feedback algorithm with “infinite” gain in the loop
(simulation)
The feedback was
simulated with K=50000 (see p.5). It was sufficient to achieve perfect
discounting the illumination.
|
Edwin H. Land's mondrian under “white” illumination |
Edwin H. Land's mondrian under “dim” illumination |
Edwin H. Land's mondrian under “yellow” illumination |
Uniform gray surface under “yellow” illumination |
 |
 |
 |
 |
 |
 |
|
|
|
 |
The feedback algorithm with amplification in the loop performs perfect colour
constancy: returns output images which do not depend on the illumination.
|
|
