Method of detecting a specific object in an image signal

• US 6,542,625 B1
• Filed: 01/06/2000
• Issued: 04/01/2003
• Est. Priority Date: 01/08/1999
• Status: Expired due to Term

• Current Assignee: LG Electronics, Inc. (LG Corporation)
• Original Assignee: LG Electronics, Inc. (LG Corporation)
• Inventors: Lee, Jin Soo, Kim, Hyeon Jun
• Primary Examiner(s): AHMED, SAMIR ANWAR
• Application Number:
  US09/478,495
• Time in Patent Office: 1,181 Days
• Field of Search: 382/103, 382/117, 382/118, 351/200, 351/204, 348/77, 348/78, 348/169, 348/526, 340/5.53, 340/5.83, 356/71, 396/18, 396/51
• US Class Current: 382/118
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

A method of detecting a specific object in an image signal is disclosed. In the present invention, the final object region can be determined by selecting one of two object regions detected by different methods or by combining the two object regions with assigned weights.

Claims

1. A method of detecting a specific object in an image signal comprising:

(a) detecting a first image difference between a first two frames having a time difference of Δ

t to obtain a first candidate region of the specific object;

(b) detecting a second image difference between a second two frames having a time difference of Δ

t′

to obtain a second candidate region of the specific object, where Δ

t>

Δ

t′

; and

(c) determining a final region of the specific object utilizing the first candidate region and the second candidate region.

2. A method of claim 1, wherein in (a) detecting the first image difference comprises:

detecting a difference between a gray value of a certain pixel in an image frame at time t and a gray value of a corresponding pixel of the same position in an image frame at time t′

; and

assigning a value of “

1”

to a gray value of a pixel in the first image difference if the detected difference between the gray values is larger than a predetermined threshold value.

3. A method of claim 1, wherein (a) further comprises:

determining whether a movement of the specific object exists based upon the first image difference;

searching for a color region representing a characteristic of the specific object if the movement of the specific object exists;

obtaining a color difference image if the color region exists by executing an AND operation with respect to the color region and the first image difference; and

detecting MBRs using the color difference image and determining MBRs which satisfies a predetermined condition as the first candidate region.

4. A method of claim 3, wherein detecting MBRs comprises:

dividing the color difference image into grid images of a predetermined size wherein if more than a predetermined number of pixels within a grid image have values of “

1,”

the corresponding grid is assigned a value of “

1;

”

determining connected regions of grid images having values of “

1;

” and

obtaining MBRs wherein each MBR includes a connected region, and determining MBRs as a first object candidate region if a MBR satisfies a condition that a density of grids having the value of “

1”

is larger than a predetermined threshold value and a condition that the aspect ratio of the MBR is within a predetermined range.

5. A method of claim 1, wherein (b) further comprises:

determining whether a movement of an object feature exists based upon the second image difference;

detecting a position of an object feature if the movement of an object feature exists; and

determining a region of certain distance which surrounds an object feature as the second candidate region.

6. A method of claim 5, wherein detecting the second image difference comprises:

detecting a difference between a gray value of a certain pixel in an image frame at time t and a gray value of a corresponding pixel of the same position in an image frame at time t′

; and

assigning a value of “

1”

to a gray value of a pixel in the first image difference if the detected difference between the gray values is larger than a predetermined threshold value.

7. A method of claim 6, wherein detecting the position of an object feature comprises determining a position where a high density of pixels having gray values of “

1”

as the position of an object feature.

8. A method of claim 5, wherein determining the second candidate region comprises adjusting and arranging a template size appropriately to match a detected position of an object feature, and determining the region of the template as the second candidate region.

9. A method of claim 1, wherein in (c), determining the final region comprises:

template-matching the first and second candidate regions; and

combining the first and second candidate regions to obtain a final region if a ratio of an intersection of the template matched candidate regions to the second candidate region obtained is greater than a predetermined threshold value.

10. A method of claim 9, wherein if said ratio is not greater than the predetermined threshold value, performing a separate template matching for each of the first and second candidate regions, and determining one or both as final face regions if a template matched result exceeds the predetermined threshold value.

11. A method of claim 1, wherein in (c), determining the first candidate region as the final region if a second candidate region cannot be obtained.

12. A method of claim 1, wherein (a) is performed prior to (b) and (b) is performed for a second two frames between the first two frames.

13. A method of claim 1, wherein (a) and (b) are simultaneously performed.

14. A method of claim 1, wherein (a) further comprises:

determining whether a movement of the specific object exists based upon the first image difference;

searching for a color region representing a characteristic of the specific object if the movement of the specific object exists;

obtaining a color difference image if the color region exists by executing an AND operation with respect to the color region and the first image difference;

detecting MBRs using the color difference image and determining MBRs which satisfies a predetermined condition as the first candidate region; and

wherein (b) further comprises;

determining whether a movement of an object feature exists based upon the second image difference;

detecting a position of an object feature if the movement of an object feature exists; and

determining a region of certain distance which surrounds an object feature as the second candidate region.

15. A method of claim 14, wherein the specific object is a face and the object feature is an eye or a mouth.