This document provides an overview of OpenCV, an open-source computer vision library with over 2500 optimized algorithms, supporting languages like C++, Python, and Java, and applicable in various fields such as object recognition and mobile robotics. It highlights the modular structure, various modules and functionalities, including image processing and machine learning capabilities, along with installation guidance and problem-solving resources. The work is licensed under Creative Commons, permitting sharing and adaptation with proper attribution.

1. Computer Vision
OpenCV
Luigi De Russis
Politecnico di Torino
Dipartimento di Automatica e Informatica (DAUIN)
Torino - Italy
luigi.derussis@polito.it
This work is licensed under the Creative Commons (CC BY-SA)
License. To view a copy of this license, visit
http://creativecommons.org/licenses/by-sa/3.0/

2. What is OpenCV?
 Open source Computer Vision library
 http://opencv.org/
 Originally developed by Intel
 Has more than 2500 optimized algorithms
 C/C++/Python API
 it is written natively in C++
 Cross-platform
includes a
 also available for Android and iOS
Java API
 Released under a BSD license (it’s free)
 Current release version: 2.4.3RC (October 2012)
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 2

3. Various applications…
 Human-Computer Interaction (HCI)
 Object Identification
 Object Recognition
 Face Recognition
 Gesture Recognition
 Motion Tracking
 Image Processing
 Mobile Robotics
 … and so on.
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 3

4. Why OpenCV? (I)
 Best competitor here: Matlab
 Pros
 Specific
 OpenCV was made for image processing
 Matlab is quite generic
 Speed
 around 30 frames processed per seconds in real time image
processing (OpenCV)
 around 4-5 frames processed per seconds in real time image
processing (Matlab)
 Efficient
 Matlab needs more system resources than OpenCV
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 4

5. Why OpenCV? (II)
 Cons
 Easy of use
 Integrated Development Environment
 you can use Eclipse, Netbeans, Visual Studio, Qt, XCode, … a
simple text editor for OpenCV
 Memory management
 Two more “pros”…
 Price (!)
 OpenCV Wrappers
 SimpleCV, JavaCV, Emgu CV, JavacvPro, …
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 5

6. Modules (I)
 OpenCV has a modular structure, i.e., the package
includes several shared or static libraries:
 core
 basic structures and algorithms
 imgproc
 Image Processing algorithms (such as image filtering,
geometrical image transformations, histograms, etc.)
 video
 video analysis (such as motion estimation and object tracking)
 highgui
 built-in simple UI
 in addition, we will use Qt
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 6

7. Modules (II)
 calib3d
 camera calibration and 3D reconstruction
 features2d
 2D features framework (feature detectors, descriptors, and
descriptor matchers)
 objdetect
 detection of objects and other items (e.g., faces, eyes, mugs,
people, …)
 ml
 machine learning classes used for statistical classification,
regression and clustering of data
 gpu
 GPU-accelerated algorithms
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 7

8. Data structures (I)
 We speak about C++ API
 All the OpenCV classes and functions are placed
into the cv namespace
 Mat
 the primary image structure in OpenCV 2.x
 overcomes the “old” IplImage/CvMat problems
(OpenCV 1.x/C API)
 automatic memory management (more or less)
 two data parts:
 matrix header (contains information about the matrix)
 a pointer to the matrix containing the pixel values
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 8

9. Data structures (II)
 Point_ (Point2f, Point, Point2d)
 2D point
 defined by x, y coordinates
 Point first(2, 3);
 Size_ (Size, Size2f)
 2D size structure
 specify the size (width and height) of an image or
rectangle
 Rect
 2D rectangle object
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 9

10. Image I/O
 Image I/O
 imread
 legge un’immagine da file e lo salva in un oggetto di tipo Mat
Mat imread(const string& filename, int flags=1)
 imwrite
 salva un’immagine su file
bool imwrite(const string& filename, InputArray img,
const vector<int>& params=vector<int>())
 imshow
 mostra un’immagine a schermo (in una finestra creata
precedentemente)
void imshow(const string& winname, InputArray mat)
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 10

11. Drawing operations
 Base drawing operations
 circle
 draws a simple or filled circle with a given center and radius on a
given image
 line
 draws a line between two point in the given image
 ellipse
 draws an ellipse outline, a filled ellipse, an elliptic arc, a filled
ellipse sector, …
 rectangle
 draws a rectangle outline or a filled rectangle
 note that negative thickness will fill the rectangle
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 11

