(SEM VII) THEORY EXAMINATION 2022-23 DIGITAL IMAGE PROCESSING

DIGITAL IMAGE PROCESSING (KEC071)

B.Tech SEM VII – Complete Solved Question Paper (2022–23)
                                                                                                                 ⏱ Time: 3 Hours | 📊 Marks: 100

SECTION A

Attempt all questions in brief (2 × 10 = 20 marks)

(a) Define image segmentation

Image segmentation is the process of dividing an image into meaningful and non-overlapping regions based on characteristics such as intensity, color, texture, or boundaries, so that objects in the image can be analyzed easily 

(b) Advantages of Wiener Filter

The Wiener filter reduces noise and blur simultaneously. It minimizes the mean square error between the original image and the restored image and performs well when noise statistics are known 

(c) Explain Hit-or-Miss transformation

Hit-or-Miss transformation is a morphological operation used to detect specific shapes or patterns in a binary image by matching foreground and background pixels using structuring elements 

(d) What is meant by pixel depth?

Pixel depth refers to the number of bits used to represent each pixel in an image. Higher pixel depth allows more intensity levels and better image quality 

(e) Need of picture compression

Picture compression reduces storage space and transmission bandwidth by removing redundant and irrelevant information while maintaining acceptable image quality 

(f) Explain Homomorphic filtering

Homomorphic filtering separates illumination and reflectance components of an image. It enhances contrast and normalizes brightness by applying frequency-domain filtering 

(g) Properties of Slant Transform

Slant transform has good energy compaction, orthogonality, and is suitable for images with linear intensity variations, making it useful in image compression 

(h) Operating modes of JPEG format

JPEG operates in:                                        Baseline mode

Progressive mode                                       Lossless mode

These modes support different compression and transmission requirements 

(i) Color image smoothing

Color image smoothing reduces noise in color images by applying filters independently or jointly to RGB or other color components while preserving edges 

(j) Problems in region-based segmentation

Problems include over-segmentation, sensitivity to noise, difficulty in defining homogeneity criteria, and high computational complexity 

SECTION B

Attempt any THREE (10 × 3 = 30 marks)

(a) Application areas of image processing

Image processing is widely used in medical imaging, remote sensing, satellite imaging, industrial inspection, biometric systems, robotics, surveillance, computer vision, and multimedia applications. It improves analysis, automation, and decision-making across domains 

(b) Why Hadamard Transform is suitable for DIP

Hadamard transform uses only addition and subtraction operations, making it computationally efficient. It provides good energy compaction and is useful in image compression and noise reduction.

Mathematically, the Hadamard matrix is orthogonal and recursive, allowing fast computation 

(c) Difference between image restoration and enhancement

(d) Short notes

(i) Band-pass filter technique
Band-pass filters allow frequencies within a specific range, removing low-frequency background noise and high-frequency noise, useful in edge enhancement.

(ii) Minimum error square filtering
This filter minimizes mean square error between original and restored image and is widely used in image restoration problems 

(e) Working of color picture histogram processing

Color histogram processing redistributes intensity values of color components to enhance contrast. It can be applied separately on RGB channels or on intensity component in color spaces like HSI 

SECTION C

Q3 (Attempt any one)

(a) Sampling and Quantization

Sampling converts a continuous image into discrete pixels, while quantization assigns intensity values to pixels.

Types:

Uniform and non-uniform sampling                       Uniform and non-uniform quantization

Both affect image resolution and quality 

(b) Properties of Fourier Transform

Fourier Transform properties include linearity, symmetry, periodicity, convolution theorem, and frequency shifting. It helps analyze image frequency components for filtering and restoration 

Q4 (Attempt any one)

(a) Image smoothing using low-pass filters

Ideal and Butterworth low-pass filters smooth images by removing high-frequency noise.

Ideal LPF has sharp cutoff but causes ringing

Butterworth LPF provides smoother transition

(b) Adaptive filters and adaptive median filter

Adaptive filters change behavior based on local image statistics. Adaptive median filter removes impulse noise effectively while preserving edges better than standard median filter 

Q5 (Attempt any one)

(a) Image compression system                             Image compression system consists of:

Source encoder                                                    Quantizer

Entropy encoder

(Draw functional block diagram in exam)

It reduces redundancy for efficient storage and transmission 

(b) Conversion from HSI to RGB

Steps:                                                                       Convert hue and saturation

Compute RGB components based on hue sector   Normalize RGB values

This conversion is used for display purposes 

Q6 (Attempt any one)

(a) Gaussian noise and averaging filter

Gaussian noise follows normal distribution and appears as grainy texture. Averaging filter reduces it by smoothing pixel values but may blur edges 

(b) Haar transform

Haar transform is a simple wavelet transform used for image compression.

Steps:                                                                     Compute averages and differences

Form transformation matrix                                  Apply recursively

It is fast and memory-efficient 

Q7 (Attempt any one)

(a) Edge linking using Hough Transform

Hough Transform detects geometric shapes like lines and circles by transforming image points into parameter space, enabling edge linking even in noisy images 

(b) Types of image degradations

Image degradation includes blur, noise, geometric distortion, motion blur, atmospheric turbulence, and sensor errors. These degrade image quality and require restoration techniques