Use Method->"RGBScaling" to rescale the image in the RGB space:

When a strong color cast is present, scaling the pixels in the RGB space may not fix the color cast:

Use Method->"LMSScaling" to scale the pixels in the LMS space:

Use Method->"ChromaticityScaling" to make the average color neutral (a shade of gray):

Use "RGBScaling" with no reference or target colors:

This corresponds to transforming the max color value to 1 for each channel:

"ChromaticityScaling" typically creates a more balanced image:

Notice that it is computationally more expensive:

If the color is specified using temperature and tint, Method->"ChromaticityScaling" is used:

With no reference or target colors, available methods make assumptions about the image scene:

"RGBScaling" and "LMSScaling" assume that the image contains a white object:

"ChromaticityScaling" assumes that the average color is neutral (i.e. gray):

Correct a strong blue color cast, assuming there should be some white in the picture:

If white pixels are present already, Method"LMSScaling" will have little or no effect:

The default assumption of a neutral average color may yield extreme results:

Manually specify a region where the average color is computed:

The same scene appears differently depending on its illumination:

Use a low temperature to simulate early sunrise:

5500K is the typical vertical daylight temperature:

Use a higher temperature to simulate a hazy sky:

Correct for age yellowing in a document: