ISO/IEC 10918-1 : 1993(E)
MSP(S) = 0
Figure D.2 Code_0(S) procedure
Figure D.2 [D40], = 9 cm = 352 %
The context-index S selects a storage location which contains Index(S), an index to the tables which make up the
probability estimation state machine. When coding a binary decision, the symbol being coded is either the more probable
symbol or the less probable symbol. Therefore, additional information is stored at each context-index identifying the sense
of the more probable symbol, MPS(S).
For simplicity, the flow charts in this subclause assume that the context storage for each context-index S has an additional
storage field for Qe(S) containing the value of Qe(Index(S)). If only the value of Index(S) and MPS(S) are stored, all
references to Qe(S) should be replaced by Qe(Index(S)).
The Code_LPS(S) procedure normally consists of the addition of the MPS sub-interval A Qe(S) to the bit stream and a
scaling of the interval to the sub-interval, Qe(S). It is always followed by the procedures for obtaining a new LPS
probability estimate (Estimate_Qe(S)_after_LPS) and renormalization (Renorm_e) (see Figure D.3).
However, in the event that the LPS sub-interval is larger than the MPS sub-interval, the conditional MPS/LPS exchange
occurs and the MPS sub-interval is coded.
The Code_MPS(S) procedure normally reduces the size of the probability interval to the MPS sub-interval. However, if
the LPS sub-interval is larger than the MPS sub-interval, the conditional exchange occurs and the LPS sub-interval is
coded instead. Note that conditional exchange cannot occur unless the procedures for obtaining a new LPS probability
estimate (Estimate_Qe(S)_after_MPS) and renormalization (Renorm_e) are required after the coding of the symbol (see
CCITT Rec. T.81 (1992 E)