(SEM VII) THEORY EXAMINATION 2024-25 DIGITAL IMAGE PROCESSING

DIGITAL IMAGE PROCESSING (KEC071) – COMPLETE SOLVED PAPER

Time: 3 Hours  Max Marks: 100
Instructions: Attempt all Sections

SECTION A (2 × 10 = 20 Marks)

Attempt all questions in brief

a) Grayscale image vs Binary image

Grayscale image: Pixel values range from 0–255 (8-bit), represents intensity levels.

Binary image: Pixel values are only 0 or 1 (black or white).

b) Quantization and its effect

Quantization maps continuous intensity values to discrete levels.
Effect: Higher quantization → better image quality; lower quantization → loss of detail (quantization noise).

c) Unsharp masking and application

Unsharp masking enhances edges by subtracting a blurred image from the original.
Application: Medical imaging, satellite image enhancement.

d) Purpose of bandpass filtering

Bandpass filtering allows a specific frequency band to pass while rejecting low and high frequencies.
Used to enhance edges and textures.

e) Two properties of Fourier Transform

Linearity

Frequency shifting (translation property)

f) One advantage and one limitation of DST

Advantage: Good energy compaction for certain images

Limitation: Not shift-invariant

g) Redundancy

Redundancy refers to repeated or unnecessary information in an image that can be removed without significant loss of quality.

h) Run-Length Encoding (RLE)

RLE is a lossless compression technique that stores pixel values as runs.
Used in: Fax images, bitmap graphics.

i) Principle of edge detection

Edges correspond to sudden changes in intensity. Edge detection identifies object boundaries using gradients.

j) Seed point in region growing

A seed point is the starting pixel from which a region grows based on similarity criteria.

SECTION B (10 × 3 = 30 Marks)

Attempt any three

a) Fundamental steps in Digital Image Processing

Image acquisition                                                    Image enhancement

Image restoration                                                    Color image processing

Image compression                                                  Image segmentation

Representation and description                                Recognition and interpretation

(Block diagram explanation included in exams)

b) Image smoothing using spatial domain techniques

Smoothing reduces noise using low-pass filters.
Examples:                                                                   Mean filter

Gaussian filter                                                            Median filter

c) Haar Transform

Haar Transform is a wavelet-based transform used for image compression.

Procedure:

Form Haar matrix                                                        Multiply image matrix with Haar matrix

Apply inverse transform for reconstruction

d) MPEG video compression standard

MPEG uses:                                                                 Temporal redundancy (motion compensation)

Spatial redundancy (DCT)                                  Applications: Video streaming, DVDs, multimedia systems.

e) Hough Transform

Used to detect parametric shapes.                          Detectable shapes:

Lines                                                                           Circles

Ellipses

SECTION C (10 × 5 = 50 Marks)

Attempt one from each question

Q3(a) Image file formats: JPEG, PNG, BMP, TIFF

Q3(b) Digital halftoning

Halftoning represents grayscale images using binary dots.
Significance: Used in printing and display devices.

Q4(a) Histogram-based image enhancement

Histogram equalization redistributes pixel intensities for better contrast.
Used in medical and low-light images.

Q4(b) Image restoration

Image restoration recovers degraded images.
Degradations include:                                                   Noise

Blur                                                                                   Motion distortion

Q5(a) Hadamard Transform

Orthogonal transform                                                       Fast computation

Used in image compression

Properties: Symmetry, orthogonality, energy preservation

Q5(b) Karhunen–Loève Transform (KLT)                      KLT is an optimal transform that decorrelates data.
Advantages:

Best energy compaction                                                   Used in image compression

Q6(a) Lossless vs Lossy compression

Q6(b) Huffman coding

Huffman coding assigns shorter codes to frequent symbols.
Steps:                                                                             Compute symbol probabilities

Build Huffman tree                                                        Assign binary codes

Q7(a) Region growing technique

Region growing groups neighboring pixels with similar properties.
Seed selection: Based on intensity, location, or user input.

Q7(b) Thresholding

Thresholding segments images based on intensity.