
Cameras play the important role in computer vision of transforming the signal
of interest in the 3-D world into the image that is processed by algorithms.
In reference [1], we carefully analyze the properties of CCD cameras from
the perspective of computer vision. This analysis forms the basis of methods
for noise estimation and radiometric calibration. These methods lead directly
to an algorithm that can be used to separate image variance due to camera noise
from image variance due to variation in the scene.
Last modified: 28 July 1998