructions stored in the memory.

2. A method according to claim 1 wherein a largest brightness value and a smallest brightness value of the image points of the image region of the first type are determined and the brightness value range is established such that its brightness values H fulfill the following conditions:

3. A method according to claim 1 wherein a frequency distribution of the brightness values, the saturation values and color tone values are respectively determined for the image points of the image region of the first type whose brightness value ranges lie in the established brightness value range, which frequency distribution specifies the frequency of the respective values in the image region of the first type, and the frequency distributions are respectively quantized in a predetermined number of stages and are subsequently divided by a predetermined normalization constant.

4. A method according to claim 1 wherein image points whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range are determined outside of the image region of the first type, and the colors of these determined image points are associated with the first color class.

5. A method according to claim 1 wherein at least one image region of a third type is selected which is known to correspond to a region of the document that contains usable characters but no noise characters, and the colors of those image points of the image region of the third type that are not associated with the first color class are associated with a third color class of colors corresponding to the usable characters.

6. A method according to claim 5 wherein an average brightness value of the image points of the image region of the second type is determined, a further brightness value is established on the basis of an average brightness value of the image points of the image region of the second type, and the colors of those image points of the image region of the third type that are not associated with the first color class are only associated with the third color class when the brightness values of these image points lie in the further brightness value range.

7. A method according to claim 6 wherein the further brightness value range is established such that its brightness values H fulfill the following condition:

H<

H_AVERAGE−

q

Wherein H_AVERAGE designates the average brightness value and q designates a quantity dependent on the average brightness value.

8. A method according to claim 7 wherein the quantity q that is dependent on the average brightness value fulfills the following condition:

q =0.1×

H_AVERAGE

9. A method according to claim 1 wherein in a first step the method is initially applied to a first document provided with noise characters and usable characters in order to associate the colors corresponding to the noise characters of the first document with the first color class, and in a second step the method is subsequently applied to at least one further, second document in order to add to the first color class the colors corresponding to the noise characters of the second color class.

10. A method according to claim 9 wherein in a third step the method is applied again to the first document in order to associate the colors corresponding to the character-less first document with the second color class, and in a fourth step the method is subsequently applied again to the second document in order to add to the second color class the colors corresponding to the character-less second document.

11. A method according to claim 10 wherein in a fifth step the method is applied again to the usable characters of the first document with a third color class, and in a sixth step the method is subsequently applied to the second document in order to add to the third color class the colors corresponding to the usable characters of the second document.

12. A method according to claim 5 wherein the filter properties of a color filter are established based on at least the first or the third color class, which color filter can be used in an automatic character recognition in order to separate the information corresponding to the noise characters from the information corresponding to the usable characters.

13. A method according to claim 12 wherein the filter properties of the color filter are also established on the basis of the second color class.

14. A computer system for processing color information of a document having noise characters and usable characters, the computer system comprising:

a processor; and

a memory;

wherein the processor executes instructions stored in the memory to perform steps comprising;

generating a digital image of the document comprised of image points whose color is respectively established by a brightness value, a saturation value, and a color tone value;

selecting at least one image region of a first type which is known to correspond to a region of the document that contains noise characters but no usable characters;

establishing a brightness value range based on image points of the image region of the first type;

for saturation values of those image points of the image region of the first type lying in the established brightness value range, respectively determining a frequency with which these saturation values occur in the image region of the first type;

based on those saturation values whose determined frequency exceeds a predetermined value, establishing a saturation value range containing these saturation values;

for color tone values of those image points of the image region of the first type whose brightness values lie in the established brightness value range, respectively determining a frequency with which those color tone values occur in the image region of the first type;

based on those color tone values whose determined frequency exceeds a predetermined value, establishing a color tone value range containing these color tone values;

associating the colors of those image points of the image region of the first type whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range with a first color class of colors corresponding to the noise characters;

associating the colors of those image points of the image region of the first type that are not associated with the first color class with a second color class of colors corresponding to a character-less document;

selecting at least one image region of a second type which is known to correspond to a region of the document that contains neither noise characters nor usable characters; and

excluding the colors of those image points of the image region of the second type whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range from the first color class and associating said colors with the second color class.

