Image Processing Reference
ment of an object centroid, and then the image would be moved so that light falls on
the correct location of the camera.
The orthogonal transfer CCD, developed by Tonry, Burke, and Schecter in 1997,
satisfies the requirement for providing image centroiding and long-term integration
on a single device. These devices rely on a modified gate structure in the CCD device
itself and a controller that allows the charge in the pixels to be moved around the chip.
The idea is analogous to moving a bucket around to catch the water falling from a
hose. As a result, light falling onto the CCD is always falling onto the right spot.
The orthogonal transfer CCD uses one part of the CCD as a position sensor to
detect image motion in a specific guide star, and the other portion as an imager,
where the entire image is adjusted based on the observed motion. The position sen-
sor is a minimum number of pixels that surround a guide star; it is read out at high
speed. The centroid position of the guide star is calculated and the image area
shifted to match the centroid location. The result is a long integration image that is
read out only once, with improved resolution as illustrated in Fig. 8.5.
The initial devices were small, but devices with imaging areas of 512
pixels have been constructed and used successfully. Results of using the orthogonal
Figure 8.5 Illustration of the operation of an orthogonal transfer CCD. Movement in
the position of the guide star is detected in one portion of the CCD frame and is used
to move all pixels in the imaging frame to stabilize the image. (Galaxy image cour-
tesy of NASA, Astronomy Picture of the Day.)