1 eye detection in images introduction to computational and biological vision lecturer : ohad ben...

20
1 Eye Detection in Eye Detection in Images Images Introduction To Computational and Introduction To Computational and biological Vision biological Vision Lecturer Lecturer : : Ohad Ben Shahar Ohad Ben Shahar Written by : Itai Bechor

Upload: augusta-owens

Post on 13-Dec-2015

218 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

1

Eye Detection in ImagesEye Detection in Images

Introduction To Computational and Introduction To Computational and biologibiological Vision cal Vision

LecturerLecturer : : Ohad Ben Shahar Ohad Ben Shahar

Written by : Itai Bechor

Page 2: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

2

Chapter HeadingsChapter Headings

Introduction Introduction The Main algorithm:The Main algorithm:

Detecting the face areaDetecting the face area Find a goodFind a good candidatescandidates Find the most probability For Eyes in ThFind the most probability For Eyes in Th

ee ImageImage Conclusions and ResultsConclusions and Results

Page 3: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

3

IntroductionIntroduction

Detecting Eyes has many applications:Detecting Eyes has many applications:• For Face RecognitionFor Face Recognition• May Be Use By The PoliceMay Be Use By The Police• In Security ServicesIn Security Services• Future Use In Computers Security For Future Use In Computers Security For

Login PropsesLogin Propses

Page 4: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

4

IntroductionIntroduction

The Eye is Quite Unique FeatureThe Eye is Quite Unique Feature in in the Facethe Face

It might be easy to detect it more It might be easy to detect it more than other elements in the facethan other elements in the face

The Objective is To detect the The Objective is To detect the Closest Area To the eyes or the Closest Area To the eyes or the Eyes Eyes

Page 5: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

5

The Algorithm DiagramThe Algorithm Diagram

Detect face Detect the edge

Find radius that suits eye Detect the eyes

Page 6: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

6

Images I work withImages I work with

Black and white images Black and white images Head Images On a Plain BackgroundHead Images On a Plain Background Image resolution of 150x150 to 300x30Image resolution of 150x150 to 300x30

00

Page 7: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

7

Extraction of the face Extraction of the face regionsregions

Step 1 Input Image

Step 1 Input Image

Step 2 Canny Edge detector

Step 2 Canny Edge detector

Step 3 Calculate the left and right bound

Step 3 Calculate the left and right bound

x

V(x)

N

M

Page 8: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

8

Face Region Extraction

Page 9: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

9

The Canny Edge DetectorThe Canny Edge Detector

I used Gaussian 5x5 I used Gaussian 5x5 convolution To smooth the convolution To smooth the image to clean the noiseimage to clean the noise

Page 10: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

11

Canny Edge DetectorCanny Edge Detector

Compute gradient of g(m,n) using to get:Compute gradient of g(m,n) using to get:

and and

And finally by threshold m:And finally by threshold m:

Page 11: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

12

Hough Circle TransformationHough Circle Transformation

in my program : I Find The Circles In The Image From Radius 1 to width/2.

A circle in 2d is :

The accumulator Holding the Votes For each Radius.

Edge pointEdge point

r

(Xi,Yi)

Largest vote (a,b)Largest vote (a,b)

Page 12: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

13

Hough Circle TransformationHough Circle Transformation

Page 13: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

14

Hough Circle TransformatioHough Circle Transformationn

Page 14: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

15

Selecting the Eyes Selecting the Eyes

Labeling Function That Find the best Match Between Two Circles In The Eyes

Page 15: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

16

Selecting the EyesSelecting the Eyes

Using the Following Methods:1. Calculate the Distances between each t

wo circles . 2.2. The Slope Between The Two Circles. The Slope Between The Two Circles.

3.3. The Radius similarity between two The Radius similarity between two circles.circles.

4.4. Large Number of circles in the same Large Number of circles in the same areaarea

Page 16: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

17

Experimental ResultsExperimental Results Good Results:Good Results:

Page 17: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

18

Experimental ResultsExperimental Results

Page 18: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

19

Experimental ResultsExperimental Results Bad Result: Hough Didn’t detect Bad Result: Hough Didn’t detect

eye circleseye circles

Page 19: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

20

Experimental ResultsExperimental Results Bad Result: Label Function Didn’t Bad Result: Label Function Didn’t

detect eyes.detect eyes.

Page 20: 1 Eye Detection in Images Introduction To Computational and biological Vision Lecturer : Ohad Ben Shahar Written by : Itai Bechor

21

ConclusionConclusion

The Algorithm need to be The Algorithm need to be improvedimproved

In Order To Improve it :In Order To Improve it :1.1. Need To Use A Eyes DatabaseNeed To Use A Eyes Database

2.2. There is special cameras that can There is special cameras that can detect the eye using an effect called detect the eye using an effect called The bright pupilThe bright pupil effect .effect .