We describe a general, coding strategy leading to a family of universal ima
ge compression systems designed to give good performance in applications wh
ere the statistics of the source to be compressed are not available at desi
gn time or vary over time or space. The basic approach considered uses a tw
o-stage structure in which the single source code of traditional image comp
ression systems is replaced with a family of codes designed to cover a larg
e class of possible sources. To illustrate this approach, we consider the o
ptimal design and use of two-stage codes containing collections of vector q
uantizers (weighted universal vector quantization), bit allocations for JPE
G-style coding (weighted universal bit allocation), and transform codes (we
ighted universal transform coding). Further, we demonstrate the benefits to
be gained from the inclusion of perceptual distortion measures and optimal
parsing. The strategy yields two-stage codes that significantly outperform
their single-stage predecessors. On a sequence of medical images, weighted
universal vector quantization outperforms entropy coded vector quantizatio
n by over 9 dB, On the same data sequence, weighted universal bit allocatio
n outperforms a JPEG-style code by over 2.5 dB. On a collection of mixed te
xt and image data, weighted universal transform coding outperforms a single
, data-optimized transform code (which gives performance almost identical t
o that of JPEG) by over 6 dB.