MPEG-2 Video Coding - Digital Video and Audio Broadcasting Technology - page 120

for digital cameras and produces excellent picture quality. JPEG stands for
Joint Photographic Experts Group, i.e. a group of experts in the coding of
still frames. The basic algorithm used in JPEG is the Discrete Cosine
Transform (Fig.7.13.), or DCT in brief. This DCT also forms the central
algorithm for the MPEG video coding method.

y

x

0 1 2 3 4 5 6 7

1

Π k ( z + ) N − 1

2

∑

F = f cos( );

k z

N

z = 0

DCT IDCT

1

Π k ( z + )

2

N − 1

2

∑

f = F cos( );

z k

N N

k = 0

250

200

150

110

100

50

50

20 10 3 20 8
0 1 234567 n

10

Fig. 7.14. One-dimensional Discrete Cosine Transform

The human eye perceives fine structures in a picture differently from
coarse structures. In analog video test engineering it was already known
that low-frequency picture disturbances, i.e. picture disturbances which
correspond to coarse image structures or interfere with these are perceived
more readily than high-frequency disturbances, i.e. those corresponding to
fine image structures or interfering with these.