SlideShare a Scribd company logo
MET’S School of Engineering
Department of Computer Science and Engineering
Academic Year 2015-2016
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 1
 
SKETCH TO PHOTO MATCHING SYSTEM
Submitted by
Chippy Thomas (GWAOCCS004)
S4-M.Tech CSE (2014-16 Batch)
19-03-2016
May 24, 2016 2
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Introduction
System that match sketch with mugshot
For the identification of suspects
Use when eyewitness are only evidence
Sketch/composites
-hand-drawn
-software generated
-surveillance composites
•mugshot
19-03-2016 2
May 24, 2016 3
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Introduction – contd…
So matching sketch and photo
Direct matching is not possible
Ie..We don’t match them pixel by pixel
Match by finding high-level features from both the sketch
and the photo
Features such as the structural distribution and the shape
of the eyes, nose and chin etc.
19-03-2016 3
May 24, 2016 4
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Introduction – contd…
19-03-2016 3
May 24, 2016 5
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Hand-drawn composite
19-03-2016 4
•Drawn by forensic artist
•Based on description provided by one or more witness
•Artist need proper training
May 24, 2016 6
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Software-generated composites
19-03-2016 5
•Drawn by operator using software
•Most used software's are identikit and FACES
•Operator need training
May 24, 2016 7
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Surveillance composites
19-03-2016 6
•Drawn by forensic artist
•Based on poor quality surveillance images
May 24, 2016 8
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Example of Verbal Description
The man was about 5ft. 6in. in
height, and 34 or 35 years of
age, with dark complexion and
dark mustache turned up at the
ends . . . He had a heavy
mustache curled up, and dark
eyes and bushy eyebrows. He
had no side whiskers, and his
chin was clean shaven. He
looked like a foreigner."
19-03-2016 7
Objective
SKETCH TO PHOTO MATCHING SYSTEM
Develop a system that can automatically match sketch with
mugshots
System that manage every types of sketch such as
-forensic sketches
-composite sketches
-surveillance sketches
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 8
• Law enforcement agencies disseminate a sketch to media
outlets
• Hope that someone will recognise the suspect
• And provide information leading to an arrest.
19-02-2016
2
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Existing system
19-03-2016 9
 
