Verfahren zur Kodierung von durch Vektoren dargestellten digitalen Dateien und geeignete Dekodierungsverfahren

摘要

A vector representative of a point in an animated image is generated by a quantification process carried out on a mesh in a space of n dimensions. The vector contains the values taken by an input signal (E) at the sampling point. The coding method consists in allocating to each vector a code (CN) representative of its magnitude, a second code CT, which identifies the type of vector, and a code CI (N,T) which identifies the vector relative to other vectors of same magnitude and type. The magnitude code CN is generated using a Huffman code calculated using the magnitude statistics. The type code is generated from the statistics of all the vectors and a Huffman code. The individual code CI(N,T) is generated assuming that all the vectors of same type have an equal probability.