(SEM V) THEORY EXAMINATION 2024-25 IMAGE PROCESSING

Subject Code: BCS057
Maximum Marks: 70
Time: 3 Hours
Paper ID: 310313

Question Paper Overview

SECTION A (2 × 7 = 14 Marks)

(Short conceptual questions – fundamental image processing concepts)

a. What is the RGB color model?
b. Define the 2D mathematical preliminaries in image processing.
c. Describe the purpose of spatial filtering in image enhancement.
d. Define image restoration in digital image processing.
e. Describe edge detection in image segmentation.
f. Define region-based segmentation.
g. Discuss the MPEG standard for video compression.

SECTION B (Attempt any three × 7 = 21 Marks)

a. Discuss the fundamentals of color images, including digital representation in RGB and HSI models.
b. Describe the Butterworth filter and its use in image enhancement.
c. What is the purpose of notch filters in image restoration, and how do they remove periodic noise?
d. Describe the thresholding technique used in image segmentation and how it separates objects from the background.
e. Explain how Run-Length Encoding (RLE) works in data compression.

SECTION C (Attempt one part from each question × 7 = 35 Marks)

Q3

(a) Describe the steps involved in digital image processing and their role in improving image quality.
OR
(b) Explain the Discrete Fourier Transform (DFT) and how it is used to transform images into the frequency domain.

Q4

(a) Differentiate between smoothing and sharpening spatial filters with examples.
OR
(b) Define homomorphic filtering and explain how it improves lighting and contrast in images.

Q5

(a) Describe Wiener filtering and how it minimizes noise in image restoration while preserving details.
OR
(b) Discuss the properties of image degradation that affect restoration and how they are corrected.

Q6

(a) Explain morphological processing in image segmentation and the role of erosion and dilation in refining features.
OR
(b) Describe dam construction in watershed segmentation and how it controls the flow of regions (watersheds).

Q7

(a) Discuss the JPEG standard for image compression, its compression techniques, and suitable image types.
OR
(b) Explain boundary description and its role in object boundary identification during image recognition.

Key Topics for Revision

1. RGB and HSI Color Models

RGB Model: Additive model using Red, Green, Blue components (used in monitors/cameras).

HSI Model: Represents Hue (color type), Saturation (purity), and Intensity (brightness). Useful for color-based segmentation.

2. Mathematical Preliminaries

Image represented as a 2D function f(x,y)f(x, y)f(x,y).

Operations include addition, subtraction, convolution, and transforms (Fourier, Laplacian).

3. Spatial Filtering

Used to enhance features or remove noise.

Smoothing filters: Average, Gaussian → reduce noise.

Sharpening filters: Laplacian, Sobel → enhance edges.

4. Image Restoration

Process of recovering a degraded image using mathematical models of degradation.

Common methods: Inverse filtering, Wiener filtering, Notch filtering.

5. Edge Detection & Segmentation

Detects intensity discontinuities (object boundaries).

Operators: Sobel, Prewitt, Canny.

Region-based segmentation: Divides image into homogeneous areas.

6. Frequency Domain Processing

Fourier Transform (FT): Converts spatial data into frequency components.

Filtering:

Low-pass: Blurs images (noise reduction).

High-pass: Sharpens images.

Notch filter: Removes periodic noise.

Butterworth filter: Smooth transition between pass and stop bands.

7. Homomorphic Filtering

Enhances contrast and lighting by separating illumination and reflectance.

Uses log and frequency domain transformations.

8. Morphological Processing

Deals with the shape of image structures.

Erosion: Removes small white noise, shrinks objects.

Dilation: Expands object boundaries, fills gaps.

Used in preprocessing and post-segmentation cleanup.

9. Watershed Segmentation

Treats grayscale image as a topographic surface.

“Dams” are constructed to prevent merging of adjacent regions.

Used for region separation and boundary extraction.

10. Wiener Filtering

Restores images degraded by noise and blur.

Balances noise suppression and detail preservation using power spectral densities.

11. Image Compression

12. Boundary Description

Represents the shape of objects via edges or contours.

Methods: Chain codes, Fourier descriptors, boundary segments.

Suggested Study Flow

Understand image representation → pixels, resolution, gray levels.

Study filtering operations (spatial & frequency domain).

Revise image enhancement and restoration methods.

Practice segmentation techniques (edge, region, morphological).

Study compression standards (JPEG, MPEG).

Revise applications in recognition and AI-based vision systems.