See also
[6]
Image Compression Through Level Lines and Wavelet Packets
[5]
Compression des images par ondelettes : intérêt d'une approche
géométrique
[4]
Un algorithme de compression d'images basé sur une
représentation mixte par lignes de niveau et paquets d'ondelettes
[3]
Perceptible Level Lines and Isoperimetric Ratio
[2]
A Compact and Multiscale Image Model Based on Level Sets
[1]
A Functional Analysis Model for Natural Images Permitting Structured Compression
Those papers are all about image representation models based on level lines.
In [1] (long paper), it is proved that the selection of the most perceptible level lines
can separate the sketch information from the microtextures of a natural image.
Application to image compression is introduced but not developed.
In [2], the sketch model is presented in a scale-space framework.
A criterion to select the most perceptible level lines which is based on the isoperimetric
ratio of shapes is introduced in [3] (short paper).
In [4], [5] (short papers) and [6] (long paper), the model detailed in [1] is applied to develop
structured image compression schemes. The article in [6] is basically the long version of the
conference proceeding [4], although the images are of different nature.
Papers [4] and [6] are about compression of natural images, where the sketch is compressed using
a geometric coder and where the microtextures are compressed within a wavelet packet basis.
The conference proceeding [5] illustrates how the model can be adapted to fit to non-natural images,
such as medical images (example of an angiography is given).