SlideShare a Scribd company logo
1 of 30
Search Engine for Images We are given a large database comprising several categories of images like: people, landscape, flowers, buses, trains, food items, animals etc. There are numerous pictures in each category.
Our Goal A user can pick any image (not necessarily inside the database) and our program (the search engine) should be able to pull out pictures from the database that are similar to the query image.
Examples of similar images
Our intention is to produce results as close to human perception as possible. In other words, our search engine automatically categorizes the pictures.
Mathematical Tools Needed The Wavelet Transform. A considerable amount of Linear algebra (particularly Inner product spaces) has to be covered before we are in a position to describe the wavelet transform. The K-means clustering algorithm. This is a powerful statistical tool.
I.T. Tools Needed We chose MATLAB 7.0 for the following reasons: It handles matrices in a convenient manner It supports a powerful programming language It can handle images in almost every format
The Method The core of this search engine is the Integrated Region Matching Scheme described by Wang, Li, and Wiederhold in the paper: SIMPLIcity: Semantics-Sensitive Integrated Matching for Picture Libraries, IEEE, 2001. We now briefly outline the method.
A brief outline of the method Partition a picture into 4x4 blocks and extract feature vectors for each block. Group the feature vectors into a number of regions (using K-means). Compute the signature of the image by combining the features of all its regions. Save this signature to a database. Repeat the process for every picture in the database.
A brief outline of the method The feature vectors contain information regarding colour, texture and shape. The query image given by the user also undergoes the same signature extracting process. Finally distances between the query signature and each image signature are calculated and sorted to give the closest matches.
In the slides that follow we will make the above ideas precise.
What is a Feature vector? It is simply a list of 6 numbers calculated for each 4x4 block. The first three numbers are averages of the Red, Green and Blue components. The next three numbers are the root mean square of wavelet coefficients arising by applying the wavelet transform once to rows and columns of the block.
Meaning of these components The first three numbers hold information about the colour. The next three numbers hold information about the texture.
Image Segmentation This is what you get when K means is applied to the set of feature vectors for various values of K. All these pictures (except the original!) were created in MATLAB.
7 regions 4 regions Original 3 regions 2 regions
Remark Figuring out an appropriate value of K for any image still remains an open problem in computer vision. But Wang, Li and Weiderhold propose a scheme that calculates a K that works quite well for this experiment.
Integrated Region Matching This method combines the properties of all the regions in a picture to measure the overall similarity between images. The payoff is that the scheme provides robustness against poor segmentation.
Integrated Region Matching Defining a similarity measure is nothing but defining a distance between sets of points (feature vectors) in a higher dimensional space. The idea of distance must be carefully chosen so that it is consistent with a person’s idea of “closeness” of two images.
Integrated Region Matching Defining a similarity measure is nothing but defining a distance between sets of points (feature vectors) in a higher dimensional space. The idea of distance must be carefully chosen so that it is consistent with a person’s idea of “closeness” of two images.
Integrated Region Matching Suppose two images A and B are represented by region sets A=r1,r2,…,rm and B=r′1,r′2,…,r′n  
Integrated Region Matching Denote by dij the distance between ri and r′j. Wang et. al. give a simple prescription for this that they found experimentally. The distance between the images is defined as the weighted sum of region-to-region matches: dA,B=i,jsijdij  
Integrated Region Matching The weights sij are the elements of the significance matrix s11⋯s1n⋮⋱⋮sm1⋯smn Wang et.al. have devised an algorithm to calculate the significance matrix.  
Meaning of the weights The weights sij capture the importance of match.  For instance if a region consists of a body of an animal and is matched to various regions in the other picture then body-to-body match will be given more weight than a body-to-background or a body-to-tree match.  
Distance between Regions Two regions r and r′ comprise of several feature vectors.  We represent these two regions by single feature vectors f and f′ constructed as follows: The ith component is the average of the ithcomponent of all FVs in that region. Now the distance between r and r′ is given by dr,r′=i=06fi−f′i2  
Experimental Results We coded up this strategy in MATLAB 7.0 and ran the program on a database of 1000 images. Some outputs are shown in the next few slides. The Top left image is the query image.
Final Remarks One group (Eagle 2) has some original contributions to make: They had come up with their own criteria of calculating the appropriate number of regions for an image.  They even implemented their own K-means routine instead of relying on MATLAB’s built-in function.
Final Remarks One of our IAYM student ShubhankarBiswas, NIT, Durgapur has presented this project at the  Seminar on Applications of Computer and Embedded Technology organized by the Variable Energy Cyclotron Centre (VECC), Kolkata, in Oct. 2009 (2 months after IAYM 2009).  His paper can be viewed at http://www.vecc.gov.in/~sacet09/index_files/Page479.htm He has acknowledged the MSF for all the guidance.

