Sharp DV-500D Technical Manual page 16

DVD-Series
2. Compression Method of Moving Pictures
When you see a TV picture, you think of it as a sequentially moving picture, thanks to the
afterimage effect that "fools" our eyes. But in reality, what you are watching is a series of still
pictures that move sequentially and change at a steady rate.
Therefore, it recognizes monotonous and complicated parts on still picture frames (spatial
information). Than, it decreases the amount of data (higher compression) for monotonous parts
and increases the amount of data (lower compression) for complicated parts.
Also, since moving pictures are a succession of still pictures, the system compares each picture
to the ones before it and memorizes upcoming ones in order to detect how much current picture
moved. It then records only the moving part or parts (time information) as data. Reproduction
is done in reverse (encode) in order to replace data (decode).
3. Steps in Spatial Information Compression (Still Picture)
(1) DCT (Discrete Cosine Transformation)
The picture is divided into small micro blocks. Each block is again divided into smaller
blocks. (See the drawing below.)
Then, the complication of each picture is detected on each block, converting it into DCT
= frequency, in order to sort monotonous parts (unnecessary information) from compli-
cated parts (necessary information) by numeric value.
(2) Quantization
The data converted into numeric value (frequency) by DCT will be more compressed in its
flat parts, where more information stays, by dividing it by a common value to make it
quantitated.
22 macro blocks
1 picture macro block
16
1 2
16
3 4
Macro block
A This part is even,
so it is easily compressed.
16
8
8
Block
B This part is complicated,
so it is difficult to compress.