See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/299357099

Fundamentals of Digital Image Processing and Basic Concept of

Classification

Article · June 2014

DOI: 10.18488/journal.65/2014.1.6/65.6.98.108

CITATIONS READS
2 7,654

3 authors, including:

Muhammad Shoyab
Japan East West Medical College Hospital
9 PUBLICATIONS   5 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Publications View project

Digital Image Processing View project

All content following this page was uploaded by Shammi Shawal on 10 March 2019.

The user has requested enhancement of the downloaded file.

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

International Journal of Chemical and Process

Engineering Research
ISSN(e): 2313-0776/ISSN(p): 2313-2558

URL: www.pakinsight.com

FUNDAMENTALS OF DIGITAL IMAGE PROCESSING AND BASIC CONCEPT

OF CLASSIFICATION

Shammi Shawal1† --- Muhammad Shoyab2 --- Suraiya Begum3

1
Department of Urban and Regional Planning, Bangladesh University of Engineering and Technology, Bangladesh
2
Medical Officer, Popular Hospital, Dhaka, Bangladesh
3
Principal Scientific Officer, Bangladesh Space Research and Remote Sensing Organization, Bangladesh

ABSTRACT
We have to classify and analyze digital images for different study and purposes. Digital images are
obtained from sources like camera, satellites, aircraft etc. Data obtained from satellites or aircraft
i.e, the space based and remote sensing data needs to be corrected as they are usually
geometrically distorted due to their acquisition system and the movements of the platform of
aircraft. Processing and pre-processing are necessary for such correction prior to image
classification. Image Processing is a technique which is used to enhance raw images received from
cameras, satellites, space probes, aircrafts etc. and Digital image processing is the technique of
processing images in the form of discrete digital brightness using digital circuits or digital
computers. Image analysis and classification starts when processing and pre-processing ends.
© 2014 Pak Publishing Group. All Rights Reserved.

Keywords: Air craft, Brightness, Distorted, Image, Processing, Space probe.

Contribution/ Originality
This paper contributes towards the concepts of digital image, digitization, image processing
and classification. It complies originally about the stages of digital image processing and its
application as well as its necessity.

1. INTRODUCTION
Digital image processing (DIP) can be defined as a transformation of an distorted image into a
modified one which helps people to detect silent features without difficulty for interpretation

† Corresponding author
DOI: 10.18488/journal.65/2014.1.6/65.6.98.108
ISSN(e): 2313-0776/ISSN(p): 2313-2558
© 2014 Pak Publishing Group. All Rights Reserved. 98
 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

necessary for image analysis in different study. It is an electronic data processing on a 2-D array of
numbers known as pixel which is the numeric representation of any image [4]. The output of image
processing can be an image or a set of characteristics related to the image. Digital Image Processing
techniques are used to manipulate the digital images using computers. Image processing system
includes treating images as 2d signal [5].Image processing system consists of a source of image
data, a processing element and a destination for the processed results. The source of image data
may be a camera, a sensor, satellite, scanner, a mathematical equation, statistical data, the Web, a
SONAR system, etc. The processing element is a computer, destination for the processed result,
and the output of the processing may be a display monitor.

1.1. Field of DIP Application

Digital image processing techniques are used

- Geographic Information Systems (GIS)
- extensively to manipulate satellite imagery
- Meteorology
- Terrain classification
- Medical science etc.

1.2. Major Tasks

2 Major Tasks in Digital Image Processing
– pictorial information improvement for human interpretation
– data processing for storage, transmission and representation

2. OBJECTIVE
The objective can be summarized as follows:
 Image Correction to remove distortion, errors, and noise during data acquisition
 Image Enhancement to modify images or increases visual appearance and interpretability
of imagery

© 2014 Pak Publishing Group. All Rights Reserved. 99

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

 Information Extraction to classify pixels or neighborhoods of pixels on the basis of their

spectral-radiometric temporal responses using computer

3. APPLIED IMAGE PROCESSING

Hardware and Software for Image Processing

Hardware Software
- Computer monitor - Photoshop
- CD-ROM, ink-jet printer - ERDAS Imagine
- Disk Drive - IDRISI, Arc View
- Internet connection - MATLAB, Visual C++
- Scanner, digital camera - ENVI, ER Mapper etc.

3.1. Image Concept

Image is a kind of language which expresses visual information to human being. Two types of
images are- Analog & Digital .An analog image is continuous and indiscrete, the natural picture
and a digital image is two-dimensional image having finite set of digital values, called picture
elements or pixels.Digital image is composed of discrete pixels, or set of pixels each having
associated integer brightness value which is called gray-level [1].

