Particular-region detection method and apparatus, and program therefor

• US 20050220347A1
• Filed: 03/31/2005
• Published: 10/06/2005
• Est. Priority Date: 03/31/2004
• Status: Active Grant

• Current Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Original Assignee: Fuji Photo Film Co Limited (Fujifilm Holdings Corporation)
• Inventors: Enomoto, Jun, Matsushita, Takafumi
• Granted Patent: US 7,613,332 B2
• Time in Patent Office: Days
• US Class Current: 382/190
• CPC Class Codes: G06T 2207/20132 Image cropping
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 2207/30216 Redeye defect
• CPC Class Codes: G06T 5/77 Retouching; Inpainting; Scratch removal
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 40/165 using facial parts and geometric relationships
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction

Abstract

In the invention, a particular-region such as red-eye region of a photographed image is detected, the image including a person as an object. At first a particular region candidate satisfying detection conditions for the particular region candidate is detected from the image. Then, a face region of the person satisfying detection conditions for the face region of the person is detected and the particular region candidate is specified as a particular region to be detected, in a condition that the particular region candidate is included in the detected face region. On the other hand, the image is divided into first and second regions at least before the detecting of the face region of the person. The detection conditions used for the detection for the particular region candidate or the detection for the face region of the person are changed between the first region and the second region of the image.

Claims

1. A particular-region detection method of detecting a particular region from an image including a person as an object, comprising:

detecting a region of the image satisfying detection conditions for particular region candidates as a particular region candidate;

detecting a face region of the person satisfying detection conditions for the face region of the person and specifying the particular region candidate as a particular region to be detected, in a condition that the particular region candidate is included in the detected face region; and

dividing the image into first and second regions at least before said detecting of the face region of the person,

wherein at least one of said detecting of the particular region candidate and said detecting of the face region of the person comprises changing the detection conditions used in said detecting between the first and second regions of the image.

2. The particular-region detection method according to claim 1, wherein:

the image is divided into the first and second regions before said detecting of the particular region candidate; and

said detecting of the particular region candidate comprises changing the detection conditions used in said detecting between the first and second regions of the image.

3. The particular-region detection method according to claim 1, wherein the particular region comprises a red-eye region.

4. The particular-region detection method according to claim 1, wherein:

the first region comprises a predetermined center region of the image; and

the second region comprises a peripheral region located around the center region.

5. The particular-region detection method according to claim 1, wherein:

the first region is set in a focused region; and

the second region is set in an unfocussed region.

6. The particular-region detection method according to claim 1, wherein:

the first region is set in a region in which luminance of the image exceeds a threshold value; and

the second region is set in a region in which luminance of the image is equal to or less than the threshold value.

7. The particular-region detection method according to claim 2, wherein the particular region candidate is a red-eye candidate, and the method further comprising setting the detection conditions for a red-eye candidate of the second region to be stricter in the red-eye detection than those of the first region, and setting the detection conditions of a face region of the first region to be stricter in the face region detection than those of the second region.

8. The particular-region detection method according to claim 1, wherein:

the image is divided into the first and second regions after said detecting of the particular region candidate and before said detecting of the face region of the person, in accordance with a detection result of the particular region candidate; and

when a plurality of particular region candidates concentrate within a predetermined range of the image, the predetermined range is set as the first region.

9. A particular-region detection apparatus for detecting a particular region from an image including a person as an object, comprising:

candidate detection means for detecting a region of the image satisfying detection conditions for particular region candidates as a particular region candidate;

face detection means for detecting a face region of the person satisfying the detection conditions for the face region of the person;

specifying means for specifying the particular region candidate as a particular region to be detected, in a condition that the particular region candidate is included in the detected face region by said face detection means;

region setting means for dividing the image into first and second regions at least before the detection of the face region of the person is executed,

wherein at least one of said candidate detection means and said face detection means changes the detection conditions used between the first and second regions of the image upon detection.

10. The particular-region detection apparatus according to claim 9, wherein said region setting means divides the image into the first region and the second region before the detection of the particular region candidate is executed, and when the detection of the particular region candidate is executed, the detection conditions used for the detection are changed between the first and second regions of the image.

11. The particular-region detection apparatus according to claim 9, wherein the particular region comprises a red-eye region.

12. The particular-region detection apparatus according to claim 10, wherein the particular region candidate is a red-eye candidate, and said candidate detection means sets the detection conditions for a red-eye candidate of the second region to be stricter in red-eye detection than those of the first region, and said face detection means sets the detection conditions for a face region of the first region to be stricter in red-eye detection than those of the second region.

13. The particular-region detection apparatus according to claim 9, wherein said region setting means divides the image into the first region and the second region after the detection of the particular region candidate and before the detection of the face region of the person, in accordance with a detection result of the particular region candidate, and when a plurality of particular region candidates concentrate within a predetermined range of the image, said region setting means sets the predetermined range as the first region, and sets a portion other than the predetermined range as the second region.

14. A program for causing a computer to detect a particular region from a fed image including a person as an object, the program comprising the procedures of:

causing calculating means of the computer to detect a region of the image satisfying detection conditions for particular region candidates as the particular region candidate;

causing the calculating means to detect a face region of the person satisfying detection conditions for the face region of the person and to set the particular region candidate as a particular region to be detected, in a condition that the particular region candidate is included in the detected face region; and

causing the calculating means to divide the image into first and second regions at least before the detection of the face region of the person,

wherein the calculating means changes the detection conditions used for at least one of the detection of the particular region candidate and the detection of the face region of the person between the first and second regions of the image.

15. The program according to claim 14, wherein the particular region comprises a red-eye region.

16. The program according to claim 14, wherein:

the particular region candidate is a red-eye candidate; and

when the red-eye candidate detection is executed, the detection conditions for a red-eye candidate of the second region are set to be stricter than those of the first region; and

when the face detection is executed, the detection conditions for a face region of the first region are set to be stricter than those of the second region.

17. The program according to claim 14, wherein:

the image is divided into the first region and the second region after the detection of the particular region candidates and before the detection of the face region of the person, in accordance with a detection result of the particular region candidate; and

when a plurality of particular region candidates concentrate within a predetermined range of the image, the predetermined range is set as the first region and a portion other than the predetermined range is set as the second region by the calculating means.