3 The image database

3.1 Database summary

The ALADIN image dataset consists of:

• The whole sky image database from the first Digitized Sky Survey (DSS-I) digitized from photographic plates and distributed by the Space Telescope Science Institute (STScI) as a set of slightly compressed FITS images (with a resolution of $1.8\hbox{$^{\prime\prime}$ }$); DSS-II is also currently being integrated into the database (see below);

• Images of crowded fields (Galactic Plane, Magellanic Clouds) at the full resolution of $0.67\hbox{$^{\prime\prime}$ }$, scanned at the Centre d'Analyse des Images (MAMA machine) in Paris;

• Global plate views ( $5\hbox{$^\circ$ }\times 5\hbox{$^\circ$ }$or $6\hbox{$^\circ$ }\times 6\hbox{$^\circ$ }$ according to the survey) are also available for all the plates contributing to the image dataset: these are built at CDS by averaging blocks of pixels from the original scans;

• Other images sets, or user-provided images, in FITS format, having suitable World Coordinate System information in the header (see e.g. Greisen & Calabretta [1995]); this functionality is currently available only for the Java stand-alone version.

3.2 Building the database contents

The ALADIN project has set up collaborations with the major groups providing digitizations of sky surveys. The original surveys are made of photographic Schmidt plates obtained at Palomar in the North, and ESO/SERC in the South, and covering the whole sky at different epochs and colours (see e.g., MacGillivray [1994]).

The database currently includes the first Digitized Sky Survey (DSS-I) produced by the Space Telescope Institute (Lasker [1992]), for the needs of the Hubble Space Telescope. To create these images, the STScI team scanned the first epoch (1950/1955) Palomar E Red and United Kingdom Schmidt J Blue plates (including the SERC J Equatorial Extension and some short V-band plates at low galactic latitude) with a pixel size of $1.7\hbox{$^{\prime\prime}$ }$ ($25~{\mu}$m). The low resolution and a light data compression (factor of 10) permit storage of images covering the full sky on a set of 102 CD-ROMs.

DSS-II images in the R-band (from Palomar POSS-II F and UK Schmidt SES, AAO-R, and SERC-ER), scanned with a $1\hbox{$^{\prime\prime}$ }$ ($15~{\mu}$m) sampling interval (see Lasker [1994]) are gradually being included into the system, and will soon be followed by DSS-II images in the B-band (POSS-II J).

In addition, high resolution digitalization of POSS-II, SERC-J, SERC-SR, SERC-I, or ESO-R plates featuring crowded regions of the sky (Galactic Plane and Magellanic Clouds) have been provided by the MAMA facility at the Centre d'Analyse des Images (CAI), Observatoire de Paris (Guibert [1992]). Sampling is $0.67\hbox{$^{\prime\prime}$ }$ per pixel ($10~{\mu}$m). Currently, these high resolution images cover about 15% of the sky, and are stored in a juke-box of optical disks, with a capacity of 500 Gigabytes.

3.3 The image server

The image server for ALADIN had to be able to deal with various survey data, in heterogeneous formats (uncompressed FITS, compressed JPEG or PMT - see Sect. 5, etc.). For that, an object-oriented design was chosen, allowing an easy manipulation of image calibrations and headers, through the use of object classes. Image compression or decompression, image reconstruction, and in a near future, part of the recalibration, are seen as class methods.

Images are currently divided into subimages of 500 $\times$ 500 pixels (DSS-I), $768 \times 768$ pixels (DSS-II), or $1024 \times 1024$ pixels (MAMA).

The 1.5 million subimages are described by records stored in a relational database, encapsulated by several classes of the image management software. When an image of the sky is requested, the original subimages containing the corresponding sky area are retrieved through SQL commands, and the resulting image is built on the fly.