May 24, 2016 11
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 10
Existing system – contd…
Problem description
SKETCH TO PHOTO MATCHING SYSTEM
•Slow and tedious process
•Fails to leverage resources available to law enforcement agencies
(mugshot databases)
•So there is a need for a method that can automatically match facial
composites to large police mugshot databases.
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 11
• Automatically match facial composites against mugshots
• For this 2 algorithms are used
• Holistic representation (developed by Klare and Jain)
• Component representation (developed by Han et al.)
19-02-2016
2
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Proposed system
19-03-2016 12
May 24, 2016 14
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 13
Proposed system – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Working
•Sketches that is given as input to the system
•From sketch features are extracted based on 2 algorithms
•There will be a database to hold the features of large mugshot
•By applying 2 algorithms it matches with the mugshot
•2 algorithms produce matching scores
•Combining matching scores of 2 algorithm
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 14
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
Figure
19-03-2016 15
Holistic algorithm
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 16
•First step is detection of eye
•Alignment and CSDN
•Feature extraction
-SIFT
-MLBP
•LDA- learn optimal subspaces & projecting into subspaces
•PCA
19-03-2016
18
SKETCH
EYE
DETECTION
ALIGNMENT
AND CSDN
PATCHWISE
SEGMENTATION
DENSE GRID
OF
KEYPOINTS
MLBP
SIFT
LDA Patch-wise
concatenation
& PCA
Sketch
Feature
Vector
Mugshot
Feature
Vector
Repeated for all patches
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE
L2
norm
19-03-2016 17
Holistic Algorithm – contd…
Holistic algorithm – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 18
•The same steps are performed for sketch and photo
•Sketch feature vector and mugshot feature vector
•These 2 features are comparing for similarity using L2
norm
•Finding score of matching
19-03-2016
Component-based algorithm
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19
•In holistic algorithm, features are extracted from each patch / grid
from images
•But component-based algorithm, features are extracted from each
component
•Per component, features are extracting
19-03-2016
May 24, 2016 21
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 20
Component-based algorithm – contd…
Modules
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 21
Eye Detection
•Accurate and reliable
•Eyes are protected and remain reliably unchanged for decades
•Eyes are relatively stable feature
19-03-2016
Modules – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 2219-02-2016
Alignment and CSDN
•composite is normalized to a fixed height and width
•transformed such that right and left eyes are at the same
position for every composite.
•The center-surround divisive normalization (CSDN)
•filter is then applied to the composite to compensate change
in modality between composite and mugshot.
19-03-2016
Modules – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 23
•SIFT and MLBP Feature Extraction
•SIFT features are extracted from dense grid of keypoints
•Multi-scale local binary pattern (MLBP) features are extracted
from a uniform patches across the face,
19-03-2016
Modules – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 24
LDA
•For both SIFT and MLBP features, optimal subspaces are learned
for each patch using linear discriminant analysis (LDA).
•projecting the patch-wise features into their respective subspaces
PCA
PCA is applied to the feature vector to reduce template size
19-03-2016
Modules – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 25
L2
norm
•It is used as classifier
•For finding similarity between mugshot and sketch features
Sum of Score Fusion
•Scores from 2 algorithms are fused using sum of score fusion
19-03-2016
Innovative idea
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 26
•Modifying existing systems with Enhanced Holistic method and
Enhanced Composite method.
•Replacing feature extraction methods and classifier
19-03-2016
Innovative idea
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 2719-03-2016
Innovative idea – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 28
•In the proposed scheme it is done by replacing the existing
systems algorithms such as SIFT and MLBP with SURF and
harlick algorithms.
•Also adding Neural network classifier instead of L
2
norm for
better accuracy.
19-03-2016
screenshots – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 2919-03-2016
•Input sketch
screenshots – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3019-03-2016
•Eye detection from sketch
screenshots – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3119-03-2016
•Aligned sketch
screenshots – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3219-03-2016
•Features extracted from sketch
screenshots – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3319-03-2016
•Eye detection from photo
screenshots – contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3419-03-2016
•Feature extraction from photo
Conclusion
SKETCH TO PHOTO MATCHING SYSTEM
3619-02-2016
•Facial composites are used by law enforcement agencies for the
identification of suspects involved in criminal activities
•Sketch to photo system effectively make the use of facial
composites
•Automatic
•Can be used with every type of composites
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 35
Number: 01 Guide Name: Mr. JOBI V DAS, AP/CSE 36
SKETCH TO PHOTO MATCHING SYSTEM
•For that 2 algorithms are using here
•Holistic algorithm and component based algorithm
•Match scores from 2 algorithms are combining
•Also filtering based on demographic information
•Along with the matched image some details such as name, age can
be displayed.
19-3-2016
Conclusion – contd…
References
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3819-02-2016
•Bonnen K. , B. F. Klare, and A. K. Jain, “Component-based
representation in automated face recognition,” IEEE Trans. Inf.
Forensics Security, vol. 8, no. 1, pp. 239–253, Jan. 2012.
•Best-Rowden L. , H. Han, C. Otto, B. Klare, and A. Jain,
“Unconstrained face recognition: Identifying a person of interest
from a media collection,” IEEE Trans. Inf. Forensics Security,Dec.
2014
•Han H. , B. F. Klare, K. Bonnen, and A. K. Jain, “Matching
composite sketches to face photos: A component-based approach,”
IEEE Trans. Inf. Forensics Security, vol. 8, no. 1, pp. 191–204,
Jan. 2013
19-03-2016
References – Contd…
SKETCH TO PHOTO MATCHING SYSTEM
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 39
•Scott J. Klum, Hu Han, Brendan F. Klare Anil K. Jain, ”The
FaceSketchID System: Matching Facial Composites to Mugshots”
IEEE Trans. Inf. Forensics Security,Dec. 2014 ,Vol. 9, no. 12, Dec
2014
19-03-2016
Quires
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 40
SKETCH TO PHOTO MATCHING SYSTEM
May 24, 2016
?
Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 41
SKETCH TO PHOTO MATCHING SYSTEM
May 24, 2016
Thank you

More Related Content

Viewers also liked

Rapid prototyping and sketching
Rapid prototyping and sketchingRapid prototyping and sketching
Rapid prototyping and sketching
Christian Glover Wilson
 