3.2. Pixel and Gray Level Concept

Image Formation
In other words, digital image is a data matrix
or array of elements which are known as image
pixels. Each pixel has an integer location and
the integer value is proportional to the
brightness at the spatial point of the pixel. The
accuracy of pixel brightness is quantized in
number of bits or gray-levels [1].

A real image is formed on a sensor when an emitted energy strikes it with sufficient intensity
to create a sensor output [1].A digital image is formed as a 2-D function, f(x, y), where x and y can
be defined as spatial coordinates, & the amplitude of f at any pair of coordinates (x, y)which is
called the intensity or grey level of the image at that point. The coordinates x, y, and the amplitude

© 2014 Pak Publishing Group. All Rights Reserved. 100

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

values of f are all finite, discrete quantities, we call the image a digital image which is obtained by
digitization of analog image [1].

Digitization
Digitization implies that a digital image is an approximation of a
real scene Pixel brightness value which typically represents gray
levels, colors, heights, opacities etc. [1]
Digitization accuracy
Resolutions express digitization accuracy.
They are of 2 types
– Pixel resolution
– Brightness resolution (color resolution)

Pixel resolution expresses pixel sampling accuracy. They are of 2

types. Pixel resolution can be described by actual size of each pixel
of a digital image, called Absolute resolution. Pixel resolution can
also be described by the size of a digital image or the amount of
pixels of the image, called Relative resolution [2].
Brightness Resolutions expresses brightness quantization accuracy.
Brightness resolution is also a kind of relative resolution because
pixel brightness is quantized into finite number of gray-levels. It
denotes total no. and data bits for gray –levels [2].

4. KEY STAGES OF DIGITAL IMAGE PROCESSING

1) Image acquisition 2) Image enhancement 3) Image restoration
4) Image pre-processing 5) Segmentation 6) Object recognition 7) Classification

4.1. Image Acquisition

An analog picture cannot be conveniently interpreted by a
computer and thus an alternative representation; the digital
image must be used. The objective of image acquisition is to
covert an analog optical image into a digital image in a faithful
manner as possible. Digital image acquisition is concerned with
generation of 2D array of integer values representing the
reflectance of actual scene at discrete special intervals [1]. This
is accomplished by
- Sampling (Digitizing the coordinate values)
© 2014 Pak Publishing Group. All Rights Reserved. - Quantization (Digitizing the amplitude values) 101
 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

Sampling means measuring the value of an image at a finite number of points normally
corresponds to the extent of the no of pixels in both vertical and horizontal directions [1].
Quantization is the representation of the measured value at the sampled point by an integer.
The number of gray levels in the equally spaced gray scale is called the quantization or gray scale
resolution of the system [1].

4.2. Digital Image Enhancement

Image enhancement makes the interpretation of complex data easier for the operator.
Fewer errors are made, more subtle features can be detected, and quantitative measurements
are facilitated. The goal of digital image enhancement is to produce approached image that is
suitable for a given application [1].

Image enhancement tasks

Image reduction, image magnification, transect extraction, contrast adjustments (linear and
non-linear), band rationing, spatial filtering, Fourier transformations, principal components
analysis, texture transformations, and image sharpening etc [2].

Image enhancement methods

Image enhancement is the modification of an image to alter its impact on the viewer.
Image enhancement method includes:
1) Contrast enhancement 2) Density slicing 3) Frequency filtering 4) Band rationing

4.3. Image Restoration

The objective of Image Restoration is to highlight fine details in the image which were
suppressed by the blur. The goal of image restoration is to improve a degrade image in some
predefined sense. Schematically this process can be visualized as

where f is the original image, g is a degraded/noisy version of the original image and f is a
restored version. Image restoration removes a known degradation [3].
The distortion may be specified by locating control points and identifying their corresponding
control points in an ideal. The distortion model then made transformation between these control
points to generate a special warping function which allows building output image pixel by pixel
(warped) [4].

© 2014 Pak Publishing Group. All Rights Reserved. 102

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

4.4. Image Pre-Processing

4.4.1. Radiometric Correction
Radiometric correction improves the fidelity of the DN’s that constitute an image.
Radiometric Corrections used for
- correcting the data for Sensor Irregularities
- remove Unwanted Sensor or Atmospheric Noise,
- converting the data so they accurately represent the
reflected or emitted radiation measured by the
sensor.

Types of radiometric correction

- Atmospheric error correction (external error)

- Topographic error correction (external error)
- Detector error or sensor error (internal error)
Atmospheric correction attempts to quantify (i.e.,
Image with nose Radiometric Corrected remove) the effect of the atmosphere at
the time an image was acquired [3].

