## Info

Old Pixel Value p Gamma

Sawtooth

Inverse Linear

Old Pixel Value p Logarithmic

Old Pixel Value p Gammalog

Old Pixel Value p Logarithmic

Old Pixel Value p Gammalog

Old Pixel Value p

Old Pixel Value p

Figure 13.7 The transfer function can assume many different forms—linear, sawtooth, inverse linear, gamma curve, logarithmic, gammalog. The graph can rise, fall, or curve. The only constraint is that for each old pixel value, there must be only one new pixel value—the transfer function must be single-valued.

Old Pixel Value p

Old Pixel Value p

### Old Pixel Value p

Figure 13.7 The transfer function can assume many different forms—linear, sawtooth, inverse linear, gamma curve, logarithmic, gammalog. The graph can rise, fall, or curve. The only constraint is that for each old pixel value, there must be only one new pixel value—the transfer function must be single-valued.

F = 1 CASE pv > 100 F = 2 END SELECT END FUNCTION F

When the computer calls F (pv), the function computes a new value that depends on the input value of pv.

In image processing software, transfer functions can be precomputed and the results stored in an array called a look-up table. The reason for this is simple: speed. If you are going to determine a new brightness for every pixel in an image that is 1024 x 1024 pixels on a side, you would need to call the transfer function 1,048,576 times. However, if you construct a look-up table for 65,536 pixel values, you have to call the look-up table 1,048,576 times, but each of the lookups is nearly instantaneous.

Look-up tables are called LUTs for short. Here is a short section of a LUT

for the function in the example above:

p

fip)