Portraitdrawing
PortraitdrawingPortraitdrawing
Portraitdrawingcjoyce104
 
How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...
How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...
How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...
Pauline Howard
 
The art-of-drawing-the-human-body
The art-of-drawing-the-human-bodyThe art-of-drawing-the-human-body
The art-of-drawing-the-human-bodyHoseo University
 
study drawing drawing portraits faces and figures
study drawing drawing portraits faces and figuresstudy drawing drawing portraits faces and figures
study drawing drawing portraits faces and figuresErdwin
 
Burne hogarth -_drawing_the_human_head
Burne hogarth -_drawing_the_human_headBurne hogarth -_drawing_the_human_head
Burne hogarth -_drawing_the_human_headErdwin
 

Viewers also liked (6)

Rapid prototyping and sketching
Rapid prototyping and sketchingRapid prototyping and sketching
Rapid prototyping and sketching
 
Portraitdrawing
PortraitdrawingPortraitdrawing
Portraitdrawing
 
How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...
How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...
How To Draw Face, Drawing Classes Online, Portraits For Drawing, Drawing The ...
 
The art-of-drawing-the-human-body
The art-of-drawing-the-human-bodyThe art-of-drawing-the-human-body
The art-of-drawing-the-human-body
 
study drawing drawing portraits faces and figures
study drawing drawing portraits faces and figuresstudy drawing drawing portraits faces and figures
study drawing drawing portraits faces and figures
 
Burne hogarth -_drawing_the_human_head
Burne hogarth -_drawing_the_human_headBurne hogarth -_drawing_the_human_head
Burne hogarth -_drawing_the_human_head
 

Similar to sketch to photo matching

The Search and Hyperlinking Task at MediaEval 2014
The Search and Hyperlinking Task at MediaEval 2014The Search and Hyperlinking Task at MediaEval 2014
The Search and Hyperlinking Task at MediaEval 2014
multimediaeval
 
Agent based modeling level 2 - edukite
Agent based modeling level 2 - edukiteAgent based modeling level 2 - edukite
Agent based modeling level 2 - edukite
EduKite
 
Code Dojo for Product Owners
Code Dojo for Product OwnersCode Dojo for Product Owners
Code Dojo for Product Owners
Tomasz Wykowski, CST
 
MLSEV. Anatomy of an ML Application
MLSEV. Anatomy of an ML ApplicationMLSEV. Anatomy of an ML Application
MLSEV. Anatomy of an ML Application
BigML, Inc
 
CIS 339 Entire Course NEW
CIS 339 Entire Course NEWCIS 339 Entire Course NEW
CIS 339 Entire Course NEW
shyamuopuop
 
From DevOps to MLOps: practical steps for a smooth transition
From DevOps to MLOps: practical steps for a smooth transitionFrom DevOps to MLOps: practical steps for a smooth transition
From DevOps to MLOps: practical steps for a smooth transition
Anne-Marie Tousch
 
A kanban scrum mashup to deliver fp projects
A kanban scrum mashup to deliver fp projectsA kanban scrum mashup to deliver fp projects
A kanban scrum mashup to deliver fp projects
India Scrum Enthusiasts Community
 
Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339
arnitaetsitty
 
Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339
arnitaetsitty
 
Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...
Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...
Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...
Spark Summit
 
Slides: Safeguarding Abila through Multiple Data Perspectives
Slides: Safeguarding Abila through Multiple Data PerspectivesSlides: Safeguarding Abila through Multiple Data Perspectives
Slides: Safeguarding Abila through Multiple Data Perspectives
Parang Saraf
 
IRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry Pi
IRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry PiIRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry Pi
IRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry Pi
IRJET Journal
 
Barcode Based Parking Management System
Barcode Based Parking Management SystemBarcode Based Parking Management System
Barcode Based Parking Management System
IRJET Journal
 
Face recognization technology
Face recognization technologyFace recognization technology
Face recognization technology
Satyanarayana Tammana
 
Licence Plate Recognition Using Supervised Learning and Deep Learning
Licence Plate Recognition Using Supervised Learning and Deep LearningLicence Plate Recognition Using Supervised Learning and Deep Learning
Licence Plate Recognition Using Supervised Learning and Deep Learning
IRJET Journal
 