Absolute Radiometric (Atmospheric) Correction

The general goal of absolute radiometric (atmospheric) correction is to turn the digital
brightness values (or DN) recorded by a remote sensing system into scaled surface reflectance
values [1].
Relative Radiometric (Atmospheric) Correction:
When required data is not available for absolute radiometric (atmospheric) correction, we can
do relative radiometric (atmospheric) correction. Relative radiometric correction may be used to
– Single-image normalization using histogram adjustment
– Multiple-data image normalization using regression

Single-Image Normalization Using Histogram Adjustment

Dark object Subtract

method can be applied
for atmospheric scattering
corrections to the image
data [1].

© 2014 Pak Publishing Group. All Rights Reserved. 103

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

4.4.2. Topographic Correction

Topographic slope and aspect also introduce radiometric distortion (for example, areas in
shadow)The goal of a slope-aspect correction is to remove topographically induced illumination
variation so that two objects having the same reflectance properties show the same brightness value
(or DN) in the image despite their different orientation to the Sun’s position [2].

Scene Illumination
In satellite remote sensing,
imagery acquired at
different times of the year
may be required (e.g., to
study phonological
cycle).These may require
sun elevation correction and
an earth-sun distance

Solar Elevation Angle: the angular elevation of the sun above the horizon Solar Zenith Angle:
The angular deviation from directly overhead (or complement of elevation). Corrections: DN/Sin
Elevation Angle or DN/Cosine Zenith Angle [2].

4.4.3. Geometric Correction

Both aerial photos and satellite imagery have geometrical errors. Satellite imageries usually
have many more errors than aerial photos. Geometric correction is necessary to pre process Remote
Sensing data so that individual picture elements (pixels) are in their proper planimetric (x, y) map
locations. Geometric corrections include correcting for geometric distortions due to sensor-Earth
geometry variations, and conversion of the data to real world coordinates (e.g. latitude and
longitude).Geometrically corrected imagery can be used to extract accurate distance, polygon area,
and direction (bearing) information [2].

Sources of geometric distortion:

Distortions appear as:
• Sensor characteristics
-optical distortion
- changes of scale over the image
- aspect ratio
- irregularities in the angular
- non-linear mirror velocity
relationships among the image
- detector geometry & scanning sequence
elements
• Viewing geometry
- displacement of objects in an
- panoramic effect
image
- earth curvature
- occlusion of one image element
• Motions of the aircraft/satellite or target
by another
© 2014 Pak Publishing Group. All Rights Reserved. 104
 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

- attitude changes (pitch, roll, etc.)

- position variations (altitude etc.)
- earth rotation

Levels of Geometric Corrections

1) Registration 2) Rectification (geo-referencing) 3) Geocoding 4)Ortho- rectification
Geometric Correction is done for Conversion of the data to real world coordinates. This is
carried by analyzing well distributed Ground Control Points (GCPs).This is done in two steps as
follows:
Geo-referencing (rectification) which involves the calculation of the appropriate
transformation from image to terrain coordinates and Geocoding which involves resampling the
image to obtain a new image in which all pixels are correctly positioned within the terrain
coordinate system [2].

Rectification Method
This method includes Selection of Ground Control Points (GCP) that can be located on both
the map and the image. In these equations (a and b) the unknowns are solved by determining the
coordinates for a set of known locations known as ground control points (GCP’s). They should be
well defined , spatially small and well distributed over the entire image [2]

Resampling
This process calculates the new pixel values from the original digital pixel values in the raw
and uncorrected image. There are three common methods for resampling of images which are as
follows:
- nearest neighbor : this process assigns each
corrected output pixel the value of the nearest
input pixel
- bilinear interpolation :this process calculates the
new output pixel value using interpolations from
the four closest input pixels
- cubic convolution :this process interpolate a new
pixel value from a larger neighborhood of 9, 16,

© 2014 Pak Publishing Group. All Rights Reserved.

25 or 36 surrounding input pixels [4]
105
 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

Image Registration
• This process applies the similar techniques as
rectification for image to image and image to map
overlays. It includes edge detection which is used
for creating image outlines, giving areas with
strong intensity contrasts. Edge detected image
preserves the useful important structural properties
filtering out un wanted things [4].

4.5. Segmentation .
This process is used to subdivide an image
. into its component regions or objects. Algorithms
of this process are based generally on one of the following 2 basic properties of intensity values
- discontinuity: to make partition of an image based on sharp changes in intensity
(such as edges)
- similarity: to make partition of an image into regions that are similar according to a
set of predefined criteria [4].

It should stop when the objects of interest in an application have been isolated.

