P. Carlini. Self-organizing maps, vector quantization, and fractal image coding. In NATO ASI Conf. Fractal Image Encoding and Analysis, Trondheim, July 1995.
This paper reviews the properties of a popular kind of neutral network, called self organizing maps, then discusses their relevance for the field of lossy image coding. In particular, vector quantization (VQ) and fractal (block based) coding are studied from a common point of view, emphasing their relationships. The latter is an exotic kind of (non-causal) predictive-VQ; clustering algorithms, directly useful in the VQ field, can be exploited to reduce ist computational complexity.
@InProceedings{Carl95,
Author = {Carlini, P.},
Title = {Self-organizing maps, vector quantization, and fractal image coding},
BookTitle = {NATO ASI Conf. Fractal Image Encoding and Analysis},
Address = {Trondheim},
Month = {July},
Year = {1995}
}
appears in Fractals, Vol. 5, Supplementary issue, April 1997