CNMES15 - Estimation con COSMIC - Alain Abran
CNMES15 - Estimation con COSMIC - Alain AbranCNMES15 - Estimation con COSMIC - Alain Abran
CNMES15 - Estimation con COSMIC - Alain Abran
COSMIC - Common Software Measurement International Consortium
 
Tableau 2018 - Introduction to Visual analytics
Tableau 2018 - Introduction to Visual analyticsTableau 2018 - Introduction to Visual analytics
Tableau 2018 - Introduction to Visual analytics
Arun K
 
Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...
Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...
Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...
Precisely
 
Do iT : Failure - Root Cause analysis : Classical approach by C3VIVO
Do iT : Failure - Root Cause analysis : Classical approach by C3VIVODo iT : Failure - Root Cause analysis : Classical approach by C3VIVO
Do iT : Failure - Root Cause analysis : Classical approach by C3VIVO
Kiran Vairagkar
 
Ijetr011814
Ijetr011814Ijetr011814
Ijetr011814
ER Publication.org
 

Similar to sketch to photo matching (20)

The Search and Hyperlinking Task at MediaEval 2014
The Search and Hyperlinking Task at MediaEval 2014The Search and Hyperlinking Task at MediaEval 2014
The Search and Hyperlinking Task at MediaEval 2014
 
Agent based modeling level 2 - edukite
Agent based modeling level 2 - edukiteAgent based modeling level 2 - edukite
Agent based modeling level 2 - edukite
 
Code Dojo for Product Owners
Code Dojo for Product OwnersCode Dojo for Product Owners
Code Dojo for Product Owners
 
MLSEV. Anatomy of an ML Application
MLSEV. Anatomy of an ML ApplicationMLSEV. Anatomy of an ML Application
MLSEV. Anatomy of an ML Application
 
CIS 339 Entire Course NEW
CIS 339 Entire Course NEWCIS 339 Entire Course NEW
CIS 339 Entire Course NEW
 
From DevOps to MLOps: practical steps for a smooth transition
From DevOps to MLOps: practical steps for a smooth transitionFrom DevOps to MLOps: practical steps for a smooth transition
From DevOps to MLOps: practical steps for a smooth transition
 
A kanban scrum mashup to deliver fp projects
A kanban scrum mashup to deliver fp projectsA kanban scrum mashup to deliver fp projects
A kanban scrum mashup to deliver fp projects
 
Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339
 
Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339Cis 339 complete class i labs cis339
Cis 339 complete class i labs cis339
 
Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...
Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...
Building a Data Warehouse for Business Analytics using Spark SQL-(Blagoy Kalo...
 
Slides: Safeguarding Abila through Multiple Data Perspectives
Slides: Safeguarding Abila through Multiple Data PerspectivesSlides: Safeguarding Abila through Multiple Data Perspectives
Slides: Safeguarding Abila through Multiple Data Perspectives
 
IRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry Pi
IRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry PiIRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry Pi
IRJET- Face Detection and Tracking Algorithm using Open CV with Raspberry Pi
 
Barcode Based Parking Management System
Barcode Based Parking Management SystemBarcode Based Parking Management System
Barcode Based Parking Management System
 
Face recognization technology
Face recognization technologyFace recognization technology
Face recognization technology
 
Licence Plate Recognition Using Supervised Learning and Deep Learning
Licence Plate Recognition Using Supervised Learning and Deep LearningLicence Plate Recognition Using Supervised Learning and Deep Learning
Licence Plate Recognition Using Supervised Learning and Deep Learning
 
CNMES15 - Estimation con COSMIC - Alain Abran
CNMES15 - Estimation con COSMIC - Alain AbranCNMES15 - Estimation con COSMIC - Alain Abran
CNMES15 - Estimation con COSMIC - Alain Abran
 
Tableau 2018 - Introduction to Visual analytics
Tableau 2018 - Introduction to Visual analyticsTableau 2018 - Introduction to Visual analytics
Tableau 2018 - Introduction to Visual analytics
 
Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...
Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...
Machine Learning & IT Service Intelligence for the Enterprise: The Future is ...
 
