SILIGURI INSTITUTE OF TECHNOLOGY

Salbari, Sukna, Siliguri – 734009

Department of Computer Science & Engineering

Student Name Ratnojit Saha

Department CSE Roll No. 11900122167
Year 4th Semester 7th Section C
Subject Name Multimed ia Systems
Subject Code OEC-CS701B
Faculty Name Anupam Mukherjee

DECLARATION BY STUDENT

I certify that this assignment is my own work in my own words.

Signature of Student: ----------------------------------

1. Answer the following: (5 marks — short answers)

a) What is MIDI?

Ans : MIDI (Musical Instrument Digital Interface) is a standardized protocol for communicating
musical performance data between electronic musical instruments, computers, and controllers.
It transmits event messages (note on/off, pitch, velocity, control changes) rather than audio,
enabling control and sequencing of instruments.

b) State the Nyquist sampling theorem.

Ans : Nyquist sampling theorem: A continuous-time bandlimited signal with maximum
frequency fmaxf_{max}fmax can be completely reconstructed from its samples if sampled at a
rate fsf_sfs such that fs>2fmaxf_s > 2 f_{max}fs>2fmax. The minimum sampling rate 2fmax2
f_{max}2fmax is called the Nyquist rate.

c) What is quantization error?

Ans : Quantization error (or quantization noise) is the difference between a continuous (or high-
precision) signal value and the nearest representable discrete level after quantization. It is
introduced when mapping analog/continuous amplitude values to a finite set of digital levels,
causing rounding or truncation error.

d) What is hypermedia?
Ans : Hypermedia is an extension of hypertext that integrates multiple types of media (text,
images, audio, video, animations) linked via hyperlinks. It allows non-linear navigation through
interconnected multimedia content (e.g., the web with multimedia pages and links).

2. What is data compression? Difference between lossy and lossless techniques.

Ans:

Data compression:
Compression is the process of encoding information using fewer bits than the original
representation by removing redundancy or irrelevancy. It reduces storage space and
transmission time.

Difference — Lossless vs Lossy

• Lossless compression

o Definition: Original data can be perfectly reconstructed from compressed data

(no information loss).
 o Examples: Huffman coding, LZW, PNG, FLAC (audio).

o Use cases: Text, executable files, source code, medical images where exact
recovery is required.

o Pros/Cons: Exact recovery; typically lower compression ratio than lossy.

• Lossy compression

o Definition: Some information is discarded during compression; reconstruction is

approximate (irreversible).

o Examples: JPEG (images), MP3 (audio), MPEG (video).

o Use cases: Multimedia (images, audio, video) where perceptual quality is

preserved and higher compression is desired.

o Pros/Cons: Higher compression ratio; possible quality degradation/artifacts.

3. Huffman coding for given frequencies — construct tree & derive codes.

Frequencies: A: 0.25, B: 0.10, C: 0.20, D: 0.15, E: 0.26, F: 0.04

Ans:

Steps (brief):

1. List symbols with probabilities and build a min-heap by frequency.

2. Repeatedly remove two lowest-frequency nodes, combine them into a new node (sum
of frequencies), and insert back.

3. Continue until one node (root) remains — this forms the Huffman tree.

4. Assign binary digits by traversing: usually left = 0, right = 1. Code for each symbol is the
path from root to its leaf.

Combining order (example):

• Combine F(0.04) + B(0.10) → node(0.14)

• Combine node(0.14) + D(0.15) → node(0.29)

• Combine C(0.20) + A(0.25) → node(0.45)

• Combine E(0.26) + node(0.29) → node(0.55)

 • Combine node(0.45) + node(0.55) → root(1.00)

Derived Huffman codes (one valid assignment):

• C : 00

• A : 01

• E : 10

• F : 1100

• B : 1101

• D : 111

4. Explain RGB and CMYK colour models and why they are complementary.

Ans:

RGB (Red, Green, Blue)

• Type: Additive colour model.

• Principle: Colours are produced by adding light of red, green, and blue primaries.
Combining all at full intensity yields white; absence yields black.

• Use cases: Displays and devices that emit light (monitors, TVs, phone screens,
projectors).

• Representation: Typically 8-bit per channel → 24-bit colour (0–255 per channel).

CMYK (Cyan, Magenta, Yellow, Key/Black)

• Type: Subtractive colour model.

• Principle: Colours are produced by subtracting (absorbing) portions of white light via
inks/paints: cyan absorbs red, magenta absorbs green, yellow absorbs blue. Combining
all ideally produces black (in practice a dark brown, so black K is added).

• Use cases: Print media (inkjet, offset printing).

• Components: C, M, Y for colour mixing; K (black) for depth and ink efficiency.

Why complementary:
 • The primary colours in one model correspond to the complements in the other (R ↔ C,
G ↔ M, B ↔ Y). RGB specifies emitted-light colour while CMYK specifies pigment/ink
absorption. A colour displayed in RGB must be converted to CMYK for printing because
printers reproduce colour by subtractive mixing; the conversion is not exact due to
gamut differences (prints often have smaller gamut).

5. Steps of JPEG compression technique.

Ans:

JPEG compression pipeline (baseline JPEG, stepwise):

1. Color space conversion: Convert input from RGB to YCbCr (luminance Y and
chrominance Cb, Cr). Human vision is more sensitive to luminance than chroma.

2. Downsampling (optional): Reduce chroma resolution (e.g., 4:2:0) because humans are
less sensitive to colour detail.

3. Block splitting: Divide image into 8×8 pixel blocks (for each channel).

4. Level shift: Subtract 128 from each pixel value to center values around zero (range
−128..127).

5. Discrete Cosine Transform (DCT): Apply 2D DCT to each 8×8 block to convert spatial
pixels to frequency-domain coefficients.

6. Quantization: Divide DCT coefficients by quantization matrix values (lossy step). Many
high-frequency coefficients become zero — main source of compression & loss.

7. Zig-zag scanning: Reorder 8×8 coefficients into a 1D sequence that groups low-frequency
first and high-frequency later to maximize runs of zeros.

8. Entropy coding: Apply lossless coding (e.g., Huffman coding or arithmetic coding) to the
serialized coefficients and run-lengths (RLE for zero runs).

9. File formatting: Store compressed data plus headers (quantization tables, Huffman
tables, image metadata) in JPEG file format.

Decompression reverses these steps (entropy decoding → inverse zig-zag → dequantize →

inverse DCT → add 128 → combine channels → convert back to RGB). Quality level is controlled
via quantization tables.