Image mosaic

4.6. Image Classification

Image classification can be defined as the process of reducing an image to information classes.
It is commonly used in photo interpretation, quantitative analysis. One of the main purposes of
satellite remote sensing and digital image processing is to interpret the observed data and classify
features for analysis. This process is usualy applied to label each pixel to particular spectral
classes and classes of interest - Information classes and Spectral classes [3].There 2 types of
classification procedure - Supervised and Unsupervised [3].

4.6.1. Supervised Classification

This type of classification is an essential tool which is used to extract quantitative information
from remote sensing image. This process includes training. The analyst has to select groups of

© 2014 Pak Publishing Group. All Rights Reserved. 106

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

training sets (pixels) that are representatives for all classes of interest. This training dataset forms
the basis for classification of the total image data. The algorithm classifies pixels with unknown
identities using samples with known values. The user should procedure start the process by
selecting and naming areas on the image corresponding to the classes of interest i.e, the information
classes. The classification algorithm then will find out all other areas of similar class [3].

Procedure
Display the image in a single band or three-band combination.
– Acquire training sets.
– Choose the classifier type.
– Perform classification.
– Refine training sets.
– Derive the accuracy assessment measures [3].

• Supervised classification usually follows

an iterative process too.

Training set is selected as the subset of the image,

which contains a spectral class associated with a
certain information class whereas the training data
contains information from all spectral bands within
the area indicated by the user or analyst. Most of the
times, there are more than one spectral class for a
given information class [3].
4.6.2. Unsupervised Classification
This type of classification approaches to use iso-data clustering .In this process, clusters of
pixels based on their similarities in spectral information are automatically classified into classes of
interest. The image is segmented into spectral classes automatically based on natural groupings of
the data.
The classification procedure is as follows:
- User has to inputs some classification parameters.
- The algorithm proceeds by finding pixels having
similar spectral properties.
- The user names each class after the classification

Spectral values of a given cover type are close together, whereas data of different classes are
well separated. After the classification, the analyst labels all identified spectral clusters. Object
Recognition & Representation is the final stage which gives the result of Digital image processing
.[3].

© 2014 Pak Publishing Group. All Rights Reserved. 107

 International Journal of Chemical and Process Engineering Research, 2014, 1(6): 98-108

5. CONCLUSION
Image Processing is necessary for different types of works like Land use study, numerical
weather prediction, mapping etc. Digital Image Processing needs for digital images like space
based and remote sensing data. It includes data operation which normally precedes for
interpretation and further manipulation for analysis of the image data to extract specific information
of interest. These operations aim to correct distorted or degraded image data to create a more
faithful representation of the original scene which helps the user or analyst to detect information
easily in a faithful manner. Preprocessing commonly comprises a series of sequential operations,
including -atmospheric correction , normalization, radiometric & geometric correction, image
registration representation etc.

6. ACKNOWLEDGEMENT
The authors would like to express their gratitude to the officials of Bangladesh Space Research
and Remote Sensing Organization (SPARRSO), especially to Md. Mozammel Haque Sarker. The
library of SPARRSO has also been very helpful in this research.

REFERENCES
[1] B. M. Namee, "Digital image processing: Digital image fundamentals." Available:
http://www.comp.dit.ie/bmacnamee/materials/dip/lectures/ImageProcessing2-
ImageProcessingFundamentals.ppt [Accessed 13 August, 2014], n.d.
[2] D. DiBiase, The nature of geographic information. John A. Dutton E-Education Institute. Available:
https://www.e-education.psu.edu/natureofgeoinfo/c8_p18.html [Accessed 13 August, 2014], n.d.
[3] J. Campbell, "Digital image classification: 4354-Remote sensing, Ini Chitoz, Erdas." Available:
http://www.scribd.com/doc/130172996/ERDAS-Digital-Image-Classification-Geography-4354-
Remote-Sensing [Accessed 13 August, 2014], 2001.
[4] C. Solomon and T. Breckon, Fundamentals of digital image processing: A practical approach with
examples in Matlab. Wiley-Blackwell. Available: www.fundipbook.com [Accessed 13 August,
2014], n.d.
[5] G. Engineers, "Introduction to image processing." Available:
http://www.engineersgarage.com/articles/image-processing-tutorial-applications [Accessed 12
September, 2014], n.d.

Views and opinions expressed in this article are the views and opinions of the authors, International Journal of
Chemical and Process Engineering Research shall not be responsible or answerable for any loss, damage or liability etc.
caused in relation to/arising out of the use of the content.

© 2014 Pak Publishing Group. All Rights Reserved. 108

View publication stats