15. A method of transforming a document into a format suitable for automatic character recognition, the method performed by a computer system with a processor and a memory, the method comprising:

generating a digital image of the document, the digital image comprised of image points, wherein each image point is associated with a color established by a brightness value, a saturation value, and a color tone value;

selecting at least one region of the digital image, wherein the at least one region is of a first type, wherein the first type has at least one noise character and no usable characters;

establishing a brightness value range for the at least one region of the first type, said brightness value range based on brightness values of at least two image points located in the at least one region of the first type wherein establishing the brightness value range is performed by the processor executing instructions stored in the memory;

determining a saturation value for at least one image point in the at least one region of the first type having a brightness value within the established brightness value range;

calculating a saturation value frequency that said saturation value occurs in the at least one image region of the first type wherein calculating the saturation value frequency is performed by the processor executing instructions stored in the memory;

establishing a saturation value range, wherein the saturation value range includes said saturation value if the calculated saturation value frequency of said saturation value exceeds a predetermined saturation threshold value;

determining a color tone value for at least one image point in the at least one region of the first type wherein establishing the saturation range is performed by the processor executing instructions stored in the memory;

having a brightness value within the established brightness value range;

calculating a color tone value frequency that said color tone value occurs in the at least one image region of the first type wherein calculating a color tone value frequency is performed by the processor executing instructions stored in the memory;

establishing a color tone value range, wherein the color tone value range includes said color tone value if the calculated color tone value frequency of said color tone value exceeds a predetermined color tone threshold value;

associating a color of an image point of the at least one image region of the first type wherein establishing the color value range is performed by the processor executing instructions stored in the memory;

with a first color class if the image point has a brightness value that lies in the established brightness value range, if the image point has a saturation value that lies in the established saturation value range, and if the image point has a color tone value that lies in the established color tone value range, wherein the first color class corresponds to noise characters;

associating a color of an image point of the at least one image region of the first type wherein associating the colors with the first color class is performed by the processor executing instructions stored in the memory;

with a second color class if the color of the image point is not associated with the first color class;

selecting at least one second region of the digital image, wherein the at least one second region of the digital image is of a second type wherein associating the colors with the second color class is performed by the processor executing instructions stored in the memory;

wherein the second type contains neither noise characters nor usable characters;

excluding a color of an image point of the at least one image region of the second type wherein excluding the colors from the first color class is performed by the processor executing instructions stored in the memory;

from the first color class if the image point has a brightness value that lies in the established brightness value range, if the image point has a saturation value lies in the established saturation value range, and if the image point has a color tone value that lies in the established color tone value range; and

associating a color of an image point of the at least one image region of the second type wherein associating said colors with the second color class is performed by the processor executing instructions stored in the memory;

with the second color class if the image point has a brightness value that lies in the established brightness value range, if the image point has a saturation value lies in the established saturation value range, and if the image point has a color tone value that lies in the established color tone value range.

16. A method according to claim 15 wherein a largest brightness value and a smallest brightness value of the image points of the image region of the first type are determined and the brightness value range is established such that its brightness values H fulfill the following conditions:

17. A method according to claim 15 wherein a frequency distribution of the brightness values, the saturation values and color tone values are respectively determined for the image points of the image region of the first type whose brightness value ranges lie in the established brightness value range, which frequency distribution specifies the frequency of the respective values in the image region of the first type, and the frequency distributions are respectively quantized in a predetermined number of stages and are subsequently divided by a predetermined normalization constant.

18. A method according to claim 15 wherein image points whose brightness values lie in the established brightness value range, whose saturation values lie in the established saturation value range, and whose color tone values lie in the established color tone value range are determined outside of the image region of the first type, and the colors of these determined image points are associated with the first color class.

19. A method according to claim 15 wherein at least one image region of a third type is selected which is known to correspond to a region of the document that contains usable characters but no noise characters, and the colors of those image points of the image region of the third type that are not associated with the first color class are associated with a third color class of colors corresponding to the usable characters.

20. A method according to claim 15 wherein an average brightness value of the image points of the image region of the second type is determined, a further brightness value is established on the basis of an average brightness value of the image points of the image region of the second type, and the colors of those image points of the image region of the third type that are not associated with the first color class are only associated with the third color class when the brightness values of these image points lie in the further brightness value range.

21. A method according to claim 20 wherein the further brightness value range is established such that its brightness values H fulfill the following condition:

H<

H_AVERAGE−

q

Wherein H_AVERAGE designates the average brightness value and q designates a quantity dependent on the average brightness value.