Method of detecting objects

• US 6,901,163 B1
• Filed: 03/29/1999
• Issued: 05/31/2005
• Est. Priority Date: 05/19/1998
• Status: Expired due to Fees

• Current Assignee: Active Silicon Limited (Solid State Plc)
• Original Assignee: Active Silicon Limited (Solid State Plc)
• Inventors: Bone, Martin Bertram, Pearce, Henry Colin, Read, Christopher Lewis
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): Alavi, Amir
• Application Number:
  US09/701,128
• Time in Patent Office: 2,255 Days
• Field of Search: 382/162, 382/165, 382/164, 382/173, 382/181, 382/220, 382/218, 382/284, 209/576, 209/580, 209/581, 358/1.9, 358/515, 358/520, 358/530, 356/71
• US Class Current: 382/165
• CPC Class Codes: B07C 5/36 Sorting apparatus characterised by the means used for distribution sorting according to destination B07C3/003, B07C3/02
• CPC Class Codes: G01J 3/463 Colour matching
• CPC Class Codes: G01J 3/50 using electric radiation detectors
• CPC Class Codes: G07D 3/14 Apparatus driven under control of coin-sensing elements

Abstract

A method of recognizing and detecting colors is disclosed, to allow recognition and sorting of objects. Pixels of a color image of an object to be recognized are received, the hue, intensity and/or saturation of each pixel is determined, the hue, intensity and/or saturation of each received pixel is allocated to a plurality of predetermined hue, intensity and/or saturation bands respectively to obtain hue, intensity and/or saturation distribution values respectively for the color image, and the distribution values for the color image are compared with the distribution values from one or more reference objects to determine if the object is recognized as a said reference object. This method of object recognition is independent of the image complexity and the number of colors on the object to be recognized. The method operates to compare a hue signature of the object to be recognized with a hue signature of a reference object.

Claims

1. A method of object recognition by colour comprising:

receiving, as a plurality of object pixels, a colour object image of the object to be recognised;

allocating the object pixels to a plurality of predetermined hue bands;

allocating the object pixels to at least one of a plurality of intensity bands, a plurality of saturation bands, and a plurality of pseudo saturation bands;

characterised by;

comparing the numbers of object pixels in each said hue band with corresponding numbers for reference pixels representing a reference object to generate a hue correlation;

comparing the numbers of object pixels in each said intensity, saturation and/or pseudo saturation band with corresponding numbers for reference pixels representing a reference object to generate a respective intensity, saturation and/or pseudo saturation correlation; and

combining said generated correlations to generate a combined correlation representative of the probability that the object to be recognised matches the reference object.

2. The method as claimed in claim 1, wherein said combined correlation is generated by averaging said generated correlations.

3. The method as claimed in claim 2, wherein said combined correlation is generated by taking a weighted average of said generated correlations.

4. The method as claimed in claim 1, wherein said predetermined hue bands are predetermined by dividing all of the visible spectrum into 360 separate hue bands.

5. The method as claimed in claim 1, wherein said comparing steps are repeated for each of a plurality of reference objects to generate a respective plurality of hue and intensity, saturation and/or pseudo saturation correlations and said combining step is repeated for each of said plurality of reference objects to generate a respective combined correlation;

the method including recognising the object by determining which of said reference objects has the closest combined correlation with the object.

6. The method as claimed in claim 5, wherein said object is recognised as a said reference object when said combined correlation is larger than a threshold correlation.

7. The method as claimed in claim 6, wherein said object is rejected as not being recognised as a said reference object when said combined correlation is smaller than said threshold.

8. The method as claimed in claim 1, further comprising:

employing sliding correlation between the numbers of object pixels in each said intensity, saturation and/or pseudo saturation band and the corresponding numbers for the said reference pixels representing the reference object to generate the respective intensity, saturation and/or pseudo saturation correlation.

9. The method of sorting objects comprising the method of object recognition by colour as claimed in claim 1 for recognising objects;

and further comprising;

sorting objects according to the result of the recognition.

10. A method of sorting objects according to their colour, comprising:

receiving, as a plurality of object pixels, a colour object image of each of a plurality of reference objects of different colours;

allocating the object pixels into a plurality of predetermined hue bands and at least one of a plurality of predetermined intensity bands, saturation bands and pseudo saturation bands to generate a reference hue distribution and a respective reference intensity, saturation and/or pseudo saturation distribution for each said reference object;

receiving as a plurality of object pixels, a colour object image of each of a plurality of objects to be recognised;

allocating the object pixels to said plurality of predetermined hue bands and at least one of said plurality of predetermined intensity bands, saturation bands and pseudo saturation bands;

characterised by;

comparing the number of object pixels in each said hue band for the objects to be recognised with the number of object pixels in each said hue band for the reference objects to generate a respective hue correlation for each object to be recognised with respect to each reference object;

comparing the number of object pixels in each said intensity, saturation and/or pseudo saturation band for the objects to be recognised with the number of object pixels in each respective said intensity, saturation and/or pseudo saturation band for the reference objects to generate a respective intensity, saturation and/or pseudo saturation correlation for each object to be recognised with respect to each reference object;

combining said generated correlations for each object to be recognised with respect to each reference object to generate combined correlations representative of the probability that the objects to be recognised match the reference objects; and

sorting said objects to be recognised in accordance with the probability that the objects match the reference objects.

11. The method as claimed in claim 10, wherein each combined correlation is generated by averaging the relevant generated correlations for each object to be recognised with respect to each reference object.

12. The method as claimed in claim 11, wherein each combined correlation is generated by taking a weighted average of the relevant generated correlations.

13. The method as claimed in claim 10, wherein said predetermined hue bands are predetermined by dividing all of the visible spectrum into 360 separate hue bands.

