UNIVERSITY OF ZIMBABWE

FACULTY OF ENGINEERING
DEPARTMENT OF ELECTRICAL ENGINEERING

DEVELOPMENT OF A GEODATA
PROCESSING SOFTWARE
PART IV
Project proposal

PREPARED BY

KONGA TAPIWA LAWRENCE (R129951D)

[email protected]

SUPERVISED BY

Dr. P. MANYERE
PROJECT COORDINATOR

Eng. E. RASHAYI
November 2015
Project submitted in partial fulfilment for the requirements of the Bachelor of Science Honors
degree in electrical engineering

INTRODUCTION
It is said that, picture is worth a thousand words. Humans rely heavily on sense of sight to
monitor the places around them and pictures are the most widely used and convenient way of
conveying information. Pictures briefly provide information about sizes, location and inter
relationships between objects while we are capable of deriving information from images.
Naturally we have the ability to see objects close to us whilst we would also like to monitor
far places and with the help of remote sensing which is defined as the science and technology
by which characteristics of objects of interest can be identified without direct contact we are
now able to do that. The vision of human beings is limited to the visible spectrum of the
electromagnetic spectrum whilst the imaging machines almost cover the whole
electromagnetic spectrum such as radiation and microwave which are not visible to the
human eye.
The captured images by the remote sensors come in digital format which means they are
represented as numbers which can be easily manipulated to obtain the required information
and the process is called Digital Image Processing. To cover a lot of ground the remote image
sensors are placed on platforms which are airborne or satellite. Recently, Unmanned Aerial
Vehicles (UAV) have demonstrated great potential in remote sensing as they are a low cost
alternative to the classical satellite and manned aerial photography. The reduced payload and
small size of UAVs limit their ability to carry heavy high quality imaging devices like those
mounted on satellites. Nowadays, with the enhanced processor performance UAVs can be
fully autonomous.
Digital Image Processing makes the image good for human interpretation or being compatible
for autonomous machine perception.

PROBLEM STATEMENT:
The traditional image acquisition sources which are the manned aircraft photography and
satellite imagery are both expensive and time consuming since it takes several hours to weeks
from the process of image acquisition to the end user. Both the manned aircraft photography
and satellite imagery offer a low resolution for free online which is of 10cm and 50 cm
respectively and the finer resolution is for sale and expensive.
In addition, images acquired by UAVs come in raw format which is not compatible with third
party software. Currently, satellite imagery is good for monitoring the earth but the repeat
cycles are infrequent and long. For example, a repeat cycle of 16 days is obtained from
Landsat. When using UAV platform for remote sensing camera movement causes blur which
in turn results in unknown image degradation (Sieberth et al. 2015).

JUSTIFICATION:
This project is mainly motivated by the need to develop a less complex, efficient and cheap
software to process raw SAR data to produce a resolution of 1.5cm/pixel.
In order to acquire a high resolution image one has to either use an impractically long antenna
or to employ short wavelengths which are easily attenuated in the atmosphere. In airborne
application the weight and size of the antenna are restricted hence, the above method is not
feasible. Another method of obtaining an image with better resolution from a radar is through
digital signal processing which is the main idea of this project.
The benefits obtained when using UAVs as a platform to obtain SAR data are given below;
You can acquire data at any time you want.
UAVs are capable to operate just below the clouds which makes them less dependent
on cloudy weather conditions.
They can operate in hostile environment where safety is a major concern like
obtaining valuable information from flooded areas, volcanoes, fog and in case of
emergency.
Since UAVs can operate without airspace requirements there is less restriction in
operation unlike in manned aircraft when flying within visual range and low altitude.
Images of high resolution are obtained since flight is closer to the vicinity of interest
which in turn allows Ground Sample Distance (GSD) of up to 1.5cm.

AIM:
The aim of this project is to develop a GeoData processing software for fully autonomous
Unmanned Aerial Vehicles (UAVs) which is able to process raw images.

OBJECTIVES:
To design an image processing software for geodata acquired by UAVs.
To develop an algorithm to process the raw image data.
To develop an algorithm that produces accurate images with maximum resolution of
1.5cm/pixel
To test and simulate the algorithm using Matlab software in collaboration with
OpenCV.
To design and develop a Graphical User Interface
To create a database to store the processed images.

METHODOLOGY
In this section of the document, the methods used to achieve the objectives of the project will
be presented. Starting by presenting the tasks to be performed, the sequence in which they
will be conducted and the time it will take to complete each task.
Tasks to be performed
i.
ii.
iii.
iv.
v.

Acquire raw SAR data from the geography department.

