Automatic white balancing of a digital image

A method for automatic white balancing of a full-color input image, the method comprising the acts of using a processor to:

• select pixels of the input image according to one or more predetermined criteria;
• determine a global reference color vector of the input image using the selected pixels;
• determine a final gain value for the R and B components of each individual pixel of the input image by combining both the global reference color vector and a local color vector corresponding to the individual pixel;
• and
• produce a white-balanced output image by adjusting the R and B components of each pixel in the input image using the corresponding final gain value;
• and
• wherein producing a white-balanced output image by adjusting the R and B components of each pixel in the input image using the corresponding final gain value comprises producing a white-balanced output image made up of pixels y(r,s)=[y(r,s)1, y(r,s)2, y(r,s)3]T according to the following equations:
• y(r,s)1=ƒ(x(r,s)1,λg({circumflex over (x)}(r,s)2,{circumflex over (x)}(r,s)1)+(1−λ)g( x2, x1)),
• y(r,s)2=x(r,s)2, and
• y(r,s)3=ƒ(x(r,s)3,λg({circumflex over (x)}(r,s)2,{circumflex over (x)}(r,s)3)+(1−λ)g( x2, x3)),
• where:
• ƒ(•) denotes an adjustment operation and g(•) denotes a gain formulation;
• x(r,s)=[x(r,s)1,x(r,s)2,x(r,s)3]T with components x(r,s)k is a pixel of the input image X;
• x=[ x1, x2, x3]T with components xk is the global reference vector
• x
• _
• k
• =
• 1
• 
• ζ
• 
• ⁢
• ∑
• (
• r
• ,
• s
• )
• ∈
• ζ
• ⁢
• x
• (
• i
• ,
• j
• )
• ⁢
• k
• ,
• ⁢
• for
• ⁢
• ⁢
• k
• =
• 1
• ,
• 2
• ,
• 3
• ;
• {circumflex over (x)}(r,s)=[{circumflex over (x)}(r,s)1,{circumflex over (x)}(r,s)2,{circumflex over (x)}(r,s)3]T with components {circumflex over (x)}(r,s)k is a reference color vector which combines properties of x and x(r,s);
• K1 is the number of rows in the input image X;
• K2 is the number of columns in the input image X;
• r=1, 2, . . . , K1 and s=1, 2, . . . , K2;
• k=1, 2, 3 correspond to the R, G, and B components of a pixel, respectively;
• and
• ζ corresponds to a set of pixels selected according to one or more predefined criteria, with |ζ| denoting the number of pixels inside ζ.

A method for automatic white balancing of an input color filter array (CFA) single-sensor image, the method comprising the acts of using a processor to:

• select pixels of the input image according to one or more predetermined criteria;
• determine global spectral characteristics of the input image using color component-wise averaging of the selected pixels;
• determine global gain values as a function of mean chrominance and luminance values;
• determine local spectral characteristics of the input image using samples from a neighborhood of a pixel under consideration and estimating a local luminance component in chrominance locations;
• determine local gain values as a function of the local luminance component and the pixel under consideration;
• determine final gain values for the chrominance components of the input image based on contributions of the local and global gain values;
• and
• produce a white-balanced output image by adjusting the input image components using the corresponding final gain values.

One or more non-transitory computer-readable media having computer-readable instructions thereon which, when executed, implement a method for automatic white balancing of an input color filter array (CFA) single-sensor image, the method comprising the acts of:

• selecting pixels of the input image according to one or more predetermined criteria;
• determining global spectral characteristics of the input image using color component-wise averaging of the selected pixels;
• determining global gain values as a function of mean chrominance and luminance values;
• determining local spectral characteristics of the input image using samples from a neighborhood of a pixel under consideration and estimating a local luminance component in chrominance locations;
• determining local gain values as a function of the local luminance component and the pixel under consideration;
• determining final gain values for the chrominance components of the input image based on contributions of the local and global gain values;
• and
• producing a white-balanced output image by adjusting the input image components using the corresponding final gain values.