More Related Content

What's hot

Pillar k means
Pillar k meansPillar k means
Pillar k meansswathi b
 
Knn Algorithm presentation
Knn Algorithm presentationKnn Algorithm presentation
Knn Algorithm presentationRishavSharma112
 
K Nearest Neighbor V1.0 Supervised Machine Learning Algorithm
K Nearest Neighbor V1.0 Supervised Machine Learning AlgorithmK Nearest Neighbor V1.0 Supervised Machine Learning Algorithm
K Nearest Neighbor V1.0 Supervised Machine Learning AlgorithmDataMites
 
Comparison of various Image Registration Techniques with the Proposed Hybrid ...
Comparison of various Image Registration Techniques with the Proposed Hybrid ...Comparison of various Image Registration Techniques with the Proposed Hybrid ...
Comparison of various Image Registration Techniques with the Proposed Hybrid ...idescitation
 
Btv thesis defense_v1.02-final
Btv thesis defense_v1.02-finalBtv thesis defense_v1.02-final
Btv thesis defense_v1.02-finalVinh Bui
 
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...Daksh Raj Chopra
 
System architecture
System architectureSystem architecture
System architectureSanjay Raj
 
PCA-SIFT: A More Distinctive Representation for Local Image Descriptors
PCA-SIFT: A More Distinctive Representation for Local Image DescriptorsPCA-SIFT: A More Distinctive Representation for Local Image Descriptors
PCA-SIFT: A More Distinctive Representation for Local Image Descriptorswolf
 
20 bayes learning
20 bayes learning20 bayes learning
20 bayes learningTianlu Wang
 
Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...
Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...
Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...wl820609
 

What's hot (20)

Project 1
Project 1Project 1
Project 1
 
Color
ColorColor
Color
 
Pillar k means
Pillar k meansPillar k means
Pillar k means
 
Project 2
Project 2Project 2
Project 2
 
Project 8
Project 8Project 8
Project 8
 
Matlab Working With Images
Matlab Working With ImagesMatlab Working With Images
Matlab Working With Images
 
Knn Algorithm presentation
Knn Algorithm presentationKnn Algorithm presentation
Knn Algorithm presentation
 
K Nearest Neighbor V1.0 Supervised Machine Learning Algorithm
K Nearest Neighbor V1.0 Supervised Machine Learning AlgorithmK Nearest Neighbor V1.0 Supervised Machine Learning Algorithm
K Nearest Neighbor V1.0 Supervised Machine Learning Algorithm
 
Comparison of various Image Registration Techniques with the Proposed Hybrid ...
Comparison of various Image Registration Techniques with the Proposed Hybrid ...Comparison of various Image Registration Techniques with the Proposed Hybrid ...
Comparison of various Image Registration Techniques with the Proposed Hybrid ...
 
Btv thesis defense_v1.02-final
Btv thesis defense_v1.02-finalBtv thesis defense_v1.02-final
Btv thesis defense_v1.02-final
 
Self-organizing map
Self-organizing mapSelf-organizing map
Self-organizing map
 
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
 
PLACES DATABASE
PLACES DATABASEPLACES DATABASE
PLACES DATABASE
 
Project 5
Project 5Project 5
Project 5
 
System architecture
System architectureSystem architecture
System architecture
 
poster
posterposter
poster
 
Project 6
Project 6Project 6
Project 6
 
PCA-SIFT: A More Distinctive Representation for Local Image Descriptors
PCA-SIFT: A More Distinctive Representation for Local Image DescriptorsPCA-SIFT: A More Distinctive Representation for Local Image Descriptors
PCA-SIFT: A More Distinctive Representation for Local Image Descriptors
 
20 bayes learning
20 bayes learning20 bayes learning
20 bayes learning
 
Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...
Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...
Dimension Reduction And Visualization Of Large High Dimensional Data Via Inte...
 

