Device for detecting red eye, program therefor, and recording medium storing the program

• US 20050226499A1
• Filed: 03/25/2005
• Published: 10/13/2005
• Est. Priority Date: 03/25/2004
• Status: Active Grant

• Current Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Original Assignee: Fuji Photo Film Co Limited (Fujifilm Holdings Corporation)
• Inventors: Terakawa, Kensuke
• Granted Patent: US 7,636,477 B2
• Time in Patent Office: Days
• US Class Current: 382/165
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction

Abstract

Red-eye area detection accuracy is improved according to a pattern of appearance of red-eye areas generated frequently in an actual photography environment. A red-eye candidate area finding unit finds red-eye candidate areas in a digital photograph image, by using reference data. A display unit displays the image with the red-eye candidate areas having been marked and preliminary corrected. A user can specify an unfound red-eye area and an erroneously specified area by using a specification unit 22. An update unit 24 updates the reference data by learning a characteristic of the unfound red-eye area and the erroneously specified area so that a probability becomes higher regarding detection of an area having a characteristic similar to the unfound red-eye area as a red-eye candidate area while becomes lower regarding detection of an area having a characteristic similar to the erroneously specified area as a red-eye area.

Claims

1. A red-eye detection device comprising:

image input reception means for receiving input of a digital photograph image;

red-eye candidate area finding means for finding red-eye candidate areas in the digital photograph image by using reference data defining a characteristic of a red-eye area;

display means for displaying the digital photograph image so that a user can confirm positions of the red-eye candidate areas that have been found;

specifying means for enabling the user to specify a red-eye area that has not been found as one of the red-eye candidate areas, as an unfound red-eye area; and

update means for carrying out an update of the reference data so as to improve a probability of finding an area having a similar characteristic to the unfound red-eye area as one of the red-eye candidate areas, by learning a characteristic of the unfound red-eye area that has been specified.

2. The red-eye detection device according to claim 1, further comprising correction means for carrying out image correction so as to suppress apparent red eye, on the red-eye candidate areas that have been found and the unfound red-eye area that has been specified.

3. The red-eye detection device according to claim 1, further comprising:

history storage means for storing history of the update of the reference data by the update means;

monitoring means for monitoring a change of detection accuracy based on frequency of specification of the unfound red-eye areas from the specifying means; and

resetting means for resetting the reference data to a state at one time in the past stored by the history storage means when the detection accuracy becomes lower than a predetermined threshold.

4. The red-eye detection device according to claim 2, further comprising:

history storage means for storing history of the update of the reference data by the update means;

monitoring means for monitoring a change of detection accuracy based on frequency of specification of the unfound red-eye areas from the specifying means; and

resetting means for resetting the reference data to a state at one time in the past stored by the history storage means when the detection accuracy becomes lower than a predetermined threshold.

5. The red-eye detection device according to claim 1, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and the characteristic of the unfound red-eye area that has been specified.

6. The red-eye detection device according to claim 2, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and the characteristic of the unfound red-eye area that has been specified.

7. The red-eye detection device according to claim 3, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and the characteristic of the unfound red-eye area that has been specified.

8. The red-eye detection device according to claim 4, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and the characteristic of the unfound red-eye area that has been specified.

9. The red-eye detection device according to claim 1, wherein the user is able to specify an area that has been found as one of the red-eye candidate areas but is not actually a red-eye area as an erroneously specified area by using the specifying means, and

the update means carries out the update of the reference data so as to lower a probability of finding an area having a similar characteristic to the erroneously specified area as one of the red-eye candidate areas, by learning a characteristic of the erroneously specified area that has been specified.

10. The red-eye detection device according to claim 9 further comprising correction means for carrying out image correction on the red-eye candidate areas excluding the erroneously specified area and on the unfound red-eye area so as to suppress apparent red eye.

11. The red-eye detection device according to claim 9 further comprising:

history storage means for storing history of the update of the reference data by the update means;

monitoring means for monitoring a change of detection accuracy based on frequency of specification of the unfound red-eye areas and the erroneously specified areas from the specifying means; and

resetting means for resetting the reference data to a state at one time in the past stored by the history storage means when the detection accuracy becomes lower than a predetermined threshold.

12. The red-eye detection device according to claim 10 further comprising:

history storage means for storing history of the update of the reference data by the update means;

monitoring means for monitoring a change of detection accuracy based on frequency of specification of the unfound red-eye areas and the erroneously specified areas from the specifying means; and

resetting means for resetting the reference data to a state at one time in the past stored by the history storage means when the detection accuracy becomes lower than a predetermined threshold.

13. The red-eye detection device according to claim 9, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and characteristics of the unfound red-eye area and the erroneously specified area.

14. The red-eye detection device according to claim 10, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and characteristics of the unfound red-eye area and the erroneously specified area.

15. The red-eye detection device according to claim 11, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and characteristics of the unfound red-eye area and the erroneously specified area.

16. The red-eye detection device according to claim 12, wherein the reference data in an initial state thereof are obtained by learning a sample image group comprising sample images representing images of red-eye areas and/or sample images representing images other than red-eye areas, and

the update means has storage means for storing data of the sample image group and carries out the update of the reference data by learning again a characteristic of the sample images comprising the sample image group and characteristics of the unfound red-eye area and the erroneously specified area.

17. The red-eye detection device according to claim 1 further comprising a means for showing a user a position or positions where there is a high probability of existence of unfound red-eye area or areas based on the results of detection of a red-eye or the face, thereby assisting the user to specify a red-eye area or areas.

18. The red-eye detection device according to claim 17, wherein the means for showing a user the position comprises means for showing the position of the possible unfound red-eye area in an enlarged scale at a position where a high possibility of existence of another eye is expected based on the results of detection of red-eye area in the case where one red-eye was found, or showing the position of the possible unfound red-eye areas in an enlarged scale at positions where a high possibility of existence of eyes is expected based on the results of detection of a face in the case where a face was detected.

19. A red-eye detection program for causing a computer to function as:

image input reception means for receiving input of a digital photograph image;

red-eye candidate area finding means for finding red-eye candidate areas in the digital photograph image by using reference data defining a characteristic of a red-eye area;

display means for displaying the digital photograph image so that a user can confirm positions of the red-eye candidate areas;

specifying means for enabling the user to specify a red-eye area that has not been found as one of the red-eye candidate areas, as an unfound red-eye area; and

update means for updating the reference data so as to improve a probability of finding an area having a similar characteristic to the unfound red-eye area as one of the red-eye candidate areas, by learning a characteristic of the unfound red-eye area.

20. A computer-readable recording medium storing a red-eye detection program for causing a computer to function as:

image input reception means for receiving input of a digital photograph image;

red-eye candidate area finding means for finding red-eye candidate areas in the digital photograph image by using reference data defining a characteristic of a red-eye area;

display means for displaying the digital photograph image so that a user can confirm positions of the red-eye candidate areas;

specifying means for enabling the user to specify a red-eye area that has not been found as one of the red-eye candidate areas, as an unfound red-eye area; and

update means for updating the reference data so as to improve a probability of finding an area having a similar characteristic to the unfound red-eye area as one of the red-eye candidate areas, by learning a characteristic of the unfound red-eye area.