Digital Image Processing: Lab 4: Image Enhancement in The Spatial Domain

Digital Image Processing
Lab 4: Image Enhancement in the Spatial Domain
Waqar Tanveer
Department of Computer Engineering
Bahria University, Islamabad
October 12, 2020
Digital Image Processing 1 / 24

Objective
For image enhancement, we will cover the following topics:
Mathematical and Logical Operations on Images
Masking Operations
Spatial Filtering

Mathematical Operations: Addition
Addition of images is similar to addition of matrices.
The images should be of identical size so that addition can be

done.
The size of the output image would be same as that of input

images. Only the pixel values of output image are the sum of
pixels values of input images.
Mathematical form of image addition is given as:

f (x, y ) = g (x, y ) + h(x, y )
where,
f (x, y ) : Output Image
g (x, y ) : Input Image 1
h(x, y ) : Input Image 2
Mathematical Operations: Addition
Figure: Addition of Images

Mathematical Operation: Subtraction
Subtraction of images is similar to subtraction of matrices.
The images should be of identical size so that subtraction can

be done.
The size of the output image would be same as that of input

images. Only the pixel values of output image are the
difference of pixels values of input images.
Mathematical form of image subtraction is given as:
f (x, y ) = g (x, y ) − h(x, y )
where,
Mathematical Operations: Subtraction
Mostly image subtraction is used for removal of background
from images, image matching, missing object tracing etc.
Figure: Subtraction of Images

Mathematical Operations: Multiplication
Superimpose texture on images.
Convolution and correlation of images.
The mathematical form of image multiplication is given as:

where, f (x, y ) = g (x, y ). ∗ h(x, y )
Figure: Multiplication of Images

Logical Operations
Figure: Logical operations on image

Average of Images
A noisy image can be represented by:
f (x, y ) = g (x, y ) + η(x, y )

where,
f (x, y ) : Averaged Image
g (x, y ) : Input Image
η(x, y ) : Noise
The noise is random and the content is fixed
The noise can be removed by taking more noisy images of the

same object and averaging them out

Average of Images
Figure: Averaging the Image

Spatial Filtering
The process of filtering is also known as convolving a mask

with an image.
A mask is a filter. Concept of masking is also known as

spatial filtering.
The general process of filtering and applying masks is consists

of moving the filter mask from point to point in an image.
At each point (x,y) of the original image, the response of a

filter is calculated by a pre defined relationship.

Spatial Filtering
Figure: Masking
Spatial Filtering
Figure: Masking over image

Problems during masking of image
Problem arises when part of the mask is located outside the

image plane, we can use following techniques to overcome this
problem:
Discard the problem pixels (e.g. 512x512 input 510x510

output if mask size is 3x3).
Zero padding: expand the input image by padding zeros

(512x512 input 514x514 output) Zero padding is not good, it
creates artificial lines or edges on the border.
We normally use the gray levels of border pixels to fill up the

expanded region (for 3x3 mask). For larger masks a border
region equal to half of the mask size is mirrored on the
expanded region.

MATLAB Code for Zero Padding
Figure: Zero Padding in MATLAB

Effect of Zero Padding
Figure: Effect of Zero Padding

Mask operation near borders
Figure: Masking around borders

MATLAB code for weighted mask operation
Figure: MATLAB code for mask operation

Types of Spatial Filters
Generally we have two types of spatial filters:

1. Linear filters or smoothing filters.
2. Frequency Domain Filters.
The linear filters are further divided into two categories:

1. Box Filter.
2. Weighted average filter.
Smoothing filters are used for blurring/noise reduction.

Blurring is usually used in pre-processing steps e.g., to remove
small details from an image prior to object extraction, or to
bridge small gaps in lines or curves.

Smoothing Filters
Figure: Smoothing Filters

Tasks
Task 1
Take any two images of your choice and perform image
addition, image subtraction and image multiplication. Do the
analysis on the results.
Task 2
Take any image of your choice, corrupt it by adding Gaussian
Noise and Salt and Pepper Noise. Compare and analyze
the result.
Hint: Use ”imnoise” MATLAB command.
Task 3
Perform noise removal by averaging multiple samples of an
image corrupted by gaussian noise. Show the effect of
averaging 5, 25 and 50 samples of noisy image. And comment
on the results.

Tasks
Task 4
Take any image, corrupt it by adding gaussian noise and the
filter the image using 3x3 weighted smoothing mask. Do
analysis on result.
Task 5
Apply box filter mask of size 3x3,5x5,9x9,15x15 and 35x35 on
any gray scale image of your choice , show and comment on
the results.
Hint: Use ”fspecial” and ”imfilter” MATLAB commands.

Digital Image Processing: Lab 4: Image Enhancement in The Spatial Domain

Uploaded by

Copyright:

Available Formats

Digital Image Processing: Lab 4: Image Enhancement in The Spatial Domain

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Digital Image Processing: Lab 4: Image Enhancement in The Spatial Domain

Uploaded by

Copyright:

Available Formats

Digital Image Processing

Lab 4: Image Enhancement in the Spatial Domain

October 12, 2020

Digital Image Processing 1 / 24

For image enhancement, we will cover the following topics:

Mathematical and Logical Operations on Images

Digital Image Processing 2 / 24

Addition of images is similar to addition of matrices.

The images should be of identical size so that addition can be

The size of the output image would be same as that of input

Mathematical form of image addition is given as:

Figure: Addition of Images

Digital Image Processing 4 / 24

The images should be of identical size so that subtraction can

The size of the output image would be same as that of input

Mathematical form of image subtraction is given as:

f (x, y ) = g (x, y ) − h(x, y )

Figure: Subtraction of Images

Digital Image Processing 6 / 24

Convolution and correlation of images.

The mathematical form of image multiplication is given as:

Figure: Multiplication of Images

Figure: Logical operations on image

Digital Image Processing 8 / 24

A noisy image can be represented by:

f (x, y ) = g (x, y ) + η(x, y )

The noise is random and the content is fixed

The noise can be removed by taking more noisy images of the

Digital Image Processing 9 / 24

Figure: Averaging the Image

The process of filtering is also known as convolving a mask

A mask is a filter. Concept of masking is also known as

The general process of filtering and applying masks is consists

At each point (x,y) of the original image, the response of a

Digital Image Processing 11 / 24

Figure: Masking over image

Digital Image Processing 13 / 24

Problem arises when part of the mask is located outside the

Discard the problem pixels (e.g. 512x512 input 510x510

Zero padding: expand the input image by padding zeros

We normally use the gray levels of border pixels to fill up the

Digital Image Processing 14 / 24

Figure: Zero Padding in MATLAB

Digital Image Processing 15 / 24

Figure: Effect of Zero Padding

Digital Image Processing 16 / 24

Figure: Masking around borders

Digital Image Processing 17 / 24

Figure: MATLAB code for mask operation

Digital Image Processing 18 / 24

Generally we have two types of spatial filters:

The linear filters are further divided into two categories:

Smoothing filters are used for blurring/noise reduction.

Digital Image Processing 19 / 24

Figure: Smoothing Filters

Digital Image Processing 20 / 24

Digital Image Processing 21 / 24

Digital Image Processing 22 / 24