Do iT : Failure - Root Cause analysis : Classical approach by C3VIVO
Do iT : Failure - Root Cause analysis : Classical approach by C3VIVODo iT : Failure - Root Cause analysis : Classical approach by C3VIVO
Do iT : Failure - Root Cause analysis : Classical approach by C3VIVO
 
Ijetr011814
Ijetr011814Ijetr011814
Ijetr011814
 

Recently uploaded

1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样
1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样
1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样
9ib5wiwt
 
Business and Corporate Case Update (2024)
Business and Corporate Case Update (2024)Business and Corporate Case Update (2024)
Business and Corporate Case Update (2024)
Wendy Couture
 
原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样
原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样
原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样
9ib5wiwt
 
EMPLOYMENT LAW AN OVERVIEW in Malawi.pptx
EMPLOYMENT LAW  AN OVERVIEW in Malawi.pptxEMPLOYMENT LAW  AN OVERVIEW in Malawi.pptx
EMPLOYMENT LAW AN OVERVIEW in Malawi.pptx
MwaiMapemba
 
Bharatiya Nagarik Suraksha Sanhita power.pptx
Bharatiya Nagarik Suraksha Sanhita power.pptxBharatiya Nagarik Suraksha Sanhita power.pptx
Bharatiya Nagarik Suraksha Sanhita power.pptx
ShivkumarIyer18
 
定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样
定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样
定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样
9ib5wiwt
 
new victimology of indonesian law. Pptx.
new victimology of indonesian law. Pptx.new victimology of indonesian law. Pptx.
new victimology of indonesian law. Pptx.
niputusriwidiasih
 
NATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptx
NATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptxNATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptx
NATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptx
anvithaav
 
Roles of a Bankruptcy Lawyer John Cavitt
Roles of a Bankruptcy Lawyer John CavittRoles of a Bankruptcy Lawyer John Cavitt
Roles of a Bankruptcy Lawyer John Cavitt
johncavitthouston
 
How to Obtain Permanent Residency in the Netherlands
How to Obtain Permanent Residency in the NetherlandsHow to Obtain Permanent Residency in the Netherlands
How to Obtain Permanent Residency in the Netherlands
BridgeWest.eu
 
Car Accident Injury Do I Have a Case....
Car Accident Injury Do I Have a Case....Car Accident Injury Do I Have a Case....
Car Accident Injury Do I Have a Case....
Knowyourright
 
XYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdf
XYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdfXYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdf
XYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdf
bhavenpr
 
Rokita Releases Soccer Stadium Legal Opinion
Rokita Releases Soccer Stadium Legal OpinionRokita Releases Soccer Stadium Legal Opinion
Rokita Releases Soccer Stadium Legal Opinion
Abdul-Hakim Shabazz
 
ADR in criminal proceeding in Bangladesh with global perspective.
ADR in criminal proceeding in Bangladesh with global perspective.ADR in criminal proceeding in Bangladesh with global perspective.
ADR in criminal proceeding in Bangladesh with global perspective.
Daffodil International University
 
Military Commissions details LtCol Thomas Jasper as Detailed Defense Counsel
Military Commissions details LtCol Thomas Jasper as Detailed Defense CounselMilitary Commissions details LtCol Thomas Jasper as Detailed Defense Counsel
Military Commissions details LtCol Thomas Jasper as Detailed Defense Counsel
Thomas (Tom) Jasper
 
Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...
Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...
Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...
Syed Muhammad Humza Hussain
 
Understanding about ITR-1 and Documentation
Understanding about ITR-1 and DocumentationUnderstanding about ITR-1 and Documentation
Understanding about ITR-1 and Documentation
CAAJAYKUMAR4
 
Donald_J_Trump_katigoritirio_stormi_daniels.pdf
Donald_J_Trump_katigoritirio_stormi_daniels.pdfDonald_J_Trump_katigoritirio_stormi_daniels.pdf
Donald_J_Trump_katigoritirio_stormi_daniels.pdf
ssuser5750e1
 
Abdul Hakim Shabazz Deposition Hearing in Federal Court
Abdul Hakim Shabazz Deposition Hearing in Federal CourtAbdul Hakim Shabazz Deposition Hearing in Federal Court
Abdul Hakim Shabazz Deposition Hearing in Federal Court
Gabe Whitley
 
ALL EYES ON RAFAH BUT WHY Explain more.pdf
ALL EYES ON RAFAH BUT WHY Explain more.pdfALL EYES ON RAFAH BUT WHY Explain more.pdf
ALL EYES ON RAFAH BUT WHY Explain more.pdf
46adnanshahzad
 

Recently uploaded (20)