Viewers also liked

Search Engine Marketing Lecture
Search Engine Marketing LectureSearch Engine Marketing Lecture
Search Engine Marketing LectureIgor ter Halle
 
Search Engines Presentation
Search Engines PresentationSearch Engines Presentation
Search Engines PresentationJSCHO9
 
Search Engine Powerpoint
Search Engine PowerpointSearch Engine Powerpoint
Search Engine Powerpoint201014161
 
Introduction to SEO
Introduction to SEOIntroduction to SEO
Introduction to SEORand Fishkin
 

Viewers also liked (6)

Search Engine Marketing Lecture
Search Engine Marketing LectureSearch Engine Marketing Lecture
Search Engine Marketing Lecture
 
Search Engines Presentation
Search Engines PresentationSearch Engines Presentation
Search Engines Presentation
 
Search engines
Search enginesSearch engines
Search engines
 
Search Engine Demystified
Search Engine DemystifiedSearch Engine Demystified
Search Engine Demystified
 
Search Engine Powerpoint
Search Engine PowerpointSearch Engine Powerpoint
Search Engine Powerpoint
 
Introduction to SEO
Introduction to SEOIntroduction to SEO
Introduction to SEO
 

Similar to search engine for images

A COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVAL
A COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVALA COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVAL
A COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVALcscpconf
 
Image Reconstruction Using Sparse Approximation
Image Reconstruction Using Sparse ApproximationImage Reconstruction Using Sparse Approximation
Image Reconstruction Using Sparse ApproximationChristopher Neighbor
 
Research Inventy : International Journal of Engineering and Science
Research Inventy : International Journal of Engineering and ScienceResearch Inventy : International Journal of Engineering and Science
Research Inventy : International Journal of Engineering and Scienceinventy
 
Query Image Searching With Integrated Textual and Visual Relevance Feedback f...
Query Image Searching With Integrated Textual and Visual Relevance Feedback f...Query Image Searching With Integrated Textual and Visual Relevance Feedback f...
Query Image Searching With Integrated Textual and Visual Relevance Feedback f...IJERA Editor
 
EFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVAL
EFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVALEFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVAL
EFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVALsipij
 
Content based image retrieval based on shape with texture features
Content based image retrieval based on shape with texture featuresContent based image retrieval based on shape with texture features
Content based image retrieval based on shape with texture featuresAlexander Decker
 
IRJET- 3D Vision System using Calibrated Stereo Camera
IRJET- 3D Vision System using Calibrated Stereo CameraIRJET- 3D Vision System using Calibrated Stereo Camera
IRJET- 3D Vision System using Calibrated Stereo CameraIRJET Journal
 
Image Processing
Image ProcessingImage Processing
Image ProcessingTuyen Pham
 
3D Reconstruction from Multiple uncalibrated 2D Images of an Object
3D Reconstruction from Multiple uncalibrated 2D Images of an Object3D Reconstruction from Multiple uncalibrated 2D Images of an Object
3D Reconstruction from Multiple uncalibrated 2D Images of an ObjectAnkur Tyagi
 
Image recogonization
Image recogonizationImage recogonization
Image recogonizationSANTOSH RATH
 
Cross domain sentiment classification via spectral feature alignment
Cross domain sentiment classification via spectral feature alignmentCross domain sentiment classification via spectral feature alignment
Cross domain sentiment classification via spectral feature alignmentlau
 
An Assessment of Image Matching Algorithms in Depth Estimation
An Assessment of Image Matching Algorithms in Depth EstimationAn Assessment of Image Matching Algorithms in Depth Estimation
An Assessment of Image Matching Algorithms in Depth EstimationCSCJournals
 
BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...
BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...
BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...ijscmcj
 

Similar to search engine for images (20)

A COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVAL
A COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVALA COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVAL
A COMPARATIVE ANALYSIS OF RETRIEVAL TECHNIQUES IN CONTENT BASED IMAGE RETRIEVAL
 
