| George Wilson 2005-05-05, 8:55 am |
|
Thomas Richter wrote:
> Hi,
>
A[Q][color=darkred]
>
> I would say this is more or less an "ad hoc" choice made to
> generate suitable golomb codes for the first few symbols where
> the statistics of the source is not yet well-known and the encoder
> had no chance to adapt itself correctly to the statistics.
>
each[color=darkred]
>
> k is choosen such that the Golomb code is optimal (or near-optimal)
> for the statistics of the residuals for the given context. For the
> full length proof for this see the following:
>
> M.J. Weinberger, G. Seroussi, G. Sapiro: The LOCO-I Lossless Image
> Compression Algorithm: Principles and Standardization into JPEG-LS,
> IEEE Transactions on Image Processing, Vol. 9, No. 8, August 2000.
>
> See especially Thm. 1 in Chapter C. What this says is that a golomb
> code with the given parameter k is optimal for a Laplacian
distribution
> described by S_t and N_t aka the "A" register in JPEG-LS.
>
> Does this answer the question?
>
> So long,
> Thomas
Yes sure.
Thanks Thomas.
George Wilson
|