METHOD OF IMAGE PROCESSING

• US 20100226533A1
• Filed: 08/16/2007
• Published: 09/09/2010
• Est. Priority Date: 08/16/2006
• Status: Active Grant

• Current Assignee: Zebra Technologies Corporation
• Original Assignee: Imperial Innovations Limited (Imperial College London)
• Inventors: Ng Sing Kwong, Jeffrey, Bharath, Anil Anthony
• Granted Patent: US 8,718,321 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06T 7/74 involving reference images or patches
• CPC Class Codes: G06V 10/42 Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation

Abstract

The present invention relates to a method of identifying a target object in an image using image processing. It further relates to apparatus and computer software implementing the method. The method includes storing template data representing a template orientation field indicative of an orientation of each of a plurality of features of a template object; receiving image data representing the image; processing the image data to generate an image orientation field indicating an orientation corresponding to the plurality of image features; processing the image orientation field using the template orientation field to generate a match metric indicative of an extent of matching between at least part of the template orientation field and at least part of the image orientation field; and using the match metric to determine whether or not the target object has been identified in the image. Image and/or template confidence data is used to generate the match metric.

Claims

1. A method of identifying a target object in an image including a plurality of image features, wherein said method includes:

a) storing template data, said template data representing a template orientation field indicative of an orientation of each of a plurality of features of a template object;

b) receiving image data representing said image;

c) processing said image data to generate an image orientation field indicating an orientation corresponding to said plurality of image features;

d) processing said image data to generate image confidence data based on at least one characteristic for use in identifying said target object in a given image, said characteristic being indicative of an increased likelihood that at least one part of said given image represents at least part of said target object, relative to other parts of said given image;

e) processing said image orientation field using said template orientation field and said image confidence data to generate a match metric indicative of an extent of matching between at least part of said template orientation field and at least part of said image orientation field; and

f) using said match metric to determine whether or not said target object has been identified in said image.

2. A method according to claim 1, wherein said method includes applying said image confidence data to said image orientation field to generate a modified image orientation field indicating both an orientation corresponding to said plurality of image features and a likelihood that at least one part of said image orientation field represents at least part of said target object.

3. A method according to claim 1 or 2, wherein said characteristic indicated by said image confidence data relates to an extent of image data errors expected in said image data and said method includes using said image confidence data to generate said match metric in accordance with said expected extent of image data errors.

4. A method according to any preceding claim, wherein said characteristic indicated by said image confidence data relates to a possible occlusion of part of the target object in the image and said method includes using said image confidence data to generate said match metric in accordance with said possible occlusion.

5. A method according to any preceding claim, wherein said image data is video image data.

6. A method according to claim 5, wherein said characteristic relates to a possible movement of the target object.

7. A method according to claim 6, wherein said method includes using an identified movement of said target object to determine a modified shape of said target object which is modified from a shape of said target object represented by said template data.

8. A method according to any preceding claim, wherein said image data represents said image in colour and said characteristic relates to at least one colour indicative of said target object.

9. A method according to claim 8, wherein said target object is a human face and said characteristic relates to a colour indicative of human skin.

10. A method according to any preceding claim, including generating the match metric using deformation data indicative of a deformation of template data with respect to image data.

11. A method according to claim 10, wherein the deformation data is determined using a difference taken between template phase data indicative of a phase of at least one of the plurality of features of the template data, and image phase data indicative of a phase of at least one of the plurality of image features.

12. A method according to claim 10 or 11, wherein the deformation data is determined using a difference taken between template angle data indicative of an angle of at least one of the plurality of features of the template data, and image angle data indicative of an angle of at least one of the plurality of image features.

13. A method according to any preceding claim, wherein at least one of said image data, said template data, said image confidence data, said template phase data, said image phase data, said template angle data, said image angle data and said deformation data is arranged according to a matrix format.

14. A method according to any preceding claim, wherein each of said image features represents, in said image, a boundary which separates a region of higher light intensity from a region of lower light intensity.

15. A method according to claim 14, wherein said boundary separates two regions, each having a similar light intensity, from each other.

16. Apparatus arranged to perform the method of any of claims 1 to 15.

17. Computer software arranged to perform the method according to any of claims 1 to 15.

18. A data carrier storing the computer software of claim 17.

19. A method of identifying a target object in an image including a plurality of image features, wherein said method includes:

a) storing template data, said template data representing a template orientation field indicative of an orientation of each of a plurality of features of a template object;

b) receiving template confidence data indicating a likelihood that at least one part of said template accurately represents at least part of said target object;

c) receiving image data representing said image;

d) processing said image data to generate an image orientation field indicating an orientation corresponding to said plurality of image features;

e) processing said image orientation field using said template orientation field and said template confidence data to generate a match metric indicative of an extent of matching between at least part of said template orientation field and at least part of said image orientation field; and

f) using said match metric to determine whether or not said target object has been identified in said image.

20. A method according to claim 19, wherein said method includes applying said template confidence data to said template orientation field to generate a modified template orientation field indicating both an orientation of each of said plurality of features of said template object and a likelihood that at least one part of said template orientation field represents at least part of said target object.

21. A method according to claim 19 or 20, wherein said target object is a human face and a size of said template object corresponds to an expected radius of said human face.

22. A method according to any of claims 19 to 21, including generating the match metric using deformation data indicative of a deformation of template data with respect to image data.

23. A method according to claim 22, wherein the deformation data is determined using a difference taken between template phase data indicative of a phase of at least one of the plurality of features of the template data, and image phase data indicative of a phase of at least one of the plurality of image features.

24. A method according to claim 22 or 23, wherein the deformation data is determined using a difference taken between template angle data indicative of an angle of at least one of the plurality of features of the template data, and image angle data indicative of an angle of at least one of the plurality of image features.

25. A method according to any of claims 19 to 24, wherein at least one of said image data, said template data, said template confidence data, said template phase data, said image phase data, said template angle data, said image angle data and said deformation data is arranged according to a matrix format.

26. A method according to any of claims 19 to 25, wherein each of said image features represents, in said image, a boundary which separates a region of higher light intensity from a region of lower light intensity.

27. A method according to claim 26, wherein said boundary separates two regions, each having a similar light intensity, from each other.

28. Apparatus arranged to perform the method of any of claims 19 to 27.

29. Computer software arranged to perform the method according to any of claims 19 to 27.

30. A data carrier storing the computer software of claim 29.