25. Discuss frequency domain filtering and its advantages.

Frequency domain filtering involves modifying the frequency components of an image. Advantages

include better handling of periodic noise and more efficient computation for large filters.

26. Explain the process of image smoothing using Gaussian filters.

Gaussian filters smooth images by reducing noise and detail. They use a kernel based on the

Gaussian function to blur the image, preserving edges better than a mean filter.

27. Describe the different types of noise and noise reduction methods.

Types include Gaussian, Salt-and-Pepper, Poisson, and Speckle noise. Reduction methods include

median filtering, Gaussian filtering, and adaptive filtering.

28. Discuss spatial filtering and its types with examples.

Spatial filtering modifies pixel values based on neighbors. Types: linear (e.g., averaging), and

non-linear (e.g., median). Used for noise reduction and edge enhancement.

29. Explain the steps involved in image restoration.

Steps: model degradation, estimate degradation function, apply inverse filtering or Wiener filtering,

and post-processing.

30. Describe image segmentation based on region growing.

Region growing starts with seed points and adds neighboring pixels with similar properties. It stops

when no more similar neighbors are found.

31. Compare and contrast various edge detection techniques.

Techniques: Sobel (gradient-based), Prewitt (similar to Sobel), Canny (multi-stage, optimal),

Laplacian (second derivative). Canny offers the best trade-off.

Operations include erosion, dilation, opening, and closing. These shape-based operations are used

in binary image processing for structure analysis.

33. Explain JPEG compression algorithm.

JPEG uses DCT (Discrete Cosine Transform), quantization, and entropy coding. It is a lossy method

that reduces image file size significantly.

34. Describe how HSI color model works in image analysis.

HSI separates color into Hue (color type), Saturation (intensity of color), and Intensity (brightness). It

mimics human perception and aids color-based processing.

35. Explain connected component labeling in segmentation.

It identifies connected pixel regions with the same intensity. It assigns a unique label to each

connected region using 4 or 8 connectivity.

36. Explain the process of digitizing an image with sampling and quantization. Illustrate with

diagrams.

Sampling divides the image into pixels; quantization assigns each pixel a value. This creates a

digital version of the analog image. (Diagram not shown in PDF).

37. Describe the architecture of a typical image processing system with a neat block

diagram.

Components: image acquisition, preprocessing, segmentation, feature extraction, recognition, and

interpretation. (Diagram not shown in PDF).

38. Derive the 2D Fourier Transform and explain its application in frequency domain
processing.

F(u,v) = Sum_x Sum_y f(x,y) * exp(-j2*pi(ux/M + vy/N)). It transforms an image to the frequency

domain for filtering and analysis.

39. Compare various image enhancement techniques in both spatial and frequency domain.

Spatial: histogram equalization, contrast stretching. Frequency: high-pass and low-pass filtering.

Spatial methods are intuitive, frequency ones are more powerful.

40. Discuss different types of spatial filters and give their mathematical formulations.

Types: averaging filter, Laplacian filter, Sobel filter. Example: Averaging filter kernel is a matrix of

equal values summing to 1.

41. Explain the role of image restoration and describe the Wiener filter technique in detail.

Restoration aims to recover the original image. Wiener filter minimizes mean square error between

estimated and original image, using noise and signal estimates.

42. Describe in detail the various steps of image compression. Compare JPEG and MPEG

compression.

Steps: transformation, quantization, encoding. JPEG is for still images; MPEG compresses video

using inter-frame redundancy.

43. Illustrate segmentation techniques using thresholding, edge detection, and region-based

approaches.

Thresholding: separates objects based on intensity. Edge detection: finds boundaries.

Region-based: grows regions with similar properties.

44. Explain the morphological operations: erosion, dilation, opening, and closing with
diagrams.

Erosion shrinks objects, dilation expands them. Opening removes small objects, closing fills small

holes. (Diagrams not shown in PDF).

45. Describe in detail the principles of color image processing and various color models.

It processes images in color spaces like RGB, HSV, YCbCr. It enables tasks like segmentation and

recognition using color features.

46. Explain in detail the image representation and description for object recognition.

Representation: boundary, region-based. Description: shape, color, texture features. Used in pattern

recognition and classification.

47. Discuss edge linking and boundary detection techniques with suitable examples.

Edge linking connects edge points using methods like Hough Transform or thresholding. Boundary

detection traces object outlines.

48. Explain restoration techniques using inverse and Wiener filtering in the presence of

noise.

Inverse filtering reverses degradation. Wiener filtering adds noise modeling to produce more

accurate restoration.

49. Discuss the applications of digital image processing in medical imaging and remote

sensing.

Medical: MRI, CT enhancement, segmentation. Remote sensing: land use mapping, environmental

monitoring using satellite images.

50. Explain template matching technique for object detection and recognition.
Template matching compares image regions with a predefined template. It uses correlation to detect

matching areas.