**ISO/IEC 10918-1 : 1993(E)**

After a non-differential frame decoding process computes the IDCT and produces a block of reconstructed image samples,

an inverse level shift shall restore the samples to the unsigned representation by adding 2

P 1

and clamping the results to

the range 0 to 2

P 1

.

**A.3.2**

**Orientation of samples for FDCT computation**

Figure A.4 shows an image component which has been partitioned into

8

×

8

blocks for the FDCT computations. Figure

A.4 also defines the orientation of the samples within a block by showing the indices used in the FDCT equation of A.3.3.

The definitions of block partitioning and sample orientation also apply to any DCT decoding process and the output

reconstructed image. Any sample added by an encoding process to complete partial MCUs shall be removed by the

decoding process.

C

i

00

s

s

s

s

s

s

s

s

s

01

10

11

70

71

77

17

07

TISO0810-93/d017

Top

Left

Right

Bottom

**Figure A.4 Partition and orientation of 8 x 8 sample blocks**

FIGURE A.4 [D17] 6 cm = 234 %

**A.3.3**

**FDCT and IDCT (informative)**

The following equations specify the ideal functional definition of the FDCT and the IDCT.

NOTE These equations contain terms which cannot be represented with perfect accuracy by any real implementation. The

accuracy requirements for the combined FDCT and quantization procedures are specified in Part 2 of this Specification. The accuracy

requirements for the combined dequantization and IDCT procedures are also specified in Part 2 of this Specification.

FDCT:

IDCT:

*S*

*C*

*C*

*s*

*x*

*u*

*y*

*v*

*s*

*C C*

*S*

*x*

*u*

*y*

*v*

*vu*

*u*

*v*

*x*

*y*

*yx*

*yx*

*u*

*v*

*u*

*v*

*vu*

=

+

+

=

+

+

=

=

=

=

1

4

2

1

16

2

1

16

1

4

2

1

16

2

1

16

0

7

0

7

0

7

0

7

cos

(

)

cos

(

)

cos

(

)

cos

(

)

where

*C*

*C*

*u v*

*u*

*v*

,

=

=

1

2

0

for

,

*C*

*C*

*u*

*v*

,

=

1 otherwise

otherwise.

**A.3.4**

**DCT coefficient quantization **

(informative)** and dequantization **(normative)

After the FDCT is computed for a block, each of the 64 resulting DCT coefficients is quantized by a uniform quantizer.

The quantizer step size for each coefficient *S*

*vu*

is the value of the corresponding element *Q*

*vu*

from the quantization table

specified by the frame parameter *Tq*

*i*

(see B.2.2).

**CCITT Rec. T.81 (1992 E)**

**27**