Method for recognizing pattern, pattern recognizer and computer program

• US 8,615,137 B2
• Filed: 07/06/2010
• Issued: 12/24/2013
• Est. Priority Date: 07/13/2009
• Status: Active Grant

• Current Assignee: Gurulogic Microsystems Oy
• Original Assignee: Gurulogic Microsystems Oy
• Inventors: Krkkinen, Tuomas
• Primary Examiner(s): Koziol, Stephen R
• Assistant Examiner(s): CHOI, TIMOTHY WING HO
• Application Number:
  US12/831,045
• Publication Number:
  US 20110007971A1
• Time in Patent Office: 1,267 Days
• Field of Search: 382/190, 382/103, 382/107, 382/181, 382209-223
• US Class Current: 382/209
• CPC Class Codes: G06V 10/7515 Shifting the patterns to accommodate for positional errors
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching

Abstract

The invention relates to a pattern recognizer, which, in order to recognize the pattern fast and with lowest possible computing power, comprises a memory (12) for storing area-specific reference values (REF) calculated on the basis of image information of image areas containing parts of the pattern to be recognized, and a processor (14) that is configured to divide (15) a received image into areas, to calculate (16) reference values (REF) area-specifically on the basis of the image information of said areas, to compare (17) the calculated reference values (REF) with the reference values (REF) stored in the memory (12) and to indicate (18) that the pattern is recognized, in case in the received image there is found a part consisting of adjacent areas, where the reference values (REF) of the areas correspond, with sufficient accuracy, to the reference values (REF) stored in the memory (12).

Claims

1. A method for determining a recognized pattern of an object in a single received original image in pixel format, comprising:

dividing the single received original pixel image into a plurality of received image areas,

calculating a plurality of area-specific reference values, one for each of the plurality of received pixel image areas, at least one of the area-specific reference values being based upon statistics of less than all of the sample pixel values in the respective received image area, at least one of the area-specific reference values being calculated based upon plurality of the sample pixel values in the respective received image area such that a total number of area-specific reference values is smaller than a total number of sample pixel values,

employing only one comparison step which includes comparing at least one of the calculated area-specific reference values with only first reference values associated with a plurality of pattern parts of objects, and

determining the recognized pattern of the object when two or more adjacent received image areas have area-specific reference values which correspond to two or more of the first reference values by a predetermined percentage.

2. The method of claim 1, wherein the area-specific reference value of the area is calculated by calculating a mean of samples comprising individual pixel values of the area.

3. The method of claim 1, wherein the area-specific reference value of the area is calculated by supersampling.

4. The method of claim 1, wherein the area-specific reference values of the areas in the part correspond, with sufficient accuracy, to the stored reference values if the proportion of areas which are found in said part and whose reference values correspond to the stored reference values of all stored reference values, exceeds a given limit value.

5. The method of claim 1, wherein

when the recognized pattern is found, information on the location of the pattern in the image is stored in the memory, and

when the recognized pattern is found a subsequent time in another image, at least the travel direction or speed of the pattern is calculated on the basis of the location of the pattern and the location of the pattern stored in the memory, and said at least travel direction or speed will be indicated.

11. A non-transitory computer readable medium containing a computer program configured to control a programmable device for implementing the method of claim 1.

12. A non-transitory computer readable medium containing a computer program configured to control a programmable device for implementing the method of claim 2.

13. A non-transitory computer readable medium containing a computer program configured to control a programmable device for implementing the method of claim 3.

14. A non-transitory computer readable medium containing a computer program configured to control a programmable device for implementing the method of claim 4.

15. A non-transitory computer readable medium containing a computer program configured to control a programmable device for implementing the method of claim 5.

16. The method of claim 1, wherein the step of dividing the single received image into a plurality of received image areas comprises dividing the single received image into a plurality of equal sized and shaped received image areas.

17. The method of claim 1, further comprising considering position data representing mutual locations of the areas corresponding to the reference values when comparing the calculated area-specific reference values with first reference values associated with a plurality of pattern parts of objects.

20. The method of claim 1, wherein the plurality of area-specific reference values corresponds to pixel image areas of identical size that comprise a plurality of pixels.

6. A pattern recognizer comprising:

a memory for storing area-specific reference values calculated on the basis of image information of image areas containing parts of a pattern to be recognized, and

a processor which is configured;

to divide a single received original pixel image into a plurality of received image areas,

to calculate a plurality of area-specific reference values, one for each of the plurality of received pixel image areas, at least one of the area-specific reference values being based upon statistics of less than all of the sample pixel values in the respective received image area, at least one of the area-specific reference values being calculated based upon a plurality of the sample pixel values in the respective received image area such that a total number of area-specific reference values is smaller than a total number of sample pixel values,

employing only one comparison step to compare the calculated area-specific reference values with first reference values associated with a plurality of pattern parts of objects, and

to determine a recognized pattern of the object when two or more adjacent received image areas have area-specific reference values which correspond to two or more of the first reference values by a predetermined percentage.

7. The pattern recognizer of claim 6, wherein the processor calculates the area-specific reference value of the area by calculating a mean of the samples consisting of individual pixel values of the received image area.

8. The pattern recognizer of claim 6, wherein the processor calculates the area-specific reference value of the area by supersampling.

9. The pattern recognizer of claim 6, wherein

when the recognized pattern is found, information on the location of the pattern in the image is stored in the memory, and

when the recognized pattern is found a subsequent time in another image, at least the travel direction or speed of the pattern is calculated on the basis of the location of the pattern and the location of the pattern stored in the memory, and said at least travel direction or speed will be indicated.

10. The pattern recognizer of claim 6, wherein

the pattern recognizer further comprises a user interface through which the user may select in the image the image areas where parts of the pattern to be recognized appear, and

the processor is configured to calculate area-specific reference values for the image areas selected through the user interface on the basis of the image information of said areas and to store said reference values in the memory.

18. The pattern recognizer of claim 6, wherein the processor is configured to divide the single received image into a plurality of received image areas by dividing the single received image into a plurality of equal sized and shaped received image areas.

19. The pattern recognizer of claim 6, wherein the processor is further configured to consider position data representing mutual locations of the areas corresponding to the reference values when comparing the calculated area-specific reference values with first reference values associated with a plurality of pattern parts of objects.

21. The pattern recognizer of claim 6, wherein the plurality of area-specific reference values corresponds to pixel image areas of identical size that comprise a plurality of pixels.