beginner_s guide to computer vision presentation

8/17/2019 Beginner_s Guide to Computer Vision Presentation

1/114

Introduction Learning Computer Vision via OpenCV Going Beyond

Beginner’s Guide to Computer Vision

Jayneil Dalal1 Sohil Patel2

1

Technical Writer/Embedded Linux EngineerIntelwww.elinux.org/Jayneil Dalal

2Computer Vision EngineerConsultant

July 24, 2013

http://www.elinux.org/Jayneil_Dalalhttp://www.elinux.org/Jayneil_Dalalhttp://find/


2/114


1 IntroductionAnatomy of the tutorialComputer Vision(CV)

2

Learning Computer Vision via OpenCVBasicsImplementing Computer Vision

3 Going Beyond

More OpenCV Codes3DThanks

http://find/


3/114


About US

We are open source advocates and evangelists.

We have recently started a hackerspace in India “Jayso Labs”where students can work on a wide varety of open sourcehardware boards(Arduino, Beaglebone Black, Raspberry Pi,Papiloboard, Pandaboard, Arndaleboard etc.) for just $10.

www.jaysolabs.com

http://www.jaysolabs.com/http://www.jaysolabs.com/http://find/


4/114


About US



www.jaysolabs.com

I d i L i C Vi i i O CV G i B d



5/114


About US



www.jaysolabs.com

I t d ti L i C t Vi i i O CV G i B d



6/114


Our Book

http://opencv.org/instant-opencv-starter.html


http://find/


7/114


Anatomy of the tutorial


2

Learning Computer Vision via OpenCVBasicsImplementing Computer Vision

3 Going Beyond



http://find/


8/114



Prerequisites

This tutorial is for beginners who have no prior experience toComputer Vision. Basic understanding of C/C++ is expected.

All the attendees will be given a DVD containing the followingUbuntu image with OpenCV 2.4.3 pre-installedVirtual box application for Windows & LinuxOpenCV tutorial codes

If you don’t have a DVD, you can download the tutorialmaterial from the link below:https://sourceforge.net/projects/oscon13cv/


https://sourceforge.net/projects/oscon13cv/https://sourceforge.net/projects/oscon13cv/http://find/


9/114



Prerequisites







10/114



Prerequisites







11/114

g p p g y


Prerequisites







12/114

g p p g y


Prerequisites







13/114

g g y


Setup

Please make sure that you have Visualization Technology(VT)enabled in your BIOS.

Then install Virtual box on your machine if you don’t have it.Create a virtual instance using the Ubuntu image providedwith atleast 1core, 1GB RAM.

Log into the live mode. DO NOT INSTALL UBUNTU!

Username=Password=opencvMake sure that the Codelite IDE has been pre-configured withthe tutorial codes.


http://find/


14/114


Setup






http://find/


15/114


Setup






http://find/


16/114


Setup






http://find/


17/114


Setup




Username=Password=opencv

Make sure that the Codelite IDE has been pre-configured withthe tutorial codes.


http://find/


18/114


Setup




Username=Password=opencv

Make sure that the Codelite IDE has been pre-configured withthe tutorial codes.


http://find/


19/114

Computer Vision(CV)


2 Learning Computer Vision via OpenCVBasicsImplementing Computer Vision

3 Going Beyond



http://find/


20/114

Computer Vision(CV)

What is it?

It is a discipline of “teaching machines how to see”.

Used in various industries - Biomedical, Robotics, Computers,Automotive, Aviation etc.


http://find/


21/114

Computer Vision(CV)

What is it?

It is a discipline of “teaching machines how to see”.

Used in various industries - Biomedical, Robotics, Computers,Automotive, Aviation etc.


http://find/


22/114

Computer Vision(CV)

Why learn Computer Vision?

Solve some of the world’s most toughest problems.

Its cool and fun. Also one does not need a PhD to implementit.


http://find/


23/114

Computer Vision(CV)

Why learn Computer Vision?

Solve some of the world’s most toughest problems.

Its cool and fun. Also one does not need a PhD to implementit.


http://find/


24/114

Computer Vision(CV)

Applications of Computer Vision

Motivation video

Used in all the latest cars today to assist drivers.

Used by a lot of companies like Facebook, Google etc. toidentify people in their photos.


http://find/


25/114

Basics



3 Going Beyond



http://find/


26/114

Basics

What is an Image?


http://find/


27/114

Basics

What is an Image?

It is nothing but a collection of ones and zeros! That is howmachines see them atleast.

It is made up of small constituent elements called ’pictureelements’ a.k.a pixels as shown below

Each pixel stores specific color information about the picture.A 1920*1080 picture has 2073600 pixels. More pixels meansclarity upto a certain point.

JPEG, PNG, BMP, GIF etc. - standardized means of

organizing and storing images.


