Image recognition scheme

• US 20020146174A1
• Filed: 01/23/2002
• Published: 10/10/2002
• Est. Priority Date: 01/25/2001
• Status: Active Grant

• Current Assignee: Ricoh Company Limited
• Original Assignee: Ricoh Company Limited
• Inventors: Shimura, Hiroshi
• Granted Patent: US 7,139,429 B2
• Time in Patent Office: Days
• US Class Current: 382/199
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 10/751 Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
• CPC Class Codes: G06V 20/95 Pattern authentication; Markers therefor; Forgery detection

Abstract

An image recognition method comprises the steps of: a) examining as to how much input image parts resemble a predetermined figure comprising a plurality of elements, and giving a first score on each image part as to how much it resembles to the element of the predetermined figure, and giving a second score on the image parts as to much a positional relationship therebetween resembles that of the elements of the predetermined figure; and b) determining as to whether or not the input image parts coincide the predetermined figure by using the first and second scores together.

Claims

1. An image recognition system comprising:

an examining part which examines as to how much input image parts resemble a predetermined figure comprising a plurality of elements, and gives a first score on each image part as to how much it resembles to the element of the predetermined figure, and gives a second score on the image parts as to how much a positional relationship therebetween resembles that of the elements of the predetermined figure; and

a determining part which determines as to whether or not the input image parts coincide with the predetermined figure by using the first respective scores and the second score synthetically.

2. The system as claimed in claim 1, wherein:

said examining part comprises a measurement part which measures the number of pixels each having a gray-scale level higher than a predetermined threshold on a foreground part, and measures the same on a background part of each image part, and a first scoring part which gives a score as to relationship between the thus-measured numbers of pixels.

3. The system as claimed in claim 2, wherein:

said measurement part does not measure the number of pixels on the background part near the boundary with the foreground part when the measured number of pixels on the foreground part is larger than a predetermined value.

4. The system as claimed in claim 1, wherein said examining part examines symmetry on each image part.

5. The system as claimed in claim 1, wherein said examining part examines area balance on each image part.

6. The system as claimed in claim 1, wherein said examining part examines angular periodicity on each image part.

7. The system as claimed in claim 1, wherein said examining part gives a plurality of primary scores on each image part with respect to various characteristic factors, and then, gives a final score thereon by combining the plurality of primary scores according to a predetermined manner.

8. The system as claimed in claim 1, further comprising a selecting part which selects one of a plurality of image parts to be sent to the determining part when the plurality of image parts are present less than a predetermined distance from each other.

9. The system as claimed in claim 1, wherein:

said determining part calculates a difference between an ideal arrangement of the respective elements of the predetermined figure and an actually measured arrangement of the respective input image parts.

10. The system as claimed in claim 1, wherein:

said determining part sums the total score on the respective input image parts and a score as to how much the arrangement of the respective image parts resembles that of the respective elements of the predetermined figure.

11. An image recognition method comprising the steps of:

a) examining as to how much input image parts resemble a predetermined figure comprising a plurality of elements, and giving a first score on each image part as to how much it resembles to the element of the predetermined figure, respectively, and giving a second score on the image parts as to much a positional relationship therebetween resembles that of the respective elements of the predetermined figure; and

b) determining as to whether or not the input image parts coincide the predetermined figure by using the first and second scores synthetically.

12. The method as claimed in claim 11, wherein:

said step a) comprises the steps of a1) measuring the number of pixels each having a gray-scale level higher than a predetermined threshold on a foreground part, and measuring the same on a background part of each image part, and a2) giving a score as to relationship between the thus-measured numbers of pixels.

13. The method as claimed in claim 12, wherein:

said step a1) does not measure the number of pixels on the background part near the boundary with the foreground part when the measured number of pixels on the foreground part is larger than a predetermined value.

14. The method as claimed in claim 11, wherein said step a) examines symmetry on each image part.

15. The method as claimed in claim 11, wherein said step a) examines area balance on each image part.

16. The method as claimed in claim 11, wherein said step a) examining angular periodicity on each image part.

17. The method as claimed in claim 11, wherein said step a) gives a plurality of primary scores on each image part with respect to various factors, and then, gives a final score thereon by combining the plurality of primary scores according to a predetermined manner.

18. The method as claimed in claim 11, further comprising the step c) of selecting one of a plurality of image parts to be processed by said step b), when the plurality of image parts are present less than a predetermined distance from each other.

19. The method as claimed in claim 11, wherein:

said step b) calculates a difference between an ideal arrangement of the respective elements of the predetermined figure and an actually measured arrangement of the respective input image parts.

20. The method as claimed in claim 11, wherein:

said step b) sums the total score on the respective input image parts and a score as to how much the arrangement of the respective image parts resembles that of the respective elements of the predetermined figure.