Point Operations Chapter 5 Gamma Gamma is a

Point Operations – Chapter 5

Gamma • Gamma is a form of point operation • It’s intended purpose is to compensate for the non-linear relationship between the image capture device and the image display and (ultimately) the human viewer • That’s the technical low down – realistically it can be viewed as nothing more than a way to enhance image intensities in a non-linear fashion

Gamma • The function is: • is called the “gamma value”

0 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176 183 190 197 204 211 218 225 232 239 246 253 Gamma • Various “gamma curves” 1. 2 1 0. 8 0. 6 0. 4 0. 2

Gamma • So, what does it really do? 1. 2 1 0. 8 0. 6 0. 4 0 1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 171 181 191 201 211 221 231 241 251 0. 2 • In “words”, what does this mean? – Dark input pixel values are amplified a lot – Bright input pixel values are amplified very little

Gamma • The steepness of the curve in the dark range can often be a problem – Low intensities is where most image sensors experience a great deal of noise – Over amplifying causes bad artifacts to appear • The solution is to split the curve into two sections, a linear portion and an exponential portion

Modified gamma The point where the curve transitions from linear to exponential

0 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176 183 190 197 204 211 218 225 232 239 246 253 Modified gamma 1. 2 1 0. 8 0. 6 0. 4 0. 2

Gamma correction

Implementation • Applying an exponential function to every pixel in a large image is very time consuming so we don’t do it • Instead, we create a look-up table (LUT) of all possible pixel input/output pairs • For an 8 -bit image this is only 256 calculations and memory locations • Image pixels are used as indexes into the table • This technique can be used for any function, not just gamma

Alpha blending • Sometimes we want to overlay one image onto another • The technique used is called alpha-blending • Realistically it’s nothing more than a pixel-bypixel weighted average • Graphics processing units have this as a built in function since graphics systems (e. g. games) do this often

Alpha blending • Setting alpha to 1 makes the foreground image completely transparent • Setting alpha to 0 makes the foreground image complete opaque • Setting it between 0 and 1 allows you to “see through” the foreground image into the background image

Alpha blending

Homework • See handout