12. Color spaces
 Converting color spaces
 cvtColor
 converts an input image from one color space to another
 examples:
 cvtColor(src, dest, CV_RGB2GRAY);
 cvtColor(src, dest, CV_HSV2BGR);
 cvtColor(src, dest, CV_RGB2BGR);
 important, since images in OpenCV uses BGR instead of
RGB
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 12

13. How we can use OpenCV?
 LABINF:
 already installed under Windows
 version 2.3.1
 Qt Creator (4.7.4) is the IDE to be used
 At home:
 feel free to install OpenCV version 2.4.3
 it should be more “stable”
 you can use whatever IDE you like
 but we give full support only for Qt Creator
 Installation:
 see the next set of slides for a step-by-step guide
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 13

14. What if I got problems?
 Small problems
 drop me a line Problems with Qt and a gray scale image
 luigi.derussis@polito.it
Awesome student to me
 Normal problems Hi,
[…] I’m using “cvtColor(image, gray, CV_BGR2GRAY);” and Indexed8 to convert an image in
gray scale but the image is not shown in Qt…
 come to office hours Can you help me?
Regards,
…
 every Tuesday, 9:00 - 11:00
 LAB6, second floor of DAUIN
 please send an e-mail beforehand
 Enormous problems
 come to the dedicated lessons
 Tuesday 30th October, 9:30-12:30, DAUIN
 Thursday 8th November, 15:00-18:00, DAUIN
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 14

15. What if I got problems?
OpenCV installation
 Small problems Not-So-Awesome student to me
 drop me a line Hi,
[…] I followed the guide for installing OpenCV on my Mac but I have an error after step 3. Can
we meet on next Tuesday to solve the problem?
 luigi.derussis@polito.it Thanks!
Regards,
 Normal problems …
 come to office hours
 every Tuesday, 9:00 - 11:00
 LAB6, second floor of DAUIN
 please send an e-mail beforehand
 Enormous problems
 come to the dedicated lessons
 Tuesday 30th October, 9:30-12:30, DAUIN
 Thursday 8th November, 15:00-18:00, DAUIN
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 15

16. What if I got problems?
 Small problems
 drop me a line
 luigi.derussis@polito.it
 Normal problems Help with OpenCV
 come to office hours Good student to me
Hi,
 every Tuesday, 9:00 - 11:00 […] I see the solution of Exercise 2.1 but I don’t understand the following expressions:
- main();
- int* number;
 LAB6, second floor of DAUIN - &timer.
Can you explain to me what they are?
 please send an e-mail beforehand Regards,
…
 Enormous problems
 come to the dedicated lessons
 Tuesday 30th October, 9:30-12:30, DAUIN
 Thursday 8th November, 15:00-18:00, DAUIN
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 16

17. An mail not to be sent!!!
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 17

18. Resources
 OpenCV Wiki
 http://code.opencv.org/projects/opencv/wiki
 OpenCV 2.x Official Documentation
 http://docs.opencv.org/
 User Q&A forum
 http://answers.opencv.org/questions/
 OpenCV 2.x Tutorials
 http://docs.opencv.org/opencv_tutorials.pdf
 Books:
 Robert Laganière, OpenCV 2 Computer Vision Application Programming
Cookbook, Packt Publishing, ISBN 978-1849513241
 Gary Bradsky, Adrian Kaehler, Learning OpenCV: Computer Vision in C++ with
the OpenCV Library, O'Reilly Media, ISBN 978-1449314651 (to be published)
10/26/2012 Luigi De Russis - Computer Vision - OpenCV 18

19. License
This work is licensed under the Creative Commons
“Attribution-NonCommercial-ShareAlike Unported (CC BY-
NC-SA 3,0)” License.
You are free:
 to Share - to copy, distribute and transmit the work
 to Remix - to adapt the work
Under the following conditions:
 Attribution - You must attribute the work in the manner
specified by the author or licensor (but not in any way that
suggests that they endorse you or your use of the work).
 Noncommercial - You may not use this work for commercial
purposes.
 Share Alike - If you alter, transform, or build upon this work,
you may distribute the resulting work only under the same or
similar license to this one.
26/10/2012 Luigi De Russis - Computer Vision - OpenCV 19