ISO/IEC 10918-1 : 1993(E)
H.1.2.3.3
Default conditioning bounds
The bounds, L and U, for determining the conditioning category have the default values L
=
0 and U
=
1. Other bounds
may be set using the DAC (Define-Arithmetic-Conditioning) marker segment, as described in Annex B.
H.1.2.3.4
Initial conditions for statistical model
At the start of a scan and at each restart, all statistics bins are re-initialized to the standard default value described in
Annex D.
H.2
Lossless decoder processes
Lossless decoders may employ either Huffman decoding or arithmetic decoding. They shall be capable of using up to four
tables in a scan. Lossless decoders shall be able to decode encoded image source data with any input precision from 2 to
16 bits per sample.
H.2.1
Lossless decoder control procedures
Subclause E.2 contains the decoder control procedures. In applying these procedures to the lossless decoder the data unit
is one sample.
When the decoder is reset in the restart interval control procedure (see E.2.4) the prediction is reset to the same value
used in the encoder (see H.1.2.1). If arithmetic coding is used, the statistics are also reset.
Restrictions on the restart interval are specified in H.1.1.
H.2.2
Coding model for lossless decoding
The predictor calculations defined in H.1.2 also apply to the lossless decoder processes.
The lossless decoders, decode the differences and add them, modulo 2
16
, to the predictions to create the output. The
lossless decoders shall be able to interpret the point transform parameter, and if non-zero, multiply the output of the
lossless decoder by 2
Pt
.
In order to avoid repetition, detailed flow charts of the lossless decoding procedures are omitted.
136
CCITT Rec. T.81 (1992 E)