B i

http://find/


28/114

Basics

What is an Image?







B i

http://find/


29/114

Basics

What is an Image?







Basics

http://find/


30/114

Basics

What is an Image?







Basics

http://find/


31/114

Basics

What is an Image?







Basics

http://find/


32/114

Basics

Terminologies

Color Models - RGB, HSV, CMYK.

A mathematical method to describe colors as a tuple of numbers.Colors can be additive or subtractive


Basics

http://find/


33/114

Basics

Terminologies

Saturation: INCREASE=bright colors DECREASE=grayishcolors

Brightness: LOW=DARK tones;HIGH=LIGHT tones

Contrast: Difference between the dark and light spots in theimage.


Basics

http://find/


34/114

Terminologies





Basics

http://find/


35/114

Terminologies





Implementing Computer Vision

http://find/


36/114

1

IntroductionAnatomy of the tutorialComputer Vision(CV)

2 Learning Computer Vision via OpenCV

BasicsImplementing Computer Vision

3 Going Beyond




http://find/


37/114

Why OpenCV

World’s most popular open source computer vision library.

Available under BSD license - great for making commercialapplications!

500+ optimized algorithms for image and video analysis

Cross platform- Mac OS, Windows, Android, iOS, Linux.

C, C++, Python and Java interfaces



http://find/


38/114

Why OpenCV








http://find/


39/114

Why OpenCV








http://find/


40/114

Why OpenCV








http://find/


41/114

Why OpenCV








http://find/http://goback/


42/114

Installing OpenCV

Two approaches: install pre-compiled package or build it from

source.For Ubuntu, all you need to do is type “sudo apt-get installlibopencv-dev” in a terminal window.

http://youtu.be/b3IQyhJXK78



http://youtu.be/b3IQyhJXK78http://youtu.be/b3IQyhJXK78http://find/


43/114

Installing OpenCV








44/114

Installing OpenCV








45/114

Resources

Website: http://opencv.org/

Official user group:http://tech.groups.yahoo.com/group/OpenCV/

Tutorials

Reference manual, user guide & cheatsheet available inOpenCV/doc folder.



R

http://opencv.org/http://tech.groups.yahoo.com/group/OpenCV/http://tech.groups.yahoo.com/group/OpenCV/http://opencv.org/http://find/


46/114

Resources



Tutorials




R



47/114

Resources



Tutorials




R



48/114

Resources



Tutorials




OpenCV Modules

http://opencv.org/http://tech.groups.yahoo.com/group/OpenCV/http://tech.groups.yahoo.com/group/OpenCV/http://opencv.org/http://find/http://goback/


49/114

OpenCV Modules

Core: compact module defining basic data structures, including the densemultidimensional array Mat and basic functions used by all other modules.

Imgproc: an image processing module that includes linear and non-linear imagefiltering, geometrical image transformations (resize, affine and perspectivewarping, generic table-based remapping), color space conversion, histograms,

and so on.Video: a video analysis module that includes motion estimation, backgroundsubtraction, and object tracking algorithms.

Calib3d: basic multiple-view geometry algorithms, single and stereo cameracalibration, object pose estimation, stereo correspondence algorithms, andelements of 3D reconstruction.

Features2d: salient feature detectors, descriptors, and descriptor matchers.Objdetect: detection of objects and instances of the predefined classes (forexample, faces, eyes, mugs, people, cars, and so on).

Highgui: an easy-to-use interface to video capturing, image and video codecs, aswell as simple UI capabilities.

Gpu: GPU - accelerated algorithms from different OpenCV modules.



U & R i



50/114

Up & Running

Please run the “Hello World” program.

ExplanationPlease run the “Modify Image” program.

Explanation

BREAK



Up & Running



51/114

Up & Running



Explanation

BREAK



Up & Running

http://find/


52/114

Up & Running



Explanation

BREAK



Up & Running

http://find/


53/114

Up & Running



Explanation

BREAK



Up & Running



54/114

Up & Running



Explanation

BREAK



Types of Image

http://find/


55/114

Types of Image

There are three types of Image:

RGB: 3 channels, 0-255Grayscale: 1 channel, 0[BLACK]-255[WHITE]Black and White: 1 channel, 0 or 1



Types of Image

http://find/


56/114

Types of Image





Types of Image

http://find/


57/114

Types of Image





RGB

http://find/


58/114

RGB



Grayscale

http://find/


59/114

Grayscale



Black and White

http://find/


60/114



Data types in OpenCV

http://find/


61/114

yp p

8-bit unsigned integer (uchar), CV 8U

8-bit signed integer (schar), CV 8S

16-bit unsigned integer (ushort), CV 16U16-bit signed integer (short), CV 16S

32-bit signed integer (int), CV 32S

32-bit floating-point number (float), CV 32F

64-bit floating-point number (double), CV 64F



Image storage in OpenCV

http://find/


62/114

g g p

Mat - Primitive datatype in OpenCV to store images.