14. The method as claimed in claim 10 wherein a particular one of said objects to be recognised is recognised as a particular one of said reference objects when the relevant combined correlation is larger than a threshold correlation.

15. The method as claimed in claim 14, wherein a particular one of said objects to be recognised is rejected as not being recognised as a particular one of said reference objects when the relevant combined correlation is smaller than said threshold.

16. The method as claimed in claim 10 further characterised by:

employing sliding correlation between the numbers of object pixels in each said intensity, saturation and/or pseudo saturation band and the corresponding numbers for the said reference pixels representing any one of said reference objects to generate the respective intensity, saturation and/or pseudo saturation correlation.

17. An apparatus for object recognition by colour, comprising:

a receiving element for receiving, as a plurality of object pixels, a colour object image of an object to be recognised;

a first allocating element for allocating the object pixels to a plurality of predetermined hue bands;

a second allocating element for allocating the object pixels to at least one of a plurality of intensity bands, a plurality of saturation bands, and a plurality of pseudo saturation bands;

a first comparing element for comparing the numbers of object pixels in each said hue band with corresponding numbers for reference pixels representing a reference object to generate a hue correlation;

a second comparing element for comparing the numbers of object pixels in each said intensity, saturation and/or pseudo saturation band with corresponding numbers for reference pixels representing a reference object to generate a respective intensity, saturation and/or pseudo saturation correlation; and

a combining element for combining said generated correlations to generate a combined correlation representative of the probability that the object to be recognised matches the reference object.

18. The apparatus as claimed in claim 17, wherein said combining element is adapted to generate said combined correlation by averaging said generated correlations.

19. The apparatus as claimed in claim 17, wherein said combining element is adapted to generate said combined correlation by taking a weighted average of said generated correlations.

20. The apparatus as claimed in claim 17 including an element for predetermining said predetermined hue bands by dividing all of the visible spectrum into 360 separate hue bands.

21. The apparatus as claimed in claim 17 wherein said first and second comparing elements are adapted to repeat the comparison for each of a plurality of reference objects to generate a respective plurality of hue and intensity, saturation and/or pseudo saturation correlations, and said combining element is adapted to repeat the combining for each of said plurality of reference objects to generate a respective combined correlation;

the apparatus including an element for recognising the objects by determining which of said reference objects has the closest combined correlation with the object.

22. The apparatus as claimed in claim 21, wherein said recognising element is adapted to recognise said object as a said reference object when said combined correlation is larger than a threshold correlation.

23. The apparatus as claimed in claim 22, wherein said recognising element is adapted to reject said object as not being recognised as a said reference object when said combined correlation is smaller than said threshold.

24. An apparatus for sorting objects according to their colour, the apparatus comprising:

a receiving element for receiving, as a plurality of object pixels, a colour object image of each of a plurality of reference objects of different colours;

a first allocating element for allocating the object pixels into a plurality of predetermined hue bands and at least one of a plurality of predetermined intensity bands, saturation bands and pseudo saturation bands, to generate a reference hue distribution and a respective reference intensity, saturation and/or pseudo saturation distribution for each said reference object;

means for receiving, as a plurality of object pixels, a colour object image of each of a plurality of objects to be recognised;

second allocating means for allocating the object pixels to said plurality of predetermined hue bands and at least one of said plurality of predetermined intensity bands, saturation bands and pseudo saturation bands;

characterised by;

a first comparing element for comparing the number of object pixels in each said hue band for the objects to be recognised with the number of object pixels in each said hue hand for the reference objects to generate a respective hue correlation for each object to be recognised with respect to each reference object;

a second comparing element for comparing the number of object pixels in each said intensity, saturation and/or pseudo saturation band for the objects to be recognised with the number of object pixels in each respective said intensity, saturation and/or pseudo saturation band for the reference objects to generate a respective intensity, saturation and/or pseudo saturation correlation for each object to be recognised with respect to each reference object;

a combining element for combining said generated correlations for each object to be recognised with respect to each reference object to generate combined correlations representative of the probability that the objects to be recognised match the reference objects; and

a sorting element for sorting said objects to be recognised in accordance with the probability that the objects match the reference objects.

25. The apparatus as claimed in claim 24, wherein said combining means is adapted to generate each combined correlation for each object to be recognised with respect to each reference object by averaging the relevant generated correlations for each object to be recognised with respect to each reference object.

26. The apparatus as claimed in claim 25, wherein said combining means is adapted to generate each combined correlation by taking a weighted average of the relevant generated correlations.

27. The apparatus as claimed in claim 24, including means for predetermining said predetermined hue bands by dividing all of the visible spectrum into 360 separate hue bands.

28. The apparatus as claimed in claim 24, wherein said recognising means is adapted to recognise a particular one of said objects to be recognised as a particular one of said reference objects when the relevant combined correlation is larger than a threshold correlation.

29. The apparatus as claimed in claim 28, wherein said recognising means is adapted to reject a particular one of said objects to be recognised as not being recognised as a particular reference object when said combined correlation is smaller than said threshold.

30. A carrier medium carrying computer code for object recognition by colour, the computer code comprising instructions for:

receiving, as a plurality of object pixels, a colour object image of the object to be recognised;

allocating the object pixels to a plurality of predetermined hue bands;

allocating the object pixels to at least one of a plurality of intensity bands, a plurality of saturation bands, and a plurality of pseudo saturation bands;

characterised by;

comparing the numbers of object pixels in each said hue band with corresponding numbers for reference pixels representing a reference object to generate a hue correlation;

comparing the numbers of object pixels in each said intensity, saturation and/or pseudo saturation band with corresponding numbers for reference pixels representing a reference object to generate a respective intensity, saturation and/or pseudo saturation correlation; and

combining said generated correlations to generate a combined correlation representative of the probability that the object to be recognised matches the reference object.