Sunteți pe pagina 1din 31

1

REVERSE IMAGE SEARCH


GROUP
6

Group Members:
Vrujal Gandhi

(201103100710010)

Jimit Vaidya

(201103100710054)

Denish Jariwala

(201203100720007)

Nikunj Rana

(201203100720021)

Guided By:
Prof. Mayank Kapadia
Assistant Professor
C. G. Patel Institute of Technology

OUTLINE

Introduction

Motivation

Objective

Methods/ Algorithms/ Block Diagrams

Results

Application

Conclusion

References

Demo

INTRODUCTION

Reverse Image Search

Searching an Image by using a query Image, rather than the text.

MOTIVATION

We can find similar, relevant and unknown images by using Reverse Image
Search.

Object-1

Object-2

Object-3

OBJECTIVE

Ease of searching.

Reduced searching time.

More results based on visual similarities.

6
TEXT BASED SEARCH
SEARCH

Type keyword in
search engine

Lookup for
matching images
in database

Result
(Images)

vs

IMAGE BASED

Input Image

Result
(Similar images
to query image )

Extract the
features and
contents

Look for similar


features and
content in image
database

RIS : BLOCK DIAGRAM

Query Image

Feature Extraction

Feature Vector

e.g. Histogram/DWT/Edge

Similarity
Measurement
e.g. ED

Image Database

Feature Extraction
of each Image
e.g. Histogram/DWT/Edge

Features
Database

Retrieved
Relevant Images

TECHNIQUES

Color:

Shape:

1.

Global Histogram

1.

Canny Edge Detector

2.

Local Histogram

2.

Sobel Operator

3.

Prewitt Operator

. Texture:
1.

Discrete Wavelet
Transform

2.

Principal Component Analysis

GLOBAL HISTOGRAM

Colour Technique

Global Histogram :

Image Pixel value


Feature = [mean2(R) mean2(G) mean2(B)]

RGB Plane

10

LOCAL HISTOGRAM

Colour Technique

Local Histogram :

Image

Part - 1

Part - 2

Part - 3

Part - 4

Feature = [mean2(Rn) mean2(Gn) mean2(Bn)]


Where, n = number of parts

DISCRETE WAVELET TRANSFORM

11
Texture

Technique

To analysis texture feature using Discrete Wavelet Transform for three


level is given as,

DISCRETE WAVELET TRANSFORM

12
Texture

Technique

One Level of the Transform

The DWT of a signal is calculated by passing it through a series of filters. First the
samples
are passed through a low pass filter with impulse response resulting in a
convolution of the two:

y[ n] ( x * g )[ n]

x[k ]g[n k ]

13

CANNY EDGE DETECTOR

Shape Technique

The Process of Canny edge detection algorithm can be broken down to 5


different steps:

1.

Apply Gaussian filter to smooth the image in order to remove the noise

2.

Find the intensity gradients of the image

3.

Apply non-maximum suppression to get rid of spurious response to edge


detection

4.

Apply double threshold to determine potential edges

5.

Track edge by hysteresis: Finalize the detection of edges by suppressing


all the other edges that are weak and not connected to strong edges.

Performance Measurement
Parameters

x 100 ] %

100 ] %

14

15

Similarity Measurement Parameters

E.D.

