Previous: Scientific Description
Up: The CCD--IPCS Detector
Previous Page: Scientific Description
Next Page: Overall performance of the system

CCD--IPCS granularity and Dithering

The CCD-IPCS has sensitivity variations over its field of view which are due to physical irregularities in the photocathodes inside the intensifier. The spatial variation in sensitivity is referred to as `granularity' because of the grainy appearance it gives to images.

In order to reduce the effect of granularity due to last three stages of intensification, a scheme of `dithering' was introduced for the IPCS-I (Jorden & Fordham Q.J.R.A.S., 27, 166, 1986). By moving the image around on the last three photocathodes in the intensifier, the signal in a given pixel, averaged over time, is proportional to the mean sensitivity within the scanned area. Hence, the pixel-to-pixel granularity is (typically) reduced by a factor of two. A similar technique is used with the IPCS-II but this also counteracts the fixed pattern noise due to the centroiding algorithm used in that detector; the granularity of the first photocathode is not diminished by these techniques since the image on this photocathode cannot be scanned.

The effects of granularity may be reduced by dividing an acquired image by a flat--field, so compensating for the pixel--to--pixel variations in sensitivity. The exposure required for a `correct' flat-field depends on the spatial scale of both the granularity and the information content within an image. Jenkins ( MNRAS, 226, 341, 1987) analysed granularity using Fourier techniques and showed that most of the power is on scales of many pixels (>5); little power was seen on small scales (<5 pixels).

Flat-fielding is, therefore, highly desirable when the noise in the data due to photon noise is less than the power density of the granularity at the spatial frequencies of interest. To adequately describe the granularity on a scale of P pixels, typically a flat field of at least N counts per pixel should be taken, where .

Because there is little power at small scales, using a flat field with a small differential shift (<2 pixels) between the image and the flat-field will still improve rather than degrade signal--to--noise of the resulting image.

However, in practice obtaining IPCS-II flat-fields is very difficult, because the required exposure time would be many hours because of the low count rate limit of the CCD--IPCS.