1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样
1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样
1比1制作(swansea毕业证书)英国斯旺西大学毕业证学位证书托业成绩单原版一模一样
 
Business and Corporate Case Update (2024)
Business and Corporate Case Update (2024)Business and Corporate Case Update (2024)
Business and Corporate Case Update (2024)
 
原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样
原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样
原版仿制(aut毕业证书)新西兰奥克兰理工大学毕业证文凭毕业证雅思成绩单原版一模一样
 
EMPLOYMENT LAW AN OVERVIEW in Malawi.pptx
EMPLOYMENT LAW  AN OVERVIEW in Malawi.pptxEMPLOYMENT LAW  AN OVERVIEW in Malawi.pptx
EMPLOYMENT LAW AN OVERVIEW in Malawi.pptx
 
Bharatiya Nagarik Suraksha Sanhita power.pptx
Bharatiya Nagarik Suraksha Sanhita power.pptxBharatiya Nagarik Suraksha Sanhita power.pptx
Bharatiya Nagarik Suraksha Sanhita power.pptx
 
定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样
定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样
定制(nus毕业证书)新加坡国立大学毕业证学位证书实拍图原版一模一样
 
new victimology of indonesian law. Pptx.
new victimology of indonesian law. Pptx.new victimology of indonesian law. Pptx.
new victimology of indonesian law. Pptx.
 
NATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptx
NATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptxNATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptx
NATURE, ORIGIN AND DEVELOPMENT OF INTERNATIONAL LAW.pptx
 
Roles of a Bankruptcy Lawyer John Cavitt
Roles of a Bankruptcy Lawyer John CavittRoles of a Bankruptcy Lawyer John Cavitt
Roles of a Bankruptcy Lawyer John Cavitt
 
How to Obtain Permanent Residency in the Netherlands
How to Obtain Permanent Residency in the NetherlandsHow to Obtain Permanent Residency in the Netherlands
How to Obtain Permanent Residency in the Netherlands
 
Car Accident Injury Do I Have a Case....
Car Accident Injury Do I Have a Case....Car Accident Injury Do I Have a Case....
Car Accident Injury Do I Have a Case....
 
XYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdf
XYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdfXYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdf
XYZ-v.-state-of-Maharashtra-Bombay-HC-Writ-Petition-6340-2023.pdf
 
Rokita Releases Soccer Stadium Legal Opinion
Rokita Releases Soccer Stadium Legal OpinionRokita Releases Soccer Stadium Legal Opinion
Rokita Releases Soccer Stadium Legal Opinion
 
ADR in criminal proceeding in Bangladesh with global perspective.
ADR in criminal proceeding in Bangladesh with global perspective.ADR in criminal proceeding in Bangladesh with global perspective.
ADR in criminal proceeding in Bangladesh with global perspective.
 
Military Commissions details LtCol Thomas Jasper as Detailed Defense Counsel
Military Commissions details LtCol Thomas Jasper as Detailed Defense CounselMilitary Commissions details LtCol Thomas Jasper as Detailed Defense Counsel
Military Commissions details LtCol Thomas Jasper as Detailed Defense Counsel
 
Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...
Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...
Synopsis On Annual General Meeting/Extra Ordinary General Meeting With Ordina...
 
Understanding about ITR-1 and Documentation
Understanding about ITR-1 and DocumentationUnderstanding about ITR-1 and Documentation
Understanding about ITR-1 and Documentation
 
Donald_J_Trump_katigoritirio_stormi_daniels.pdf
Donald_J_Trump_katigoritirio_stormi_daniels.pdfDonald_J_Trump_katigoritirio_stormi_daniels.pdf
Donald_J_Trump_katigoritirio_stormi_daniels.pdf
 
Abdul Hakim Shabazz Deposition Hearing in Federal Court
Abdul Hakim Shabazz Deposition Hearing in Federal CourtAbdul Hakim Shabazz Deposition Hearing in Federal Court
Abdul Hakim Shabazz Deposition Hearing in Federal Court
 
