In the first paper of this series we use the publicly available code Gim2D to model the r - and i -band images of all galaxies in a magnitude-limited sample of roughly 1800 morphologically classified galaxies taken from the Sloan Digital Sky Survey .
The model is a concentric superposition of two components , each with elliptical isophotes with constant flattening and position angle .
The disk luminosity profile is assumed exponential , while the bulge is assumed to have a de Vaucouleurs or a Sérsic profile .
We find that the parameters returned by Gim2D depend little on the waveband or bulge profile used ; their formal uncertainties are usually small .
Nevertheless , for bright galaxies the measured distribution of b / a , the apparent disk flattening , deviates strongly from the expected uniform distribution , showing that the “ disk ” identified by the code frequently corresponds to an intrinsically 3-dimensional structure rather than to a true thin disk .
We correct approximately for this systematic problem using the observed statistics of the b / a distribution and estimate , as a function of absolute magnitude , the mean fractions of galaxy light in disks and in “ pure bulge ” systems ( those with no detectable disk ) .
For the brightest galaxies the disk light fraction is about 10 % and about 80 % are “ pure bulge ” systems .
For faint galaxies most of the light is in disks and we do not detect a “ pure bulge ” population .
Averaging over the galaxy population as a whole , we find that 54 \pm 2 \% of the local cosmic luminosity density at both r and i comes from disks and 32 \pm 2 \% from “ pure bulge ” systems .
The remaining 14 \pm 2 \% comes from bulges in galaxies with detectable disks .