Image Coding with Discrete Transforms using efficient
energy-based adaptive zonal filtering
A. Palau and G. Mirchandani
ICASSP1994
In this paper, earlier work with adaptive zonal-coding is
extended to apply to a larger class of windows that
permit a better approximation of contours and adaptation to local
image content. For bit allocation, we apply a variable bit-rate
allocation procedure, specific to the zonal-window employed.
As opposed to traditional coding schemes where bit-allocation follows
block-transform coding, we develop a method where the selection
of the energy/zonal-windows and bit allocation procedures proceed
simultaneously. Results of both the sequential and simultaneous
optimizations are presented and compared to equivalent coding of
images obtained through other techniques. Including cost of bit overhead,
tests show a similar signal-to-noise ratio for the same compression factor.
It is also shown that our technique leads to a lowering of a local
error criterion which governs detail in low intensity regions.