3
Chapter 1
Introduction to Computer Vision
and OpenCV
A significant share of the information that we get from the world while we are awake is through sight. Our eyes do
a wonderful job of swiveling about incessantly and changing focus as needed to see things. Our brain does an even
more wonderful job of processing the information stream from both eyes and creating a 3D map of the world around
us and making us aware of our position and orientation in this map. Wouldn't it be cool if robots (and computers in
general) could see, and understand what they see, as we do?
For robots, seeing in itself is less of a problem—cameras of all sorts are available and quite easy to use. However,
to a computer with a camera connected to it, the camera feed is technically just a time-varying set of numbers.
Enter computer vision.
Computer vision is all about making robots intelligent enough to take decisions based on what they see.
Why Was This Book Written?
In my opinion, robots today are like personal computers 35 years ago—a budding technology that has the potential
to revolutionize the way we live our daily lives. If someone takes you 35 years ahead in time, don't be surprised to see
robots roaming the streets and working inside buildings, helping and collaborating safely with humans on a lot of
daily tasks. Don't be surprised also if you see robots in industries and hospitals, performing the most complex and
precision-demanding tasks with ease. And you guessed it right, to do all this they will need highly efficient, intelligent,
and robust vision systems.
Computer vision is perhaps the hottest area of research in robotics today. There are a lot of smart people all
around the world trying to design algorithms and implement them to give robots the ability to interpret what they see
intelligently and correctly. If you too want to contribute to this field of research, this book is your first step.
In this book I aim to teach you the basic concepts, and some slightly more advanced ones, in some of the most
important areas of computer vision research through a series of projects of increasing complexity. Starting from
something as simple as making the computer recognize colors, I will lead you through a journey that will even teach
you how to make a robot estimate its speed and direction from how the objects in its camera feed are moving.
We shall implement all our projects with the help of a programming library (roughly, a set of prewritten functions
that can execute relevant higher-level tasks) called OpenCV.
This book will familiarize you with the algorithm implementations that OpenCV provides via its built-in functions,
theoretical details of the algorithms, and the C++ programming philosophies that are generally employed while using
OpenCV. Toward the end of the book, we will also discuss a couple of projects in which we employ OpenCV’s framework
for algorithms of our own design. A moderate level of comfort with C++ programming will be assumed.
