Image Processing Reference
In-Depth Information
Color separation by prism
R-sensor
R light
G light
G-sensor
B light
B-sensor
FIGURE 8.11
Schematic diagram of three-sensor camera system.
pastels (top and bottom, respectively). It seems that the border between photography and
computer graphics is blurry. While this is allowed for fun, should some kind of guidelines
be set for photography?
8.3 Three-Chip Color Camera System
Thus far, descriptions have involved a single-sensor color camera system, which is
overwhelmingly dominant in the market because of its low cost and compact system size.
Disregarding cost and size, a three-sensor system is the mainstream choice for applications
in which picture quality is a high priority. In this system, color separation is implemented
by a prism and not by a color filter like in a single-sensor camera system.
As shown in Figure 8.11, light is guided to each of the three sensors assigned as R, G,
and B light sensors by reflections at the prism faces. No color filter is set on each sensor.
In a single-sensor camera system, only the light that passes through the color filter can
contribute as sensitivity in the sensor. The other light is absorbed or reflected by the color
filter. However, the color filter's role is to select light. Only one-third of incident light can
pass through the color filter and work, making two-thirds unavailable.
As almost all light reaches one of the sensors in a three-sensor camera, there is no such
waste. In this system, accurate sensor alignment is necessary so that R, G, and B light from
the same portion of the object arrives at the pixel of the same position in each color sensor.
Another advantage of this system is that demosaicking is unnecessary, because R, G, and
B light arrives at each pixel in each sensor. As there is no need for signal generation by dem-
osaicking, there is no possibility of it causing false signals, leading to higher picture quality.
Reference
1. T. Kuroda, The 4 dimensions of noise, IEEE International Solid-State Circuits Conference, San
Francisco, February 2007, Imaging Forum: Noise in imaging systems, pp. 1-33.
