Face detecting system and method using symmetric axis

• US 20050135679A1
• Filed: 06/22/2004
• Published: 06/23/2005
• Est. Priority Date: 12/17/2003
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Yoon, Ho Sub, Kim, Do Hyung, Lee, Jae Yeon, Soh, Jung
• Granted Patent: US 7,426,287 B2
• Time in Patent Office: Days
• US Class Current: 382/190
• CPC Class Codes: G06V 40/164 using holistic features

Abstract

A face detecting system using a symmetric axis, the system including: an image inputting unit for inputting a still image thereto; an image converting unit for reducing the inputted image to a first certain size of image for a real-time process by using a pyramidal resolution reduction way; a scanning unit for scanning the reduced image by applying a second certain size of symmetric mask; a region dividing unit for dividing a region of the scanned mask into a left region and a right region with respect to the symmetric axis; a symmetric-similarity computing unit for computing a symmetric similarity of a local region (Symmetric_measure_local) at the two divided regions and then computing a symmetric similarity of a global region (Symmetric_measure_global) to compute a symmetric similarity of a face region; and a face region extracting unit for extracting a region having a symmetric axis at which the computed symmetric similarity is above a predetermined value, as the face region.

Claims

1. A face detecting system using a symmetric axis, the system comprising:

an image inputting unit for inputting a still image thereto;

an image converting unit for reducing the inputted image to a first certain size of image for a real-time process by using a pyramidal resolution reduction way;

a scanning unit for scanning the reduced image by applying a second certain size of symmetric mask;

a region dividing unit for dividing a region of the scanned mask into a left region and a right region with respect to the symmetric axis;

a symmetric-similarity computing unit for computing a symmetric similarity of a local region (Symmetric_measure_local) at the two divided regions and then computing a symmetric similarity of a global region (Symmetric_measure_global) to compute a symmetric similarity of a face region; and

a face region extracting unit for extracting a region having a symmetric axis at which the computed symmetric similarity is above a predetermined value, as the face region.

2. The face detecting system of claim 1, wherein the first certain size is 160×

120 (width×

length).

3. The face detecting system of claim 1, wherein the second certain size is 13×

12 (width×

length).

4. The face detecting system of claim 1, wherein the second certain size is 9×

9(width×

length)

5. The face detecting system of claim 1, wherein in case that the symmetric similarity of the local region (Symmetric_measure_local) computed in the symmetric-similarity computing unit is above a certain value, the symmetric similarity of the global region (Symmetric_measure_global) is computed.

6. The face detecting system of claim 1, wherein the symmetric similarity of the local region (Symmetric_measure_local) computed in the symmetric-similarity computing unit is computed in the following Equations 1 and 2.

$\begin{array}{c}\mathrm{Symmetric\_measure}\end{array}$

_local = W 1 * Symmetric_axis ⁢

_check +  

⁢

W 2 * Upper_block ⁢

_check + W 3 * Lower_block ⁢

_check + W 4 * Other_checks [ Equation ⁢

 

⁢

1 ]

1.0=W₁+W₂+W₃+W₄  

 

(Equation

     2)

7. The face detecting system of claim 6, wherein the Symmetric_axis_check is performed by binary-coding pixels forming the symmetric axis into a symmetric-axis threshold level as in the following Equation 3, and then analyzing the binary-coded result.

$\begin{array}{c}\underset{}{\stackrel{}{\mathrm{Symmetric}}}\end{array}$

 

⁢

axis j <

T height j = 0 ⁢

 

⁢

( k , j ) = { 255 ;

if ⁢

 

⁢

( SG ⁡

( k , j ) <

Symmetric_threshold ) 0 ;

otherwise [ Equation ⁢

 

⁢

3 ]

8. The face detecting system of claim 6, wherein the Symmetric_axis_check is performed by shifting the symmetric axis of the symmetric mask up to down and left to right to check the symmetric axis while jumping without checking all regions of the symmetric mask.

9. The face detecting system of claim 6, wherein the Upper_block_check is performed by checking whether or not pixels of a block existing from 0 to above ⅓

heights of the symmetric axis include a binary-coding region corresponding to an eye.

10. The face detecting system of claim 6, wherein the Lower_block_check is performed by checking whether or not pixels of a block existing from ⅔

to lowest (T_height) heights of the symmetric axis include a binary-coding region corresponding to a mouse or a nose.

11. The face detecting system of claim 1, wherein the symmetric similarity of the global region (Symmetric_measure_global) computed in the symmetric-similarity computing unit is computed in the following Equations 4 and 5.