ALL EYES ON RAFAH BUT WHY Explain more.pdf
ALL EYES ON RAFAH BUT WHY Explain more.pdfALL EYES ON RAFAH BUT WHY Explain more.pdf
ALL EYES ON RAFAH BUT WHY Explain more.pdf
 

sketch to photo matching

  • 1. MET’S School of Engineering Department of Computer Science and Engineering Academic Year 2015-2016 Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 1   SKETCH TO PHOTO MATCHING SYSTEM Submitted by Chippy Thomas (GWAOCCS004) S4-M.Tech CSE (2014-16 Batch) 19-03-2016
  • 2. May 24, 2016 2 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Introduction System that match sketch with mugshot For the identification of suspects Use when eyewitness are only evidence Sketch/composites -hand-drawn -software generated -surveillance composites •mugshot 19-03-2016 2
  • 3. May 24, 2016 3 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Introduction – contd… So matching sketch and photo Direct matching is not possible Ie..We don’t match them pixel by pixel Match by finding high-level features from both the sketch and the photo Features such as the structural distribution and the shape of the eyes, nose and chin etc. 19-03-2016 3
  • 4. May 24, 2016 4 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Introduction – contd… 19-03-2016 3
  • 5. May 24, 2016 5 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Hand-drawn composite 19-03-2016 4 •Drawn by forensic artist •Based on description provided by one or more witness •Artist need proper training
  • 6. May 24, 2016 6 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Software-generated composites 19-03-2016 5 •Drawn by operator using software •Most used software's are identikit and FACES •Operator need training
  • 7. May 24, 2016 7 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Surveillance composites 19-03-2016 6 •Drawn by forensic artist •Based on poor quality surveillance images
  • 8. May 24, 2016 8 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Example of Verbal Description The man was about 5ft. 6in. in height, and 34 or 35 years of age, with dark complexion and dark mustache turned up at the ends . . . He had a heavy mustache curled up, and dark eyes and bushy eyebrows. He had no side whiskers, and his chin was clean shaven. He looked like a foreigner." 19-03-2016 7
  • 9. Objective SKETCH TO PHOTO MATCHING SYSTEM Develop a system that can automatically match sketch with mugshots System that manage every types of sketch such as -forensic sketches -composite sketches -surveillance sketches Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 8
  • 10. • Law enforcement agencies disseminate a sketch to media outlets • Hope that someone will recognise the suspect • And provide information leading to an arrest. 19-02-2016 2 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Existing system 19-03-2016 9
  • 11.   May 24, 2016 11 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 10 Existing system – contd…
  • 12. Problem description SKETCH TO PHOTO MATCHING SYSTEM •Slow and tedious process •Fails to leverage resources available to law enforcement agencies (mugshot databases) •So there is a need for a method that can automatically match facial composites to large police mugshot databases. Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 11
  • 13. • Automatically match facial composites against mugshots • For this 2 algorithms are used • Holistic representation (developed by Klare and Jain) • Component representation (developed by Han et al.) 19-02-2016 2 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Proposed system 19-03-2016 12
  • 14. May 24, 2016 14 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 13 Proposed system – contd…
  • 15. SKETCH TO PHOTO MATCHING SYSTEM Working •Sketches that is given as input to the system •From sketch features are extracted based on 2 algorithms •There will be a database to hold the features of large mugshot •By applying 2 algorithms it matches with the mugshot •2 algorithms produce matching scores •Combining matching scores of 2 algorithm Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 14
  • 16. SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE Figure 19-03-2016 15
  • 17. Holistic algorithm SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 16 •First step is detection of eye •Alignment and CSDN •Feature extraction -SIFT -MLBP •LDA- learn optimal subspaces & projecting into subspaces •PCA 19-03-2016
  • 18. 18 SKETCH EYE DETECTION ALIGNMENT AND CSDN PATCHWISE SEGMENTATION DENSE GRID OF KEYPOINTS MLBP SIFT LDA Patch-wise concatenation & PCA Sketch Feature Vector Mugshot Feature Vector Repeated for all patches SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE L2 norm 19-03-2016 17 Holistic Algorithm – contd…
  • 19. Holistic algorithm – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 18 •The same steps are performed for sketch and photo •Sketch feature vector and mugshot feature vector •These 2 features are comparing for similarity using L2 norm •Finding score of matching 19-03-2016
