White balance method

• US 20060164521A1
• Filed: 04/11/2005
• Published: 07/27/2006
• Est. Priority Date: 01/27/2005
• Status: Active Grant

• Current Assignee: Arima Computer (Texas) Corporation (Arima Communications Corporation)
• Original Assignee: Arima Computer (Texas) Corporation (Arima Communications Corporation)
• Inventors: Fuh, Chiou-Shann, Hsueh, Albert, Chikane, Varsha
• Granted Patent: US 7,495,696 B2
• Time in Patent Office: Days
• US Class Current: 348/223.100
• CPC Class Codes: H04N 23/88 for colour balance, e.g. white-balance circuits or colour temperature control

Abstract

A white balance method for an image formed by a digital still camera is provided. The white balance method comprises steps of: removing a color cast over a white object of the image; detecting a reference white point from the image by extracting an a pixel information of the white object; and adjusting a white balance of the white object by determining a scale factor based on the color cast and the pixel information.

Claims

1. A white balance method for an image formed by a digital still camera, comprising steps of:

(a) removing a color cast over a white object of said image;

(b) detecting a reference white point from said image by extracting an a pixel information of said white object; and

(c) adjusting a white balance of said white object by determining a scale factor based on said color cast and said pixel information.

2. The white balance method according to claim 1, wherein said step (a) further comprises steps of:

storing an original image data I_org (R_org, G_org, B_org) of said image;

performing a histogram equalization for said original image data I_org; and

converting said original image data I_org from an RGB color space to a YCrCb color space so as to obtain a histogram-equalized image data I_Hist (Y_Hist, C_rHist, C_bHist).

3. The white balance method according to claim 2, wherein said step (b) further comprises steps of:

determining all candidate white pixels from said histogram-equalized image data I_Hist with an equation (1);

Y_Hist≧

210, and
−

3≦

C_rHist, C_bHist≦

+3  

 

(1);

obtaining a relatively brightest pixel (Y_Hist^bright, C_rHist, C_bHist^bright) from said candidate white pixels, wherein said relatively brightest pixel has a relatively maximum Y_Hist value and has C_rHist and C_bHist values being relatively closest to zero;

calculating an average pixel (Y_Hist^avg, C_rHist^avg, C_bHist^avg) of said candidate white pixels;

determining all reference white pixels from said RGB color space if corresponding pixels thereof from said histogram-equalized image data I_Hist satisfy equation (2);

Y_l≦

Y_Hist≦

Y_u
C_rl≦

C_rHist≦

C_ru  

 

(2),
C_bl≦

C_bHist≦

C_bu

wherein Y_l and Y_u are a relatively maximum value and a relatively minimum value of a range from said Y_Hist^bright and said Y_Hist^avg respectively, C_rl and C_cu are a relatively maximum value and a relatively minimum value of a range from said C_rHist^bright and said C_rHist^avg respectively, and C_bl and C_bu are a relatively maximum value and a relatively minimum value of a range from said C_bHist^bright and said C_bHist^avg respectively; and

calculating an average W (R_w, G_w, B_w) of said reference white pixels as said pixel information.

4. The white balance method according to claim 3, wherein said step (c) further comprises steps of:

calculating a first scale factor (R_scale, G_scale, B_scale) from said average W(R_w, G_w, B_w) of said reference white pixels with equations (3) and (4);

R_scale=Y_w/R_w
G_scale=Y_w/G_w  

 

(3)
B_scale=Y_w/B_w
Y_w=0.299*R_w+0.587*G_w+0.114*B_w  

 

(4); and

calculating a second scale factor (R_GWA, G_GWA, B_GWA) with equations (5) and (6);

R_GWA=Y_avg/R_avg
G_GWA=Y_avg/G_avg  

 

(5)
B_GWA=Y_avg/B_avg
Y_avg=0.299*R_avg+0.587*G_avg+0.114*B_avg  

 

(6),

wherein R_avg, G_avg, and B_avg are average color components of said original image data I_org (R_org, G_org, B_org), respectively.

5. The white balance method according to claim 4, further comprising steps of:

determining said color cast as being bluish if B_Hist^avg+3≧

G_Hist^avg and B_Hist^avg>

R_Hist^avg, wherein (R_Hist^avg, G_Hist^avg, B_Hist^avg) is obtained by converting said average pixel (Y_Hist^avg, C_rHist^avg, C_bHist^avg) of said candidate white pixels from said YC_rC_b color space to said RGB color space; and

taking (R_scale, G_scale, B_GWA) as said scale factor.

6. The white balance method according to claim 4, further comprising steps of:

determining said color cast as being reddish if G_Hist^avg+3 >

R_Hist^avg>

B_Hist^avg,

wherein (R_Hist^avg, G_Hist^avg, B_Hist^avg) is obtained by converting said average pixel (Y_Hist^avg, C_rHist^avg, C_bHist^avg) of said candidate white pixels from said YC_rC_b color space to said RGB color space; and

taking (R_GWA, G_scale, B_scale) as said scale factor.

7. The white balance method according to claim 4, further comprising steps of:

determining said color cast as being greenish if R_Hist^avg>

G_Hist^avg>

B_Hist^avg,

wherein (R_Hist^avg, G_Hist^avg, B_Hist^avg) is obtained by converting said average pixel (Y_Hist^avg, C_rHist^avg, C_bHist^avg) of said candidate white pixels from said YC_rC_b color space to said RGB color space; and

taking (R_scale, G_GWA, B_scale) as said scale factor.