Object detector, object detecting method and robot

• US 20050069208A1
• Filed: 08/12/2004
• Published: 03/31/2005
• Est. Priority Date: 08/29/2003
• Status: Active Grant

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Morisada, Hidehiko
• Granted Patent: US 7,532,743 B2
• Time in Patent Office: Days
• US Class Current: 382/190
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

An object detector, an object detecting method and a robot can reduce diction errors of detecting wrong objects without increasing the volume of the computational operation to be performed to detect the right object. A face detector 101 comprises a face detecting section 110 that operates like a conventional face detecting section and is adapted to roughly select face candidates from an input image by template matching and detect face candidates by means of a support vector machine for face recognition, a non-face judging section 120 that detects non-face candidates that are judged to be non-faces and removes them from the face candidates selected by the face detecting section 110 and a skin tracker section 114 for tracking a face region after the non-face judgment. When the assumed distance between the face detector and the face as computed from the input image and the measured distance as measured by a distance sensor show a large difference, when the color variance of the face candidate is small, when the occupancy ratio of the skin color region is large and when the change in the size of the face region is large after the elapse of a predetermined time, the non-face judging section 120 judges such face candidates as non-faces and removes them from the face candidates.

Claims

1. An object detector for detecting an object of detection out of an input image, the detector comprising:

object detecting means adapted to detect object candidates judged to be the object; and

non-object detecting means adapted to detect non-object candidates judged to be non-objects and output the object candidates less the non-object candidates.

2. The object detector according to claim 1, wherein

the object of detection is a face region;

and

the object detecting means is adapted to detect face candidates judged to be the face region, satisfying certain face conditions;

while

the non-object detecting means is non-face detecting means adapted to detect non-face candidates judged to be non-faces, satisfying certain non-face conditions and output the face candidates less the non-face candidates.

3. The object detector according to claim 2, further comprising:

distance measuring means adapted to measure the distance between the object detector and the object of detection;

and the non-face detecting means being adapted to compute the assumed distance between the object detector and the object of detection on the basis of the size of the face candidate relative to the input image and judge the face candidate to be a non-face candidate when the difference between the assumed distance and the distance measured by the distance measuring means is greater than a predetermined threshold value.

4. The object detector according to claim 2, wherein the non-face detecting means is adapted to compute the variance of the face candidate in a color space and, if the variance is less than a predetermined threshold value, judge the face candidate to be a non-face candidate.

5. The object detector according to claim 2, wherein the non-face detecting means is adapted to judge the face candidate as a non-face candidate when the difference between the size of the face candidate in the current input image and the size of the face candidate in the input image of a predetermined time ago is greater than a predetermined threshold value.

6. The object detector according to claim 2, wherein the non-face detecting means is adapted to compute the occupancy ratio of the skin color region relative to the face region and judge the face candidate to be a non-face candidate when the occupancy ratio of the skin color region is higher than a predetermined threshold value.

7. The object detector according to claim 2, further comprising:

tracking means for tracking the face candidate from which the non-face candidate is removed by the non-face detecting means;

the non-face detecting means being adapted to detect the non-face candidate from the face candidate that is the object of tracking of the tracking means at a predetermined timing.

8. The object detector according to claim 2, wherein the face detecting means includes a matching result generating means adapted to determine the correlated value of the input image and a template of a predetermined size showing an average face image and discriminating means for discriminating the face candidate by pattern recognition from the input image region selected on the basis of the correlated value.

9. An object detecting method for detecting an object of detection out of an input image, the method comprising:

an object detecting step of detecting object candidates judged to be the object; and

a non-object detecting step of detecting non-object candidates judged to be non-objects and outputting the object candidates less the non-object candidates.

10. The method according to claim 9, wherein

the object of detection is a face region;

and

the object detecting step is a face detecting step of detecting face candidates judged to be the face region, satisfying certain face conditions;

while

the non-object detecting step is a non-face detecting step of detecting non-face candidates judged to be non-faces, satisfying certain non-face conditions and outputting the face candidates less the non-face candidates.

11. The method according to claim 10, further comprising:

the non-face detecting step is adapted to compute the assumed distance between the object detector and the object of detection on the basis of the size of the face candidate relative to the input image and judge the face candidate to be a non-face candidate when the difference between the assumed distance and the distance measured by distance measuring means for measuring the distance to the object of detection is greater than a predetermined threshold value.

12. The method according to claim 10, wherein the non-face detecting step is adapted to compute the variance of the face candidate in a color space and, if the variance is less than a predetermined threshold value, judge the face candidate to be a non-face candidate.

13. The method according to claim 10, wherein the non-face detecting step is adapted to judge the face candidate as a non-face candidate when the difference between the size of the face candidate in the current input image and the size of the face candidate in the input image of a predetermined time ago is greater than a predetermined threshold value.

14. The method according to claim 10, wherein the non-face detecting step is adapted to compute the occupancy ratio of the skin color region relative to the face region and judge the face candidate to be a non-face candidate when the occupancy ratio of the skin color region is higher than a predetermined threshold value.

15. The method according to claim 10, further comprising:

a tracking step of tracking the face candidate from which the non-face candidate is removed by the non-face detecting step; and

a non-face re-detecting step of detecting the non-face candidate from the face candidate that is the object of tracking at a predetermined timing.

16. A robot equipped with an object detector for detecting an object of detection out of an input image so as to autonomously act, the object detector comprising:

object detecting means adapted to detect object candidates judged to be the object; and

non-object detecting means adapted to detect non-object candidates judged to be non-objects and output the object candidates less the non-object candidates.

17. The robot according to claim 16, wherein

the object detector is adapted to detect a face region from the input image as object of detection;

and

the object detecting means is adapted to detect face candidates judged to be the face region, satisfying certain face conditions;

while

the non-object detecting means is non-face detecting means adapted to detect non-face candidates judged to be non-faces, satisfying certain non-face conditions and output the face candidates less the non-face candidates.

18. The robot according to claim 17, further comprising:

distance measuring means adapted to measure the distance between the object detector and the object of detection;

and the non-face detecting means being adapted to compute the assumed distance between the object detector and the object of detection on the basis of the size of the face candidate relative to the input image and judge the face candidate to be a non-face candidate when the difference between the assumed distance and the distance measured by the distance measuring means is greater than a predetermined threshold value.

19. The robot according to claim 17, wherein the non-face detecting means is adapted to compute the variance of the face candidate in a color space and, if the variance is less than a predetermined threshold value, judge the face candidate to be a non-face candidate.

20. The robot according to claim 17, wherein the non-face detecting means is adapted to judge the face candidate as a non-face candidate when the difference between the size of the face candidate in the current input image and the size of the face candidate in the input image of a predetermined time ago is greater than a predetermined threshold value.

21. The robot according to claim 17, wherein the non-face detecting means is adapted to compute the occupancy ratio of the skin color region relative to the face region and judge the face candidate to be a non-face candidate when the occupancy ratio of the skin color region is higher than a predetermined threshold value.