Sony SNC-CS50P - Manual (Page 5)

5
H.264
H.264 (or MPEG-4 Part 10) has been developed with the
aim of providing high-quality video at a much lower bit rate
than MPEG-4. A number of techniques for achieving
efficient compression are incorporated in H.264. One major
contributing factor is the improvement in motion prediction.
As in the case of MPEG-4, each image is divided into
blocks to predict movement. However, with H.264, the
block patterns can be a 16 x 16 pixel macroblock or any
combination of the seven options shown in Fig. 4 (e.g. 4 x
4 sub-blocks in the upper right quadrant of the
macroblock, an 8 x 8 sub-block in the upper left
quadrant, and an 8 x 16 sub-block in the lower half, as
shown in Fig. 5). The block pattern is variably determined
depending on the amount and speed of movement within
the image. If an area of the image has little movement,
the algorithm utilizes large blocks (such as 16 x 16 pixels
or 8 x 8 pixels) to predict the difference between the
previous VOP and the current image. However, where an
area of the image includes significant motion, the
algorithm utilizes smaller blocks for prediction. By
dynamically adapting the size of each block to the
amount of motion, the prediction accuracy for each block
is significantly improved. Because the predicted data is
more accurate, less image data needs to be transmitted;
therefore, compression efficiency is greatly improved
when compared to MPEG-4.
Though motion prediction using variable block sizes
increases prediction accuracy and minimizes the amount
of data to be transmitted, it does require greater
processing power within the codec.
JPEG/MPEG-4/H.264 Comparison
The difference between JPEG, MPEG-4, and H.264
compression formats has been explained in the above
sections. Here, let‘s relate picture quality to transmission
bit rate.
Fig. 6 is a graph depicting the picture quality vs. the bit
rate of these three compression formats.
*4 The vertical
axis (PSNR level) expresses the picture quality, and the
horizontal axis expresses the transmission bit rate. PSNR
(Peak Signal-to-Noise Ratio) is a metric widely used by
engineers to measure the “quality” of compressed video
images.
At a PSNR of 35 dB, JPEG images are transmitted at
approximately 260 Kb/s, while MPEG-4 transmits at
approximately 85 Kb/s and H.264 transmits at 50 Kb/s. To
put this into perspective, MPEG-4 requires approximately
one-third of the bandwidth used by JPEG, and H.264
requires just one-fifth.
In summary, both MPEG-4 and H.264 are ideal for image
transfer over a network because they require much less
network bandwidth than JPEG.
Fig. 5 H.264 Combination Block Pattern
40
Video Parameters:
•10 frames/s
•QCIF (176 x 144 pixels)
•10 seconds of video (100 frames)
JPEG
PSNR
(dB)
Bit rate (Kb/s)
38
36
35
34
32
30
28
0 100 200 300
H.264
MPEG-4
Fig. 6 Comparison Between H.264, MPEG-4,
and JPEG (picture quality vs. bit rate)
16 pixels
16 pixels
8 pixels
4 pixels 8 pixels 4 pixels
4 pixels 4 pixels
8 pixels
16 pixels
16 pixels
8 pixels
8 pixels
8 pixels
Fig. 4 H.264 Motion Compensation Blocks
*4 The graph shows just one example of comparing bit rates at which JPEG, MPEG-4, and H.264 images can be transmitted. Actual bit rates for
transmitting data using these three compression formats differ with image quality and image size settings.