Presenting by,

Prashanth B R

1AR08CS035

Dept.Of CSE.

AIeMS-Bidadi.

Sketch4Match – Content-based Image Retrieval System Using Sketches

Under the Guidance of,

Mr Gangadhar Immadi

Asst Prof, Dept of CSE

AIeMS-Bidadi.

Outline of Presentation Introduction

Objectives

Purpose of the system

Global structure and subsystems

Advantages

Tests and Results

Conclusion

References

Introduction to Sketch4Match-Content-based Image Retrieval System Using Sketches

The Content Based Image Retrieval is one of most popular, rising research areas of the digital image processing.Search engine tools- Google, Yahoo… etc.

• text based search..Goal of CBIR to extract visual content like color , text, or shape. Introduces design based on a free hand sketch --SBIRMaking search more efficient hereby.Test results show that the sketch based system allows users an intuitive access to search-tools.

Introduction to Sketch4Match-Content-based Image Retrieval System Using Sketches

In today’s corporate world huge data has to be managed, processed and stored.

The growing of data storages and revolution of internet had changed the world.

Text based search. Keywords! Is this efficient.

Why we use keywords for search- Giving unique & identifiable name to image is not difficult.Keyword acts as the human abstraction of that particular image.

Therefore it is to develop a CBIR system, which can retrieve using sketches in frequently used databases.

Drawing area to sketch the required image.

Matched images to sketch are retrieved.

Using Sketch based system can be very important and efficient in many areas of life.

Digital library. Crime prevention. photo sharing sites. search engines.

Introduction to Sketch4Match-Content-based Image Retrieval System Using Sketches

Objectives

This paper aims to introduce the problems and challenges concerned with design and creation of CBIR systems, which is based on free hand sketch.

Make convenient to retrieve data or images based on sketches so that even illiterates , who do not know to write text can also make use of system effectively.

Introducing this system into search engines makes corporate world and even other users bit more efficient in retrieval of data effectively.

Purpose of the system.

The Purpose of the system. Even though the measure of research in

sketch-based image retrieval is more, there is no widely used SBIR system.

The goal of this paper is to develop a SBIR search engine, which with free hand sketch content can be retrieved.

The most important task is to bridge the gap between the free hand sketch and the picture.

Global structure and subsystems.

1. The Global Structure of the System:

Displaying Subsystem

Preprocessing Subsystem

Feature Vector generating subsystem

Retrieval subsystem

Database Managementsubsystem

imagePreprocessed

image

ResultFeature vector

Stock index

Global Structure and subsystems.

2. The Preprocessing Subsystem:

The system is for databases containing simple images.

Preprocessing subsystem

Input OutputImage

Processed Image

Global structure and subsystems

3. The Feature Vector Preparation Subsystem: Here we use 3 Descriptor vectors which represents the content of the image.

Edge Histogram descriptor(EHD). Histogram of Oriented Gradients(HOG). Scale invariant feature transform(SIFT).

Purpose of the above descriptors, • preprocessing of free hand sketch.• Compression of free hand sketch with gallery of images.• Retrieval of matched images from the database

Global structure and subsystems

4. The Retrieval and Database Management Subsystem:

1. The Retrieval subsystem is used just to retrieve the matched image to the free hand sketch.

2. In Database Subsystem the images and their descriptors are stored and necessary mechanism for subsequent processing is provided.

3. This module consists of • The storage • The Retrieval • Data manipulation modules

Global structure and subsystems

5. The Display Subsystem:Since drawings are basis of retrieval.• The Drawing surface.• The screen for retrieval of matched images.

Advantages

Make convenient to retrieve data or images based on sketches so that even illiterates , who do not know to write text can also make use of system effectively.

Introducing this system into search engines makes corporate world and even other users bit more efficient in retrieval of data effectively.

Tests and Results

Used Test Databases:The system is tested with more than one sample database to obtain a more

extensive description. Few used databases are: 1. Flicker 160 Database 2. Microsoft Research Cambridge Object Recognition image Database3. Wang the database from corel image database.

Testing Aspects, Used Metrics: We can evaluate the effectiveness of the system forming methods, and

comparing different applied methods. This compression can be done easily through Metrics.

By applying EHD & HOG methods on these databases, we can find this system better than some systems before. So this system is more effective than the examined other systems.

Conclusion:

Among the objectives of this paper performed to design, implement and test a sketch-based image retrieval system. Two main aspects were taken into account. The retrieval process has to be highly interactive.The robustness of the method is essential in some degree of noise, which might also be in case of simple images.

Based on the test results with many databases HOG is more better in many cases than EHD.

References

[1] D. Comaniciu, and P. Meer, “Robust analysis of feature spaces: colorimage segmentation,” IEEE Conference on Computer Vision and PatternRecognition, pp. 750–755, June 1997.[2] N. Dalal, and B. Triggs, “Histograms of oriented gradients for humandetection,” IEEE Conference on Computer Vision and Pattern Recognition,pp. 886–893, July 2005.[3] T. Deselaers, D. Keysers, and H. Ney, “Features for image retrieval:an experimental comparison,” Information Retrieval, vol. 11, pp. 77–107,December 2007.[4] M. Eitz, K. Hildebrand, T. Boubekeur, and M. Alexa, “An evaluationof descriptors for large-scale image retrieval from sketched feature lines,”Computers and Graphics, vol. 34, pp. 482–498, October 2010.[5] R. Fabbri, L.D.F. Costa, J.C. Torelli, and O.M. Bruno, “2D Euclideandistance transform algorithms: a comparative survey,” ACM ComputingSurveys, vol. 44, pp. 1–44, February 2008.[6] M. Flickner, H. Sawhney, W. Niblack, J. Ashley, Q. Hiang, B. Dom, M.Gorkani, J. Hafner, D. Lee, D. Petkovic, D. Steele, and P. Yanker, “Queryby image and video content: the QBIC system,” IEEE Computer, vol. 28,pp. 23–32, 2002.

QUERIES ?

Thank you