More info here:

http://opencv.willowgarage.com/documentation/cpp/basic structures.html#cv-mat-mat

Image properties:height, width, depth, type, channels, pixels

Please run the “Matrix Operation” program.

Explanation






63/114

g g p


More info here:




Explanation




http://find/


64/114

g g


More info here:




Explanation




http://find/


65/114


More info here:




Explanation



RGB Color model

http://find/


66/114



HSV Color model

http://find/


67/114



Image Conversions

http://find/


68/114

RGB to Grayscale(reverse not possible!)

Grayscale to black and white.Makes processing much easier.

Please run the Image Conversion code.

Explanation



Image Conversions



69/114




Explanation



Image Conversions



70/114




Explanation



Image Conversions

http://find/


71/114




Explanation



Image Conversions

http://find/


72/114




Explanation



Pixel Manipulation

http://find/


73/114

Dealing with one/many pixel simultaneously.

Please run the Pixel manipulation code.

Explanation

Classic example: Image Steganography

Explanation



Pixel Manipulation

http://find/


74/114



Explanation


Explanation



Pixel Manipulation



75/114



Explanation


Explanation



Pixel Manipulation

http://find/


76/114



Explanation


Explanation



Pixel Manipulation

http://find/


77/114



Explanation


Explanation



Histogram

http://find/


78/114

It is a graph showing the number of pixels in an image at eachdifferent intensity value found in that image

Histograms are used to depict image statistics in an easilyinterpreted visual format



Histogram

http://find/


79/114





Histogram

http://find/


80/114





Plotting your own Histogram

http://find/


81/114



Importance of Histogram

http://find/


82/114

It is used in many applications to improve the visual

appearance of an image.

Trivial image comparison.




http://find/


83/114







http://find/


84/114






Equalization

http://find/


85/114

It is a method that improves the contrast in an image, inorder to stretch out the intensity range.



Equalization

http://find/


86/114




Equalization

http://find/


87/114



More OpenCV Codes



88/114

1 Introduction

Anatomy of the tutorialComputer Vision(CV)

2 Learning Computer Vision via OpenCV

BasicsImplementing Computer Vision

3 Going BeyondMore OpenCV Codes

3DThanks


More OpenCV Codes

Examples



89/114

Changing the brightness/contrast of a given image

Smoothening a given image

Edge detection

Please run all of the above codes.

Explanation


More OpenCV Codes

Examples

http://find/


90/114



Edge detection


Explanation


More OpenCV Codes

Examples

http://find/


91/114



Edge detection


Explanation


More OpenCV Codes

Examples

http://find/


92/114



Edge detection


Explanation


More OpenCV Codes

Examples



93/114



Edge detection


Explanation


3D

1 Introduction

http://find/


94/114

1 Introduction




3DThanks


3D

3D Image Processing

http://find/


95/114

Z parameter - depth.

3D Cameras use either ’time of flight’ or the structured lighttechnology.

Explanation


3D

3D Image Processing

http://find/


96/114



Explanation


3D

3D Image Processing



97/114



Explanation


3D

Hardware Devices

http://find/


98/114

Microsoft Kinect

Primesense

Softkinetic


3D

Hardware Devices



99/114

Microsoft Kinect

Primesense

Softkinetic


3D

Hardware Devices

http://find/


100/114

Microsoft Kinect

Primesense

Softkinetic


3D

Software



101/114

OpenNI library - world’s most popular open source 3D imageprocessing library.

Point Cloud Library(PCL)


3D

Software



102/114

OpenNI library - world’s most popular open source 3D imageprocessing library.

Point Cloud Library(PCL)


3D

Potential Applications

http://find/


103/114

Terrain mapping

Gaming

Medical


3D


http://find/


104/114

Terrain mapping

Gaming

Medical


3D


http://find/


105/114

Terrain mapping

Gaming

Medical


3D

Demos



106/114

Getting RGB data and depth map using Microsoft Kinect.

Hand TrackerSkeleton Tracking

Kinect hacks on Pandaboard


3D

Demos

http://find/


107/114





3D

Demos

http://find/


108/114





3D

Demos

http://find/


109/114





Thanks

1 Introduction

http://find/


110/114


2 Learning Computer Vision via OpenCVBasics



3DThanks


Thanks

Contact US

http://find/


111/114

Thank you for attending our tutorial! :-)

Jayneil: [email protected]: [email protected]

http://find/


112/114


Thanks

Contact US


113/114




Thanks

Contact US

http://find/


114/114


http://find/

beginner_s guide to computer vision presentation

Documents