Create a Graphical User interface and add functionality to it from the specifications
and requirements of the software.
Develop a MySQL database to store the developed images.
Image processing algorithms are developed using Matlab /Simulink
Testing and prototyping.

In making this project a success, many information sources will be considered which include:
i.
ii.
iii.
iv.

Library resources such as textbooks and journals

Internet
Consulting the supervisor
Consulting the Geography department.

TIME SCHEDULE
0ct

Nov

Dec

Jan

Feb

March

Apri
l

May

Literature review
specifications and
design
Implementation
Integration
development
Testing and Results
Document Preparation

EQUIPMENT
Software
i.
Matlab
ii.
MySQL

LITERATURE REVIEW:
INTRODUCTION
This literature review discusses the theoretical background of Digital Image Processing as
well as the previous work done in the development of algorithms for image processing.

BRIEF HISTORY OF DIGITAL IMAGE PROCESSING;

Digital image processing as we know it was known as digital picture processing and most of
the techniques used today were developed in 1960s at Massachusetts Institute of technology,
Bell laboratories, Jet Propulsion Laboratory and a few other research organisations, with
main application to satellite imagery, character recognition, medical imaging and
enhancement of photographs.
During this era the cost of processing images was fairly high due to the computing equipment
available. Digital image processing flourished as cheaper computers and sophisticated
hardware and software became available in the 1970s. With the increase in computing power,
images then could be processed in near real time, for some challenges like television
standards conversion.
Nowadays, with fast processors and cheap computers, digital image processing has turned out
to be the most common method for image processing.
Gradient domain image processing was recently introduced in 2002 by Raanan Fattal, a new
method to manipulate differences between pixels rather than the pixel values between them in
an image.
DIGITAL IMAGE PROCESSING:
An image may be defined as a two-dimensional function, f(x, y), where x
and y are spatial coordinates, and the amplitude of the function at any
pair of coordinates (x, y) is called the grey level of the image at that
instant. An image is digital when the values x, y and the amplitude of the
function are all discrete and finite quantities.(Salman n.d.)
DIP is a field of Digital Signal Processing (DSP) where the input is an image and the output
is also an image and is defined as the use of computer algorithms to process digital images.
The advantage of DIP over analog image processing is that it allows a number of different
algorithms to be applied and avoids signal distortion and noise to build up during processing.
TYPES OF DIGITAL IMAGES:
Basically there are four types of digital images which are Binary, Greyscale, True colour
(RGB) and Indexed images.
Binary image
This is a digital image where the value of each pixel is either 0 or 1, or just black or white.
They are very efficient for digital storage since one bit represents one pixel. Architectural
plans, fingerprints or text images can be suitably represented by binary images.
5

Greyscale images; shades of grey are used represent each pixel from a range of 0(black) to
255(white).the size of the range means that each pixel is represented by one byte. Most
natural objects can be identified using this type of images for example printed objects and X
ray imagery.
True colour or RGB image.
In this type of image, the specific colour for each pixel is described by the amount of red,
green and blue in it. Each colour component has a range from 0 to 255 which gives 16 777
216 possible colour combinations for each pixel. Every pixel is described by 24 bits which
makes this type to be also known as 24 bit colour images.

Figure1.1 True RGB image.

Indexed image.
For the convenience of file handling and storage, indexed images have a colour map or
palette which is an array of all colour combinations used in the image and is referred to by the
value or index of each pixel.
ASPECTS OF IMAGE PROCESSING.
The four basic operations of digital image processing are image restoration, enhancement,
classification and transformation which are meant for different problems and tasks. A given
algorithm may perform one or more of this operations and we need to understand what we
want to obtain from any image.
Image restoration

This process is done to obtain a faithful representation of the earth surface by a way of
correcting and calibrating the images to reverse the damages by a known cause, for example,
removing optical distortions and blur due to linear motion.
Image Enhancement
This is performed to optimize the appearance of images for visual analysis by modifying the
characteristics of the image, for example, de blurring, removing noise and improving
contrast.
Image classification or segmentation
This is a process of interpreting images using the assistance of computers by segregating
certain features of an image, for example, identifying cars, roads or trees.
Image Transformation
This operation derives new imagery from raw image band using some mathematical
treatment.

REFERENCES
Parker, J., 2011. Algorithms for image processing and computer vision. Indianapolis: Wiley
Publishing,Inc.
Salman, N.H., NEW IMAGE PROCESSING TOOLBOX USING MATLAB CODES 3IMAGE TOOLBOX MENU 4- TESTS AND RESULTS 1-Thresholding Function Test
3-3-THRESHOLD TECHNIQUES. , pp.18.
Sieberth, T. et al., 2015. UAV Image Blur - Its Influence and Ways to Correct it. , XL, pp.33
39.