bn

where, n = number of samples

16

Query Image
Query Image

17

LOCAL HISTOGRAM

Results

10

11

12

13

14

15

16

18

DISCRETE WAVELET TRANSFORM


1

10

11

12

13

14

15

16

Results

19

CANNY EDGE DETECTOR

Results

10

11

12

13

14

15

16

20

FUSION OF THREE

Results

10

11

12

13

14

15

16

21

RESULT ANALYSIS

Query Image Local Hist

DWT

Canny

Fusion

16

29

10

10

38

42

12

12

55

Average
Precision
Rate

5.2

9.2

56.25

32.50

12.50

57.50

Recall Rate

14.29

8.25

3.17

14.60

173

14

14

179

12

11

180

14

14

192

10

11

10

197

11

11

Average
Precision
Rate

12.2

7.2

12

76.25

25.00

45.00

75.00

Recall Rate

13.71

4.49

8.09

13.48

African
People

Beach

Total Images in
Database

63

89

22

RESULT ANALYSIS

Query Image Local Hist

DWT

Canny

Fusion

224

10

10

242

10

10

243

244

11

10

248

Average
Precision
Rate

9.8

3.2

9.6

61.25

20.00

6.25

60.00

Recall Rate

49.00

16.00

5.00

48.00

302

11

10

330

14

14

332

12

12

346

11

11

368

12

13

Average
Precision
Rate

12

4.8

12

75.00

30.00

6.25

75.00

Recall Rate

60.00

24.00

5.00

60.00

Building

Bus

Total Images in
Database

84

99

23

RESULT ANALYSIS

Query Image Local Hist

DWT

Canny

Fusion

401

16

16

16

16

434

16

16

16

16

454

16

16

10

16

466

16

16

16

16

497

16

16

16

16

Average
Precision
Rate

16

16

14.8

16

100.00

100.00

92.50

100.00

Recall Rate

80.00

80.00

74.00

80.00

517

14

14

521

15

10

15

529

14

14

544

14

14

553

14

11

14

Average
Precision
Rate

14.2

9.2

14.2

88.75

57.50

18.75

88.75

Recall Rate

71.00

46.00

15.00

71.00

Dragon

Elephant

Total Images in
Database

100

100

24

RESULT ANALYSIS

Query Image Local Hist

DWT

Canny

Fusion

600

15

15

611

15

11

15

627

14

14

636

16

16

644

16

16

Average
Precision
Rate

15.2

5.8

4.6

15.2

95.00

36.25

28.75

95.00

Recall Rate

76.00

29.00

23.00

76.00

700

15

15

701

16

10

16

702

16

14

16

774

15

15

15

791

16

13

16

Average
Precision
Rate

15.6

4.2

10.6

15.6

97.50

26.25

66.25

97.50

Recall Rate

78.00

21.00

53.00

78.00

Flower

Horse

Total Images in
Database

100

100

25

RESULT ANALYSIS

Query Image Local Hist

DWT

Canny

Fusion

816

11

11

818

13

13

829

13

11

13

843

12

12

852

13

13

Average
Precision
Rate

12.4

7.2

12.4

77.50

25.00

45.00

77.50

Recall Rate

62.00

20.00

36.00

62.00

937

13

10

13

942

14

14

950

13

12

974

13

13

977

13

13

Average
Precision
Rate

13.20

2.60

5.40

13.00

82.50

16.25

33.75

81.25

Recall Rate

66.00

13.00

27.00

65.00

Mountain

Food

Total Images in
Database

95

100

26

APPLICATIONS

Detect copyright violations

Architectural and Engineering Design

Track images on the internet

Art Collection

Search relevant content instantly

Crime Prevention

Forensic Lab Research

Photograph Achieves

Bio-Medical Research

Nudity-Detection Filters

Unknown Entities Identification

Face Recognition and Finding

And lots more...

27

CONCLUSION

Reverse Image Search (RIS) performs well as all three feature extraction
techniques are used together. Whichever feature extraction technique we
use has certain drawbacks and to cover them up we used the fusion of all
three feature extraction technique. It shows the relevant retrieved images as
per the query image with more accuracy by using more advanced algorithm
and reduces time taken by loading pre-defined image database in the code.

28

REFERENCES

1.

D.Jeyabharathi and Dr.A. Suruliandi, Performance Analysis of Feature Extraction and Classification Techniques in CBIR, 2013
International Conference on Circuits, Power and Computing Technologies [ICCPCT-2013]

2.

Amandeep Khokher and Rajneesh Talwar, Content-based Image Retrieval: Feature Extraction Techniques and Applications,
International Conference on Recent Advances and Future Trends in Information Technology (iRAFIT2012)

3.

Bouden Toufik and Nibouche Mokhtar, The Wavelet Transform for Image Processing Applications, Advances in Wavelet Theory and
Their Applications in Engineering, Physics and Technology, Dr. Dumitru Baleanu (Ed.), ISBN: 978-953-51-0494-0, InTech (2012)

4.

Amanbir Sandhu and Aarti Kochhar, Content Based Image Retrieval using Texture, Color and Shape for Image Analysis
International Journal of Computers & Technology ISSN: 2277-3061 Volume 3, No. 1, AUG, 2012

29

REFERENCES
5.

Ms. K. Arthi and Mr. J. Vijayaraghavan, Content Based Image Retrieval Algorithm Using Colour Models International
Journal of Advanced Research in Computer and Communication Engineering Vol. 2, Issue 3, March 2013

6.

Ahmed J. Afifi and Wesam M. Ashour, "Content-Based Image Retrieval Using Invariant Color and Texture Features" Digital
Image Computing Techniques and Applications (DICTA), 2012 International Conference on, At Fremantle, Western
Australia

7.

Dandge S.S. And Bodkhe A.P., "Content Based Image Retrieval System" Journal of Signal and Image Processing, Volume 3,
Issue 2, 2012

8.

Jun Yue, Zhenbo Li, Lu Liu and Zetian Fub,"Content-based image retrieval using color and texture fused features"
Mathematical and Computer Modelling, Volume 54, Issues 34, August 2011

30

REFERENCES
9.

Nidhi Singh,Kanchan Singh and Ashok K. Sinha,"A Novel Approach for Content Based Image Retrieval" 2nd
International Conference on Computer, Communication, Control and Information Technology( C3IT-2012) on
February 25 - 26, 2012

10.

Shankar M. Patil, "A Content Based Image Retrieval using color, texture and shape", International Journal of
Computer Science and Engineering Technology", Vol. 3 , No. 9 ,pp. 404-410, 2012.

11.

Irena Valova and Boris Rachev,"A Content Based Image Retrieval System Based on Color Features" November
2004, CODATA, Berlin Germany

31

DEMO and Q & A

THANK YOU !

S-ar putea să vă placă și