Gamma correction of images is used to optimize the usage of data type depth when encoding an image by taking advantage of the non-linear manner in which humans perceive light and color. This non-linearity must be compensated to achieve correct color reproduction. To do this, luminance of each of the linear red, green, and blue components is reduced to a non-linear form using an inverse transformation. This process is called

gamma correction

.

The Intel IPP functions use the following basic equations to convert an RGB image to a gamma-corrected R'G'B' image:

for

R

,

G

,

B

< 0.018

for

R

,

G

,

B

≥

0.018

Note that the channel intensity values are normalized to fit in the range [0..1]. The gamma value is equal to 1/0.45 = 2.22 in conformity with ITU Rec.709 specification (see [ITU709]).