Image Reconstruction Using Sparse Approximation
Image Reconstruction Using Sparse ApproximationImage Reconstruction Using Sparse Approximation
Image Reconstruction Using Sparse Approximation
 
Pca analysis
Pca analysisPca analysis
Pca analysis
 
Research Inventy : International Journal of Engineering and Science
Research Inventy : International Journal of Engineering and ScienceResearch Inventy : International Journal of Engineering and Science
Research Inventy : International Journal of Engineering and Science
 
Query Image Searching With Integrated Textual and Visual Relevance Feedback f...
Query Image Searching With Integrated Textual and Visual Relevance Feedback f...Query Image Searching With Integrated Textual and Visual Relevance Feedback f...
Query Image Searching With Integrated Textual and Visual Relevance Feedback f...
 
EFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVAL
EFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVALEFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVAL
EFFICIENT IMAGE RETRIEVAL USING REGION BASED IMAGE RETRIEVAL
 
Content based image retrieval based on shape with texture features
Content based image retrieval based on shape with texture featuresContent based image retrieval based on shape with texture features
Content based image retrieval based on shape with texture features
 
IRJET- 3D Vision System using Calibrated Stereo Camera
IRJET- 3D Vision System using Calibrated Stereo CameraIRJET- 3D Vision System using Calibrated Stereo Camera
IRJET- 3D Vision System using Calibrated Stereo Camera
 
Image Processing
Image ProcessingImage Processing
Image Processing
 
IMAGE RETRIEVAL USING QUADRATIC DISTANCE BASED ON COLOR FEATURE AND PYRAMID S...
IMAGE RETRIEVAL USING QUADRATIC DISTANCE BASED ON COLOR FEATURE AND PYRAMID S...IMAGE RETRIEVAL USING QUADRATIC DISTANCE BASED ON COLOR FEATURE AND PYRAMID S...
IMAGE RETRIEVAL USING QUADRATIC DISTANCE BASED ON COLOR FEATURE AND PYRAMID S...
 
mini prjt
mini prjtmini prjt
mini prjt
 
3D Reconstruction from Multiple uncalibrated 2D Images of an Object
3D Reconstruction from Multiple uncalibrated 2D Images of an Object3D Reconstruction from Multiple uncalibrated 2D Images of an Object
3D Reconstruction from Multiple uncalibrated 2D Images of an Object
 
D010332630
D010332630D010332630
D010332630
 
Image recogonization
Image recogonizationImage recogonization
Image recogonization
 
Cross domain sentiment classification via spectral feature alignment
Cross domain sentiment classification via spectral feature alignmentCross domain sentiment classification via spectral feature alignment
Cross domain sentiment classification via spectral feature alignment
 
Sentence generation
Sentence generationSentence generation
Sentence generation
 
FULL PAPER.PDF
FULL PAPER.PDFFULL PAPER.PDF
FULL PAPER.PDF
 
IJET-V2I6P17
IJET-V2I6P17IJET-V2I6P17
IJET-V2I6P17
 
An Assessment of Image Matching Algorithms in Depth Estimation
An Assessment of Image Matching Algorithms in Depth EstimationAn Assessment of Image Matching Algorithms in Depth Estimation
An Assessment of Image Matching Algorithms in Depth Estimation
 
BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...
BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...
BEHAVIOR STUDY OF ENTROPY IN A DIGITAL IMAGE THROUGH AN ITERATIVE ALGORITHM O...
 