  • 20. Component-based algorithm SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19 •In holistic algorithm, features are extracted from each patch / grid from images •But component-based algorithm, features are extracted from each component •Per component, features are extracting 19-03-2016
  • 21. May 24, 2016 21 SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 20 Component-based algorithm – contd…
  • 22. Modules SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 21 Eye Detection •Accurate and reliable •Eyes are protected and remain reliably unchanged for decades •Eyes are relatively stable feature 19-03-2016
  • 23. Modules – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 2219-02-2016 Alignment and CSDN •composite is normalized to a fixed height and width •transformed such that right and left eyes are at the same position for every composite. •The center-surround divisive normalization (CSDN) •filter is then applied to the composite to compensate change in modality between composite and mugshot. 19-03-2016
  • 24. Modules – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 23 •SIFT and MLBP Feature Extraction •SIFT features are extracted from dense grid of keypoints •Multi-scale local binary pattern (MLBP) features are extracted from a uniform patches across the face, 19-03-2016
  • 25. Modules – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 24 LDA •For both SIFT and MLBP features, optimal subspaces are learned for each patch using linear discriminant analysis (LDA). •projecting the patch-wise features into their respective subspaces PCA PCA is applied to the feature vector to reduce template size 19-03-2016
  • 26. Modules – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 25 L2 norm •It is used as classifier •For finding similarity between mugshot and sketch features Sum of Score Fusion •Scores from 2 algorithms are fused using sum of score fusion 19-03-2016
  • 27. Innovative idea SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 26 •Modifying existing systems with Enhanced Holistic method and Enhanced Composite method. •Replacing feature extraction methods and classifier 19-03-2016
  • 28. Innovative idea SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 2719-03-2016
  • 29. Innovative idea – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 28 •In the proposed scheme it is done by replacing the existing systems algorithms such as SIFT and MLBP with SURF and harlick algorithms. •Also adding Neural network classifier instead of L 2 norm for better accuracy. 19-03-2016
  • 30. screenshots – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 2919-03-2016 •Input sketch
  • 31. screenshots – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3019-03-2016 •Eye detection from sketch
  • 32. screenshots – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3119-03-2016 •Aligned sketch
  • 33. screenshots – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3219-03-2016 •Features extracted from sketch
  • 34. screenshots – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3319-03-2016 •Eye detection from photo
  • 35. screenshots – contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3419-03-2016 •Feature extraction from photo
  • 36. Conclusion SKETCH TO PHOTO MATCHING SYSTEM 3619-02-2016 •Facial composites are used by law enforcement agencies for the identification of suspects involved in criminal activities •Sketch to photo system effectively make the use of facial composites •Automatic •Can be used with every type of composites Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 19-03-2016 35
  • 37. Number: 01 Guide Name: Mr. JOBI V DAS, AP/CSE 36 SKETCH TO PHOTO MATCHING SYSTEM •For that 2 algorithms are using here •Holistic algorithm and component based algorithm •Match scores from 2 algorithms are combining •Also filtering based on demographic information •Along with the matched image some details such as name, age can be displayed. 19-3-2016 Conclusion – contd…
  • 38. References SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 3819-02-2016 •Bonnen K. , B. F. Klare, and A. K. Jain, “Component-based representation in automated face recognition,” IEEE Trans. Inf. Forensics Security, vol. 8, no. 1, pp. 239–253, Jan. 2012. •Best-Rowden L. , H. Han, C. Otto, B. Klare, and A. Jain, “Unconstrained face recognition: Identifying a person of interest from a media collection,” IEEE Trans. Inf. Forensics Security,Dec. 2014 •Han H. , B. F. Klare, K. Bonnen, and A. K. Jain, “Matching composite sketches to face photos: A component-based approach,” IEEE Trans. Inf. Forensics Security, vol. 8, no. 1, pp. 191–204, Jan. 2013 19-03-2016
  • 39. References – Contd… SKETCH TO PHOTO MATCHING SYSTEM Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 39 •Scott J. Klum, Hu Han, Brendan F. Klare Anil K. Jain, ”The FaceSketchID System: Matching Facial Composites to Mugshots” IEEE Trans. Inf. Forensics Security,Dec. 2014 ,Vol. 9, no. 12, Dec 2014 19-03-2016
  • 40. Quires Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 40 SKETCH TO PHOTO MATCHING SYSTEM May 24, 2016 ?
  • 41. Number: 04 Guide Name: Mr. JOBI V DAS, AP/CSE 41 SKETCH TO PHOTO MATCHING SYSTEM May 24, 2016 Thank you