CONTENT-BASED IMAGE RETRIEVAL

A picture speaks more than a thousand words !!

Presented By:
D.SRIKANTH V.M.SRI KRISHNA G.SRIRAM B.ABHILASH

INTRODUCTION

INTRODUCTION

Image Retrieval system for retrieving images from large database of digital images Common method of image retrieval utilizes metadata / keywords Manual image annotation is time consuming Locating desired image from small database is possible, where as in large database more effective techniques are needed
3

EXISTING SYSTEM

QBIC supports users to retrieve image by colour, shape and texture

QBIC provides several query methods

Simple Query Mutli-Feature Query Mutli-Pass Query

EXISTING SYSTEM

Photo Book system supports users to retrieve image by colour, shape and texture

Photo Book provides set of matching algorithms, divergence, vector space angle, histogram and Fourier peak

PROPOSED SYSTEM

Currently most widely used image search engine is GOOGLE. It provides its users with textual annotation. Not many images are annotated with proper description so many relevant images go unmatched CBIR uses Quadratic Distance & Integrated Regional Matching (I.R.M)

Quadratic Distance yield metric distance IRM is non-metric and gives result that are not optimal

PROPOSED SYSTEM

Our proposed system uses modified IRM and colour feature which overcomes above mentioned disadvantages We also provide an interface where user can give query images as input, automatically extracts the colour feature and compared with the images in database, retrieve the matching image
7

HARDWARE REQUIREMENTS
System Configuration:

Pentium III Processor with 700 MHz Clock Speed 256 MB RAM 20 GB HDD, 32 Bit PCI Ethernet Card.
8

SOFTWARE REQUIREMENTS
Operating System

Windows NT/2000 (Client/Server).

Software requirements

Java, JDK 1.4, J2SDK 1.4, Swings, RMI and Java Network Programming.
9

MODULES

10

MODULES

ADMINISTRATOR MODULE USER MODULE SEARCHING MODULE

11

ADMINISTRATOR MODULE

Maintaining the image database.

Update the database according to the users request. Classify the images for efficient searching.

12

USER MODULE

Upload the query images.

13

SEARCHING MODULE

Searching based on a given image. Integrate the search with the existing application. Combine querying techniques with content independent metadata.

14

IMAGE FEATURES

Texture (Laws, Gabor filters, local binary partition) Color (histograms, grid layout, wavelets) Shape (first segment the image, then use statistical or structural shape similarity measures) Objects and their Relationships

15

IMAGE FEATURE / HISTOGRAMS

Query Image User Image Database
Colour Measure

Retrieved Images

Images

Histogram

Comparison

16

TIGER IMAGE AS A COLOUR GRAPH

sky image above tiger above sand abstract regions

17

above adjacent inside grass above adjacent

Global Shape Properties: Tangent-Angle Histograms

135

0 30

45

135

18

Gridded Colour
Gridded colour distance is the sum of the color distances in each of the corresponding grid squares. 1 3 2 4 1 3 2 4

19

Object Detection: Rowleys Face Finder

1. Convert to gray scale 2. Normalize for lighting 3. Histogram equalization 4. Apply neural in 32 x 32 windowsnet(s) trained on 16K a pyramid structure images
20

UML DIAGRAMS

21

CLASS DIAGRAM
INPUT package image rawimj1 integer : package_imagecolomns1 integer : package_imagerows1 package_image_tracker1 integer : package_pix1 integer : package_pix3 integer : fileno float : he1 string : str public void main string() package input() DISPLAY private : thread image private : imagetodisplay ptivate : imagearray integer : noimgs integer : currentimage integer : sleeptime integer : imgcols1 integer : imgrows1 integer : pix1 integer : pix3 float : hes float : hes1 integer : fileno1 integer : n integer : l integer;k integer : m string : str1 string : str2 string : str3 string : str0 integer : x integer : y void init() void start() void suspend() void destroy() void run() void paint() void input123()

HISTOGRAM integer : imgno string : imgname float : he1 public histogram()

22

USE CASE DIAGRAM

query image

DBA image database user

visual content description

visual content description feature vector feature dabase includes

similarity comparsion

23
retrieval result

SEQUENCE DIAGRAM
USER
User Image Visual Content Feature Vector Similarity Result

Query Image()

Description()

Feature Vector()

Compare Similarity()

Retrive Result()

24

SEQUENCE DIAGRAM
DBA
DBA Image Visual Content Database Similarity Result User Create image Database()

Visual Content Description() Feature Database()

Includes()

Retrive result()

User()

25

HOME PAGE

26

HOME PAGE

27

HOME PAGE

28

HOME PAGE

29

CONCLUSION

30

CONCLUSION

Satisfactory progress Its easy to compute. Its more stable than the color histogram, QBIC, Photo Book methods.

31