QB301 (a) Develop the technique of median filter in image restoration and discuss it in

detail.
A median filter is a nonlinear digital filtering technique used to reduce noise in an image or a
signal. It is particularly effective at preserving edges and details while smoothing out random
variations or outliers. The median filter works by replacing each pixel value in an image with the
median value of neighboring pixels within a specified window or kernel.
1. Median Filter as a Smoothing Filter:
- The median filter is categorized as a type of smoothing filter.
- Smoothing filters are employed to reduce noise in images or signals while preserving
essential details and edges.
2. Absence of Mask:
• Unlike some other filters that use masks or kernels, the median filter doesn't employ a
mask. The median filter doesn't use a mask or kernel in the traditional sense because it
operates differently from linear filters that rely on a weighted sum of pixel values within
a defined neighborhood.
• The median filter's strength lies in its non-linear nature, robustness to outliers, and ability
to preserve edges. By not using a mask, it can efficiently address noise-related issues in
images while maintaining simplicity in its operation.

3. Sorting and Choosing Median:

- The process involves selecting a 3x3 sub-image from the original image.
- For a given window of pixel values, the first step is to sort them in ascending order.
- The pixel values within this sub-image are arranged in ascending order.
- The first four values are then ignored, and the remaining values are used to calculate the
median.
• 3 5 7 2 10 20 30 9 4 – sub image
• 2,3,4,5,7,9,10,20,30
4. Non-Linear Spatial Filtering:
- The median filter is classified as a non-linear spatial filtering technique.
- Non-linear filters, as opposed to linear ones, do not rely on a linear combination of pixel
values.
Types of Median Filtering:
1. Median Filtering Smoothing:
- The primary purpose of median filtering is smoothing.
- By replacing each pixel value with the median of neighboring pixels, the filter effectively
reduces noise while preserving edges.

2. Max Filter:
- Rule (R) for Max Filter: (R = Max)
- The Max filter is utilized to highlight the brightest points in an image.
- It replaces each pixel with the maximum value within the specified window.
Median Filter Maximum:
It replaces each pixel value with the maximum value within the specified window. This filter is
useful in scenarios where you want to highlight bright regions or edges in an image. By replacing
each pixel with the maximum value in the neighborhood
3. Min Filter:
- Rule (R) for Min Filter: (R = Min)
- The Min filter is employed to emphasize the darkest or smallest points in an image.
- It replaces each pixel with the minimum value within the specified window.
Median Filter Minimum:
It replaces each pixel value with the minimum value within the specified window. This is often
employed to enhance darker details or edges.
QB302 (a) Interpret the Wiener Filtering and its significance in detail.
Based on these conditions, the minimum of the error function in the frequency domain by
the expression.
• This result is known as the wiener filter after N.Wiener,who proposed the concepts in the
year shown. The filter which consists of the term inside the brackets also is commonly
referred to as the minimum mean square error filter or the least square error filter.
• We include references at the end of sources containing detailed derivations of the wiener
filter. The restored image in the spatial domain is given by the inverse Fourier transform
of the frequency domain estimate F(u,v).
• If the noise is zero,then the noise power spectrum vanishes and the wiener filter reduces
to the inverse filter.
• However, the power spectrum of the undegraded image seldom is known. Where k is a
specified constant.
• Example illustrates the power spectrum of wiener filtering over direct inverse filtering.
The value of K was chosen interactively to yield the best visual results.
• It illustrates the full inverse filtered result similarly is the radially limited inverse filter.
• These images are duplicated here for convenience in making comparisons.
• As expected, the inverse filter produced an unusable image. The noise in the filter.
• The wiener filter result is by no means perfect, but it does give us a hint as to image
content.
• The noise is still quite visible, but the text can be seen through a “curtain” of noise.
QB303 (a) Summarize the different morphological technique in image processing.
Morphological operations in image processing involve the manipulation of image structures
using shapes.

1. Morphology: It refers to the study of shapes and structures. In image processing,

morphological operations deal with modifying or analyzing the geometric structure of an image.
2. Dilation: Dilation is a morphological operation that expands or thickens regions in an image.
It involves placing a structuring element (a predefined shape like a square or circle) at each pixel
and assigning the maximum pixel value within that neighborhood to the center pixel. This
process enhances or enlarges bright regions in the image.
3. Erosion: Erosion is the counterpart to dilation. It shrinks or thins regions in an image. Similar
to dilation, a structuring element is used, but this time the minimum pixel value within the
neighborhood is assigned to the center pixel. Erosion is useful for removing small details or
noise from an image.

These morphological techniques are fundamental in image processing for tasks such as noise
reduction, feature extraction, and shape analysis.
QB304 (b) Demonstrate the process of Thresholding and Linking in detail.
QB305 (b) Illustrate the concept as well pros and cons of region-growing techniques in
detail.
Region growing:
Region growing is a simple region-based image segmentation method. It is also classified as a
pixel-based image segmentation method since it involves the selection of initial seed points. This
approach to segmentation examines neighboring pixels of initial ―seed points‖ and determines
whether the pixel neighbors should be added to the region. The process is iterated on, in the same
manner as general data clustering algorithms.
Region-based segmentation:
The main goal of segmentation is to partition an image into regions. Some segmentation methods
such as "Thresholding" achieve this goal by looking for the boundaries between regions based on
discontinuities in gray levels or color properties. Region-based segmentation is a technique for
determining the region directly. The basic formulation for Region-Based Segmentation is:

Advantages：
1. Region growing methods can correctly separate the regions that have the same properties we
define.
2. Region growing methods can provide the original images which have clear edges the good
segmentation results.
3. The concept is simple. We only need a small numbers of seed point to represent the property
we want, then grow the region.
4. We can determine the seed points and the criteria we want to make.
5. We can choose the multiple criteria at the same time.
6. It performs well with respect to noise.
Disadvantages：
1. The computation is consuming, no matter the time or power.
2. Noise or variation of intensity may result in holes or oversegmentation.
3. This method may not distinguish the shading of the real images
QC301 (b) Develop the Boundary segmentation techniques based on region growing,
splitting and merging in detail.
Region-based segmentation:
The main goal of segmentation is to partition an image into regions. Some segmentation methods
such as "Thresholding" achieve this goal by looking for the boundaries between regions based on
discontinuities in gray levels or color properties. Region-based segmentation is a technique for
determining the region directly. The basic formulation for Region-Based Segmentation is:

Region growing:
Region growing is a simple region-based image segmentation method. It is also classified as a
pixel-based image segmentation method since it involves the selection of initial seed points. This
approach to segmentation examines neighboring pixels of initial ―seed points‖ and determines
whether the pixel neighbors should be added to the region. The process is iterated on, in the same
manner as general data clustering algorithms.
Region Splitting and Merging:
1. Region Growing from Seed Points:
- The procedure involves growing regions from a set of seed points.
- An alternative approach initially subdivides an image into arbitrary, disjointed regions.
- Regions are then merged and/or split to satisfy certain conditions.
2. Merging Adjacent Regions:
- Regions Rj and Rk are merged if the condition P(Rj ∪ Rk) = TRUE is satisfied.
 - The process continues until no further merging or splitting is possible.
3. Texture Quantification:
- Mean and standard deviation of pixels in a region are used to quantify the texture of the
region.
Role of Thresholding in Image Formation:
1. Reflectance and Illumination Model:
- The image f(x, y) is modeled as the product of a reflectance component r(x, y) and an
illumination component i(x, y).
- The model is represented as: ƒ(x, y) = i(x, y) r(x, y).
2. Histogram Bimodality:
- The reflectance function's histogram is bimodal.
- Placing a single global threshold T in the histogram valley can easily partition it.
3. Logarithmic Transformation:
- Taking the natural logarithm of the image equation yields a sum: z(x, y) = ln ƒ(x, y) = ln i(x,
y) + ln r(x, y).
4. Histogram Convolution:
- If i(x, y) and r(x, y) are independent, the histogram of z(x, y) is given by the convolution of
the histograms of i(x, y) and r(x, y).
- Nonuniformity in illumination can cause distortion in the histogram of r(x, y).
5. Compensating for Nonuniformity:
- When access to the illumination source is available, the illumination pattern is projected onto
a constant, white reflective surface, yielding g(x, y) = k i(x, y).
- Normalizing the image ƒ(x, y) by dividing it by g(x, y) results in h(x, y) = ƒ(x, y) / g(x, y) =
r(x, y) / k.
6. Segmentation with Thresholding:
- If r(x, y) can be segmented using a single threshold T, then h(x, y) can be segmented using a
single threshold of value T/k.
 QC302 (a) Demonstrate the concept of removing the unwanted spurious points in Edge of
an image.
Edge thinning – 4 marks
Advantage – 4 marks
Algorithm – 6 marks
Second order approach – 1 marks

Edge thinning is a technique used to remove the unwanted spurious points on the edge of an
image. This technique is employed after the image has been filtered for noise (using median,
Gaussian filter etc.), the edge operator has been applied (like the ones described above) to detect
the edges and after the edges have been smoothed using an appropriate threshold value. This
removes all the unwanted points and if applied carefully, results in one pixel thick edge elements.
Advantages:
1) Sharp and thin edges lead to greater efficiency in object recognition.
2) If you are using Hough transforms to detect lines and ellipses then thinning could give much
better results.
3) If the edge happens to be boundary of a region, then thinning could easily give the image
parameters like perimeter without much algebra.

Algorithm:
There are many popular algorithms used to do this, one such is described below:
1) Choose a type of connectivity, like 8, 6 or 4.
2) 8 connectivity is preferred, where all the immediate pixels surrounding a particular pixel are
considered.
3) Remove points from North, south, east and west.
4) Do this in multiple passes, i.e. after the north pass, use the same semi processed image in the
other passes and so on.
5) Remove a point if: The point has no neighbors in the North (if you are in the north pass, and
respective directions for other passes.) The point is not the end of a line. The point is isolated.
Removing the points will not cause to disconnect its neighbors in any way.
6) Else keep the point. The number of passes across direction should be chosen according to the
level of accuracy desired.

Second-order approaches to edge detection

Some edge-detection operators are instead based upon second-order derivatives of the intensity.
This essentially captures the rate of change in the intensity gradient. Thus, in the ideal continuous
case, detection of zero-crossings in the second derivative captures local maxima in the gradient.

The early Marr-Hildreth operator is based on the detection of zero-crossings of the Laplacian
operator applied to a Gaussian-smoothed image. It can be shown, however, that this operator will
also return false edges corresponding to local minima of the gradient magnitude. Moreover, this
operator will give poor localization at curved edges. Hence, this operator is today mainly of
historical interest.