ELEN0450 · MULTIMEDIA SYSTEMS
Sampling and quantization
Lecture 1 — Part B · Exercises
Gavage Cédric
REMINDER
Formula sheet
Sampling condition f_s ≥ 2 · f_max
Nyquist frequency f_N = 1 / (2 · T_s) = f_s / 2
Quantization levels N bits → 2^N levels
Dither levels n×n matrix → n² + 1 levels
Raw audio bit rate f_s × N × n_channels
Raw video bit rate W × H × bpp × fps
Unit conversion 1 byte = 8 bits
EXERCISE 1
Give the minimum sampling frequency for:
a) A telephone signal band-limited to 4 kHz
b) Music covering the full range of human hearing,
up to 20 kHz
c) A medical ultrasound signal occupying
the 2 – 10 MHz band
The same second of signal
a) Telephone speech
4 kHz
b) Music
20 kHz
c) Medical ultrasound
10 MHz
EXERCISE 2
A sunset sky fades smoothly from orange to blue. Store it on too few bits
and the gradient breaks into visible bands.
a) How many distinct levels for 1, 8, 10 and 16 bits?
b) A grayscale image has 4096 shades.
On how many bits is it coded?
c) A colour image is coded on 24 bits, 8 per channel.
How many colours can it represent?
What “N bits” actually means
1 bit
2 bits
3 bits
4 bits
EXERCISE 3
Call a friend and their voice sounds noisy. Put the same voice on a CD
and it sounds like they are in the room.
Compute the raw bit rate of each one.
What factor separates them, and where does it come from?
Two devices, two sets of numbers
Digital telephone
8 kHz
8 bits
mono
CD audio
44.1 kHz
16 bits
stereo
EXERCISE 4
A YouTube video is nothing but a stack of still pictures, thirty of them
every second. Let us open one up and count what is inside.
For a 640 × 480 video, colour at 8 bits per channel, 30 frames per second:
a) How many pixels does the frame contain in total, and how many bytes are required to store this single frame?
b) What is the video bit rate, expressed in MB/s and in Mb/s
c) How many bytes is it for an hour ?
Three sampling axes
W × H
space
fps
time