Pre-Filter for Object Detection

• US 20080181455A1
• Filed: 12/10/2007
• Published: 07/31/2008
• Est. Priority Date: 01/30/2007
• Status: Active Grant

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Greig, Darryl
• Granted Patent: US 8,300,973 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

A method of pre-processing an image to identify processes for subsequent processing of the image, comprising the steps of:

• a) investigating portions of the image using a spatial filter;
• b) calculating for a first plurality of regions within a portion of the image under investigation respective metrics as a function of intensity within those regions;
• c) selecting combinations of regions within the portion of the image under investigation and processing them to obtain a second plurality of filter values, where the second plurality is greater than the first plurality; and
• d) comparing the filter values with process thresholds for subsequent processes so as to identify subsequent processes that can be skipped.

Claims

1. A method of pre-processing an image to identify processes for subsequent processing of the image, comprising the steps of:

a) investigating portions of the image using a spatial filter;

b) calculating for a first plurality of regions within a portion of the image under investigation respective metrics as a function of intensity within those regions;

c) selecting combinations of regions within the portion of the image under investigation and processing them to obtain a second plurality of filter values, where the second plurality is greater than the first plurality; and

d) comparing the filter values with process thresholds for subsequent processes so as to identify subsequent processes that can be skipped.

2. A method as claimed in claim 1, in which the spatial filter comprises at least first and second overlapping regions with the second region being enclosed by the first region.

3. A method as claimed in claim 2, in which the spatial filter has three regions, and a third region is enclosed by the second region.

4. A method as claimed in claim 2, in which the overlapping regions are concentric.

5. A method as claimed in claim 1, in which the spatial filter comprises a centrally disposed first region and at least four further regions disposed symmetrically around the central region.

6. A method as claimed in claim 5, in which the at least four further regions are equidistant from the first region.

7. A method as claimed in claim 6, in which each of the regions is rectangular and the centres of the rectangles are equally displaced from the centre of the first region.

8. A method as claimed in claim 1, in which during investigation of a portion of the image the sizes of one or more regions within the spatial filter are varied.

9. A method as claimed in claim 1, in which the spatial filter is scanned over the image.

10. A method as claimed in claim 3, in which the spatial filter comprises first, second and third squares of side W^o, W^m and W^I, and an average value is calculated for the area enclosed by each square.

11. A method as claimed in claim 3 in which the spatial filter comprises first, second and third squares and an average value is calculated for the area bounded between the first and second squares, an average value is calculated for the area bounded between the second and third squares, and for the area bounded by the third square.

12. A method as claimed in claim 10, in which the average values are identified as A(W^o), A(W^m) and A(W^I) for the first, second and third squares respectively and a parameter D for the area under investigation is calculated by:

D=A(W^o)−

2A(W^m)+A(W^I)

for a given portion of the image.

13. A method as claimed in claim 12, in which the parameter is compared with a threshold to indicate the number of times that the threshold is exceeded or not exceeded, as appropriate.

14. A method as claimed in claim 10, in which values are calculated for a plurality of square sizes.

15. A method as claimed in claim 13, in which the method is performed a plurality of times to build a confidence value for a feature under investigation being within the portion of the image.

16. A method as claimed in claim 15, in which the confidence values from different portions of the image are combined to obtain a metric for comparison with process thresholds.

17. A method as claimed in claim 1, in which the subsequent processes are orientation specific image recognition processes.

18. A method as claimed in claim 17, in which the subsequent processes are face detection processes.

19. A method as claimed in any claim 1, in which only subsequent processes where the filter values exceed an associated threshold are marked for execution.

20. A method as claimed in claim 1, in which an one of average intensity of the image within the portions of the spatial filter and a variance within or between the portions of the spatial filter are compared.

21. A method as claimed in claim 1, in which the spatial filter exhibits at least four degrees of rotational symmetry.

22. A method as claimed in claim 1, in which threshold values are varied within the calculation of the pre-filter value depending on the combination of test regions selected.

23. An image processor arranged to perform the method of claim 1.

24. A computer program adapted to cause a computer to perform the method of claim 1.

25. A method of selecting an orientation specific face detection algorithm from a plurality of such algorithms comprising scanning a test mask over an image under investigation, the test mask comprising three concentric shapes, forming an intensity value for the area bounded by each of the shapes, and processing the intensity values to obtain a metric indicating a presence of a face in the image.