Object Detection in an Image

• US 20110299783A1
• Filed: 05/03/2011
• Published: 12/08/2011
• Est. Priority Date: 05/06/2010
• Status: Active Grant

• Current Assignee: STMicroelectronics Pvt Ltd. (STMicroelectronics NV), STMicroelectronics SAS (STMicroelectronics NV)
• Original Assignee: STMicroelectronics SAS (STMicroelectronics NV)
• Inventors: Chotard, Ludovic, Sanches, Michel, Schwambach, Vitor, Chandra, Mahesh
• Granted Patent: US 8,805,081 B2
• Time in Patent Office: Days
• US Class Current: 382/195
• CPC Class Codes: G06F 18/2148 characterised by the process organisation or structure, e.g. boosting cascade
• CPC Class Codes: G06V 10/7747 Organisation of the process, e.g. bagging or boosting
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

The invention concerns a method of performing, by an image processing device, object detection in an image comprising: performing one or more tests of a test sequence for detection of a first object on pixels values of a plurality of at least partially overlapping sub-regions (310, 312, 314) of a first search window (108); generating a cumulative score based on results of said one or more tests on said plurality of sub-regions; comparing said cumulative score with a threshold value; and based on said comparison, selectively performing one or more of said tests of said test sequence on at least one further sub-region of said first search window, said at least one further sub-region at least partially overlapping each of said plurality of sub-regions.

Claims

1. A method of performing, by an image processing device, object detection in an image comprising:

performing one or more tests of a test sequence for detecting a first object, on pixel values of a plurality of at least partially overlapping sub-regions of a first search window;

generating a cumulative score based on results of said one or more tests on said plurality of sub-regions;

comparing said cumulative score with a threshold value; and

based on said comparison, selectively performing one or more of said tests of said test sequence on at least one further sub-region of said first search window, said at least one further sub-region at least partially overlapping each of said plurality of sub-regions.

2. The method of claim 1, wherein each of said one or more tests comprises:

determining a stage score based on selected pixel values of the sub-region;

comparing said stage score with a stage threshold value; and

selectively performing one or more further tests on pixel values of the sub-region based on said comparison.

3. The method of claim 1, wherein each of said one or more tests comprises analysing one or more Haar-like features within said plurality of sub-regions.

4. The method of claim 1, wherein said plurality of sub-regions comprises first, second and third sub-regions.

5. The method of claim 1, wherein if said comparison indicates that said cumulative score is not greater than said threshold, performing one or more tests of said test sequence on a plurality of at least partially overlapping sub-regions of a second search window, wherein said second search window is offset with respect to the first search window by a plurality of pixels in at least one direction.

6. The method of claim 5, wherein prior to performing said one or more tests on said plurality of sub-regions of the second search window, performing one or more tests of a second test sequence for detecting a second object on pixel values of said at least partially overlapping sub-regions of the first search window.

7. The method of claim 1, wherein said first search window is between 10 and 30 percent larger than each of said plurality of sub-regions.

8. The method of claim 1, wherein said pixel values are integral values each equal to a sum of pixel values of a determined area.

9. The method of claim 1 wherein said object comprises a face.

10. A device for performing object detection in an image comprising:

a memory storing said image; and

an object detection block configured to;

perform one or more tests of a test sequence for detection of a first object on pixels values of a plurality of at least partially overlapping sub-regions of a first search window;

generate a cumulative score based on results of said one or more tests on said plurality of sub-regions;

compare said cumulative score with a threshold value; and

based on said comparison, selectively perform one or more tests of said test sequence on at least one further sub-region of said first search window, said at least one further sub-region at least partially overlapping each of said plurality of sub-regions.

11. The device of claim 10 wherein the device comprises a digital camera.

12. An object detection unit for detecting an object in an image comprising:

a resize unit configured to resize said image;

an image integration unit configured to generate an integral image window by calculating, for each pixel value in the resized image, an integral value equal to the sum of one or more of said pixel values; and

a processing block configured to process said integral image window to detect an object, wherein said resize unit, image integration unit and processing block are configured to operate in parallel.

13. The object detection unit of claim 12, wherein said resize unit, image integration unit and processing block are configured such that while a first image is being resized by said resize unit, a first window of a second image is being integrated by said image integration unit, and a further window of said second image is being processed by said processing block to detect an object.

14. The object detection unit of claim 12, wherein said image integration unit is further configured to determine an integral square image.

15. The object detection unit of claim 12, further comprising a memory arranged to store:

said image;

a resized image for processing by said integration unit; and

at least part of an image being resized by said resize unit.

16. The object detection unit of claim 12, further comprising a circular buffer arranged to store said integral image window during the objection detection performed by said processing block.

17. The object detection unit of claim 12, further comprising a programmable memory configured to store at least one test sequence applied during the object detection performed by said processing block.

18. The object detection unit of claim 12, wherein said processing block comprises a microprocessor adapted to execute a first instruction set and an arithmetic unit adapted to perform specific operations.

19. The object detection unit of claim 12, further comprising a buffer arranged to store said integral image window prior to processing by said processing block, wherein each value of said integral image for a given pixel position is determined based on a sum of pixels of a rectangle, one corner of the rectangle corresponding to the given pixel position, and a diagonally opposite corner of the rectangle corresponding to a corner of the image, wherein said integral image is processed in overlapping strips of said image extending from one edge of the image to an opposite edge, and wherein all the values stored in said buffer are overwritten at the start of each new strip.

20. A method of performing, by an image processing device, object detection in an image comprising:

receiving the image;

storing the image in a first memory;

performing the following operations by running instructions on a processor;

performing one or more tests of a test sequence for detecting a first object, on pixel values of a plurality of at least partially overlapping sub-regions of a first search window;

generating a cumulative score based on results of said one or more tests on said plurality of sub-regions;

comparing said cumulative score with a threshold value; and

based on said comparison, selectively performing one or more tests of said test sequence on at least one further sub-region of said first search window, said at least one further sub-region at least partially overlapping each of said plurality of sub-regions; and

providing an indication of the detected object.

21. The method of claim 20 wherein providing an indication of the detected object comprises superimposing an indicator over a display of the image.

22. The method of claim 20 wherein the pixel values comprise luminance values.

23. The method of claim 20 further comprising capturing the image using an image sensor.

24. The method of claim 20 wherein the image is an image of a first resolution and further comprising storing the image in a second resolution.