$\begin{array}{c}\mathrm{Symmetric\_measure}\end{array}$

_global=W5*Gray_symmetric⁢

_difference+W6*Gray_threshold⁢

_symmetric⁢

_difference+W7*Edge_threshold⁢

_symmetric⁢

_difference[Equation⁢

 

⁢

4]
1.0=W₅+W₆+W₇  

 

(Equation

     5)

12. A face detecting method using a symmetric axis, the method comprising the steps of:

(a) inputting a still image thereto;

(b) reducing the inputted image to a first certain size of image for a real-time process by using a pyramidal resolution reduction way;

(c) scanning the reduced image by applying a second certain size of symmetric mask;

(d) dividing the scanned mask into a left region and a right region with respect to the symmetric axis;

(e) computing a symmetric similarity of a local region (Symmetric_measure_local) at the two divided regions and then computing a symmetric similarity of a global region (Symmetric_measure_global) to compute a symmetric similarity of a face region; and

(f) extracting a region having a symmetric axis at which the computed symmetric similarity is above a predetermined value, as the face region.

13. The face detecting method of claim 12, wherein in case that the symmetric similarity of the local region computed in the step (e) is above a certain value, the symmetric similarity of the global region (Symmetric_measure_global) is computed.

14. The face detecting method of claim 12, wherein the symmetric similarity of the local region (Symmetric_measure_local) computed in the step (e) is computed in the following Equations 1 and 2.

$\begin{array}{c}\mathrm{Symmetric\_measure}\end{array}$

_local = W 1 * Symmetric_axis ⁢

_check +  

⁢

W 2 * Upper_block ⁢

_check + W 3 * Lower_block ⁢

_check + W 4 * Other_checks [ Equation ⁢

 

⁢

1 ]

1.0=W₁+W₂+W₃+W₄  

 

(Equation

     2)

15. The face detecting method of claim 14, wherein the Symmetric_axis_check is performed by binary-coding pixels forming the symmetric axis into a symmetric-axis threshold level, and then analyzing the binary-coded result.

16. The face detecting method of claim 14, wherein the Symmetric_axis_check is performed by shifting the symmetric axis of the symmetric mask up to down and left to right to check the symmetric axis while jumping without checking all regions of the symmetric mask.

17. The face detecting method of claim 14, wherein the Upper_block_check is performed by checking whether or not pixels of a block existing from 0 to above ⅓

heights of the symmetric axis include a binary-coding region corresponding to an eye.

18. The face detecting method of claim 14, wherein the Lower_block_check is performed by checking whether or not pixels of a block existing from ⅔

to lowest (T_height) heights of the symmetric axis include a binary-coding region corresponding to a mouse or a nose.

19. The face detecting method of claim 12, wherein the symmetric similarity of the global region computed in the step (e) is computed in the following Equations 6 to 9.

$\begin{array}{c}\mathrm{Gray\_symmetric}\end{array}$

_difference = ( Equation ⁢

 

⁢

6 ) ∑

i = 0 , j = 0 j <

T width , j <

T height ⁢

( 255 - abs ⁡

( LTG ⁡

( i , j ) - BTG ⁡

( l - i , j ) )  

Gray_threshold ⁢

_symmetric ⁢

_difference = ∑

i = 0 , j = 0 j <

T width , j <

T height  

⁢

{ 255 ;

if ⁢

 

⁢

( LTGT ⁡

( i , j ) ⩵

255 ⁢

 

⁢

and ⁢

 

⁢

LTGT ⁡

( l - i , j ) ⩵

255 ) 0 ;

otherwise ( Equation ⁢

 

⁢

7 ) Edge_threshold ⁢

_symmetric ⁢

_difference = ∑

i = 0 , j = 0 j <

T width , j <

T height  

⁢

{ 255 ;

if ⁢

 

⁢

( LTET ⁡

( i , j ) ⩵

255 ⁢

 

⁢

and ⁢

 

⁢

LTET ⁡

( l - i , j ) ⩵

255 ) 0 ;

otherwise ( Equation ⁢

 

⁢

8 ) LTGT ⁢

( i ++ , j ++ ) i = 0 , j = 0 i <

T width , j <

T height = { 255 ;

if ⁢

 

⁢

( LTG ⁡

( i , j ) <

A_gray ⁢

_threshold ⁢

_value 0 ;

otherwise { Equation ⁢

 

⁢

9 ) RTGT ⁡

( i -- , j ++ ) l = T width , j = n i <

0 , j <

T height = { 255 ;

if ⁢

 

⁢

( RTG ⁡

( l , j ) <

B_gray ⁢

_threshold ⁢

_value 0 ;

otherwise