Page 140 - Computational Colour Science Using MATLAB

8

Characterization of Cameras

8.1 Introduction

Some general comments regarding characterization can be found in Chapter 7,
Section 7.1. For input devices such as scanners and cameras it is important to
note that effective characterization is only practicable if the device does not
perform automatic white-point balancing. Automatic white-point balancing is
where the RGB values of each pixel in the captured image are transformed so
that the pixel for the brightest patch in the image scene is denoted as white with
equal RGB values (normally R ¼ G ¼ B ¼ 255). White-point balancing can be
useful if the aim is to capture a pleasing image, since the human visual system is
able to discount the colour of the light source so that surfaces tend to retain their
daylight appearance. For colorimetric characterization, however, this setting
should be disabled if at all possible.
The most efficient method for characterizing a digital camera or scanner is to
image a chart containing a set of colours of known tristimulus values (Johnson,
2002). Such charts commonly include neutral patches that may be used to
linearize the camera RGB outputs and coloured patches that may be used to
characterize a transform from linearized RGB values to CIE XYZ values. In the
late 1980s a working group of the ANSI IT8 (Image Technology Committee
No. 8) was created to define standard targets to be used in the characterization
of scanners and printers (McDowell, 2002). The IT8 committee chose to
colorimetrically define the colours that should appear in the target, but then
allow individual manufacturers to produce targets to meet these requirements.
Two standards were developed, ANSI IT8.7/1 and ANSI IT8.7/2, for
transmission and reflectance modes, respectively, and they were combined into
a single ISO standard (ISO 12641:1997). Two further charts that are sometimes
used for device characterization are the Macbeth ColorChecker chart (which
contains 24 patches) and the Macbeth ColorChecker DC chart (which contains
over 200 patches).

Computational Colour Science Using MATLAB. By Stephen Westland and Caterina Ripamonti.
& 2004 John Wiley & Sons, Ltd: ISBN 0 470 84562 7