search engine for images

  • 1. Search Engine for Images We are given a large database comprising several categories of images like: people, landscape, flowers, buses, trains, food items, animals etc. There are numerous pictures in each category.
  • 2. Our Goal A user can pick any image (not necessarily inside the database) and our program (the search engine) should be able to pull out pictures from the database that are similar to the query image.
  • 4. Our intention is to produce results as close to human perception as possible. In other words, our search engine automatically categorizes the pictures.
  • 5. Mathematical Tools Needed The Wavelet Transform. A considerable amount of Linear algebra (particularly Inner product spaces) has to be covered before we are in a position to describe the wavelet transform. The K-means clustering algorithm. This is a powerful statistical tool.
  • 6. I.T. Tools Needed We chose MATLAB 7.0 for the following reasons: It handles matrices in a convenient manner It supports a powerful programming language It can handle images in almost every format
  • 7. The Method The core of this search engine is the Integrated Region Matching Scheme described by Wang, Li, and Wiederhold in the paper: SIMPLIcity: Semantics-Sensitive Integrated Matching for Picture Libraries, IEEE, 2001. We now briefly outline the method.
  • 8. A brief outline of the method Partition a picture into 4x4 blocks and extract feature vectors for each block. Group the feature vectors into a number of regions (using K-means). Compute the signature of the image by combining the features of all its regions. Save this signature to a database. Repeat the process for every picture in the database.
  • 9. A brief outline of the method The feature vectors contain information regarding colour, texture and shape. The query image given by the user also undergoes the same signature extracting process. Finally distances between the query signature and each image signature are calculated and sorted to give the closest matches.
  • 10. In the slides that follow we will make the above ideas precise.
  • 11. What is a Feature vector? It is simply a list of 6 numbers calculated for each 4x4 block. The first three numbers are averages of the Red, Green and Blue components. The next three numbers are the root mean square of wavelet coefficients arising by applying the wavelet transform once to rows and columns of the block.
  • 12. Meaning of these components The first three numbers hold information about the colour. The next three numbers hold information about the texture.
  • 13. Image Segmentation This is what you get when K means is applied to the set of feature vectors for various values of K. All these pictures (except the original!) were created in MATLAB.
  • 14. 7 regions 4 regions Original 3 regions 2 regions
  • 15. Remark Figuring out an appropriate value of K for any image still remains an open problem in computer vision. But Wang, Li and Weiderhold propose a scheme that calculates a K that works quite well for this experiment.
  • 16. Integrated Region Matching This method combines the properties of all the regions in a picture to measure the overall similarity between images. The payoff is that the scheme provides robustness against poor segmentation.
  • 17. Integrated Region Matching Defining a similarity measure is nothing but defining a distance between sets of points (feature vectors) in a higher dimensional space. The idea of distance must be carefully chosen so that it is consistent with a person’s idea of “closeness” of two images.
  • 18. Integrated Region Matching Defining a similarity measure is nothing but defining a distance between sets of points (feature vectors) in a higher dimensional space. The idea of distance must be carefully chosen so that it is consistent with a person’s idea of “closeness” of two images.
  • 19. Integrated Region Matching Suppose two images A and B are represented by region sets A=r1,r2,…,rm and B=r′1,r′2,…,r′n  
  • 20. Integrated Region Matching Denote by dij the distance between ri and r′j. Wang et. al. give a simple prescription for this that they found experimentally. The distance between the images is defined as the weighted sum of region-to-region matches: dA,B=i,jsijdij  
  • 21. Integrated Region Matching The weights sij are the elements of the significance matrix s11⋯s1n⋮⋱⋮sm1⋯smn Wang et.al. have devised an algorithm to calculate the significance matrix.  
  • 22. Meaning of the weights The weights sij capture the importance of match. For instance if a region consists of a body of an animal and is matched to various regions in the other picture then body-to-body match will be given more weight than a body-to-background or a body-to-tree match.  
  • 23. Distance between Regions Two regions r and r′ comprise of several feature vectors. We represent these two regions by single feature vectors f and f′ constructed as follows: The ith component is the average of the ithcomponent of all FVs in that region. Now the distance between r and r′ is given by dr,r′=i=06fi−f′i2  
  • 24. Experimental Results We coded up this strategy in MATLAB 7.0 and ran the program on a database of 1000 images. Some outputs are shown in the next few slides. The Top left image is the query image.
  • 25.
  • 26.
  • 27.
  • 28.
  • 29. Final Remarks One group (Eagle 2) has some original contributions to make: They had come up with their own criteria of calculating the appropriate number of regions for an image. They even implemented their own K-means routine instead of relying on MATLAB’s built-in function.
  • 30. Final Remarks One of our IAYM student ShubhankarBiswas, NIT, Durgapur has presented this project at the Seminar on Applications of Computer and Embedded Technology organized by the Variable Energy Cyclotron Centre (VECC), Kolkata, in Oct. 2009 (2 months after IAYM 2009). His paper can be viewed at http://www.vecc.gov.in/~sacet09/index_files/Page479.htm He has acknowledged the MSF for all the guidance.