Image template masking
View Patent ↗A method of identifying images as matching comprises comparing two image templates using a master mask to select corresponding codes from the templates. The master mask excludes blocks from the matching process and/or weights blocks according to their known or expected reliability.
1. A method of determining whether a first iris image matches a second iris image, comprising:
extracting from said first and second iris images respective first and second templates, each template comprising a spatial arrangement of template codes;
overlaying on said first template a master mask comprising a spatial arrangement of master mask codes, each master mask code being representative of an expected variability or matching reliability within a population of iris images of different individuals of a template code of corresponding spatial location;
comparing said overlaid first template with said second template; and
determining whether a match exists in dependence upon said comparison, said match indicating that said first and second iris images are images of the same human or animal individual.
2. A method as claimed in claim 1 including the step of generating the master mask from a plurality of existing images.
3. A method as claimed in claim 2 in which the plurality of existing iris images includes a plurality of iris image pairs that are known to match.
4. A method as claimed in claim 3 in which the master mask is generated by applying a matching algorithm to said pairs of iris images that are known to match and defining each said master mask code in dependence upon a variability or matching reliability of a template code of a corresponding spatial location for said pairs.
5. A method as claimed in claim 4 in which said variability or matching reliability is defined in dependence upon an average taken across all said pairs.
6. A method as claimed in claim 5 including setting a value of a least reliable master mask code, re-applying said matching algorithm using said value, and setting a value of a second least reliable master mask code.
7. A method as claimed in claim 6 including iteratively re-applying said matching algorithm to set all values of said master mask code until a matching error reaches a noise value.
8. The method of claim 2 wherein the master mask is generated from template codes of said plurality of existing images.
9. A method as claimed in claim 1 in which said template codes are binary codes.
10. A method as claimed in claim 9 in which said master mask codes are binary codes.
11. A method as claimed in claim 10 in which said master mask codes are in one-to-one correspondence with said template codes.
12. A method as claimed in claim 1 in which said master mask codes comprise weightings to be applied to corresponding spatial locations of said first template.
13. A method as claimed in claim 1 in which determination of a match is dependent upon an average distance metric, for example a Hamming distance, between said overlaid first template codes and said second template codes.
14. A method as claimed in claim 1 in which the determination of a match is dependent upon an average weighted distance metric, for example a weighted Hamming distance, between said overlaid first template codes and said second template codes.
15. A method as claimed in claim 1 in which said master mask is overlaid in a plurality of spatial positions on said first template to find a position of best match.
16. A method as claimed in claim 1 in which said first or second templates, or both, have individual masks which are applied in addition to said master mask.
17. A method as claimed in claim 1 in which said master mask is applied to said first template to create a stored composite template, said stored composite template then being compared with said second template.
18. A method as claimed in claim 1 in which said one of said first and second iris images is a candidate for verification against the other of said first and second iris images.
19. A method as claimed in claim 1 in which one of said first and second iris images is retrieved from a stored database for one-to-many matching against the other of said first and second iris images.
20. A method as claimed in claim 1 in which one of said first and second iris images is compared with a plurality of the other of said first and second iris images.
21. A method as claimed in claim 1 in which said second template is compared at a plurality of locations against said overlaid first template to find a second template position of best match.