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

**A.2.3**

**Interleaved order (Ns > 1)**

When Ns > 1, each scan component Cs

i

is partitioned into small rectangular arrays of H

k

horizontal data units by V

k

vertical data units. The subscripts k indicate that H

k

and V

k

are from the position in the frame header component-

specification for which C

k

=

Cs

i

. Within each H

k

by V

k

array, data units are ordered from left-to-right and top-to-bottom.

The arrays in turn are ordered from left-to-right and top-to-bottom within each component.

As shown in the example of Figure A.3, Ns

=

4, and MCU

1

consists of data units taken first from the top-left-most region

of Cs

1

, followed by data units from the corresponding region of Cs

2

, then from Cs

3

and then from Cs

4

. MCU

2

follows the

same ordering for data taken from the next region to the right for the four components.

0

1

2

3

4

5

0

1

2

3

0

1

2

3

4

5

0

1

0

1

2

0

1

2

3

0

1

2

0

1

TISO0800-93/d016

Cs : H = 2, V = 2

Cs : H = 2, V = 1

Cs : H = 1, V = 2

Cs : H = 1, V = 1

1

1

1

1

2

2

2

3

3

3

4

4

4

MCU = d d d d d d d d d ,

MCU =

d d d d d d d d d ,

MCU = d d d d d d d d d ,

MCU = d d d d d d d d d ,

00

01

10

11

02

03

12

13

04

05

14

15

20

21

30

31

01

10

11

02

03

12

04

05

20

30

00

01

00

10

00

2

3

4

01

02

02

11

10

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

4

4

4

4

Cs data units

Cs

Cs

Cs

1

2

3

4

**Figure A.3 Interleaved data ordering example**

FIGURE A.3 [D16] 7,5 cm = 293 %

**A.2.4**

**Completion of partial MCU**

For DCT-based processes the data unit is a block. If x

i

is not a multiple of 8, the encoding process shall extend the number

of columns to complete the right-most sample blocks. If the component is to be interleaved, the encoding process shall also

extend the number of samples by one or more additional blocks, if necessary, so that the number of blocks is an integer

multiple of H

i

. Similarly, if y

i

is not a multiple of 8, the encoding process shall extend the number of lines to complete the

bottom-most block-row. If the component is to be interleaved, the encoding process shall also extend the number of lines

by one or more additional block-rows, if necessary, so that the number of block-rows is an integer multiple of V

i

.

NOTE It is recommended that any incomplete MCUs be completed by replication of the right-most column and the bottom

line of each component.

For lossless processes the data unit is a sample. If the component is to be interleaved, the encoding process shall extend

the number of samples, if necessary, so that the number is a multiple of H

i

. Similarly, the encoding process shall extend

the number of lines, if necessary, so that the number of lines is a multiple of V

i

.

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

**A.3**

**DCT compression**

**A.3.1**

**Level shift**

Before a non-differential frame encoding process computes the FDCT for a block of source image samples, the samples

shall be level shifted to a signed representation by subtracting 2

P 1

, where P is the precision parameter specified in B.2.2.

Thus, when P

=

8, the level shift is by 128; when P

=

12, the level shift is by 2048.

**26**

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