4 Object detection and photometry

The identification, photometry and classification of the objects have been performed with the program SExtractor (Source-Extractor, Bertin 1995).

First, a background map was constructed and subtracted from the image. The sky map is a bilinear interpolation between the meshes of a grid with a mesh size of $56 \times 56$ pixel (or $128 \times 128$ pixel in the case of the more extended Fornax dwarf ellipticals). In addition, a median filter of $5 \times 5$ pixel has been applied in order to suppress possible overestimations due to bright stars. Then the image was convolved with a Gaussian with a FWHM slightly larger than that for stellar images in order to favour the detection of marginally resolved objects. A FWHM of $2\hbox{$^{\prime\prime}$}$ was chosen for all fields except the background fields B3 and B4, which were convolved with a Gaussian of $1\hbox{$.\!\!^{\prime\prime}$}5$ FWHM. Objects were found with a threshold of about 2 sigmas above the sky level. The level of the lowest isophote in V and I above which objects were detected is given in Table 1 for the different fields. The minimum number of connected pixels for a detection was chosen to be 5 in all fields. Composite objects were deblended by a multithresholding algorithm. In all fields the identifications have been controlled by eye, obvious multi-identifications of the same object were removed and objects that had been missed by the finding algorithm were added. The number of added objects was always below 2% of the total number.

The finding completeness starts to drop in V at magnitudes between 22.0 and 22.5 mag (except the NGC 1399 SW field: 21.0) and in I at magnitudes between 21.0 and 21.5 mag (NGC 1399 SW field: 20.0) depending on the seeing and exposure times of the different fields. The finding limit for low surface brightnesses varies between 23.0 and 24.0 mag arcsec^-2 peak surface brightness in V ($\mu_{\rm peak}$ is the surface brightness of the central pixel as given by SExtractor). This latter limit is about the same as in the FCC, whereas the limiting total magnitude is about 2 magnitudes fainter than in the FCC.

The photometry was done via elliptical apertures whose ellipticity and position angle are defined by the second order moments of the light distribution. Total magnitudes are computed in two different ways. For isolated objects the flux is measured within an aperture calculated by a further development of Kron's "first moment'' algorithm (Kron 1980; see also Infante 1987). For overlapping objects, i.e. which have neighbours within the elliptical aperture, the fraction of flux lost by the isophotal magnitude (using the detection threshold as the lowest isophote) is estimated and corrected assuming that the intensity profiles have Gaussian wings because of atmospheric blurring.

The color determination was done by measuring aperture magnitudes from circular apertures with $3\hbox{$^{\prime\prime}$}$ diameter in both filters. Peak surface brightnesses ($\mu_{\rm peak}$ in mag arcsec^-2) are calculated from the peak intensity at the central pixel. Note that for objects with apparent core radii intrinsically smaller than the seeing the derived central surface brightness is a lower limit compared to the true central surface brightness due to the convolution with the seeing. A more detailed analysis of the surface brightness profiles of the brighter galaxies in our sample is given in Sect. 6.

The photometric calibration was done by applying the calibration equations for aperture photometry given in Kissler-Patig et al. (1997).