in a Fingerprint Digital Library Sung Hee Park 1 Jonathan P Leidig 1 Lin Tzy Li 134 Edward A Fox 1 Nathan J Short 2 Kevin E Hoyle 2 A Lynn Abbott 2 and Michael S Hsiao ID: 649625
Download Presentation The PPT/PDF document "Experiment and Analysis Services" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Experiment and Analysis Services in a Fingerprint Digital Library
Sung Hee Park1, Jonathan P. Leidig1, Lin Tzy Li1;3;4, Edward A. Fox1, Nathan J. Short2, Kevin E. Hoyle2, A. Lynn Abbott2, and Michael S. Hsiao21 Digital Library Research Laboratory, Virginia Tech, USA 2 Department of Electrical and Computer Engineering, Virginia Tech, USA3 Institute of Computing, University of Campinas, Brazil4 CPqD Foundation, Campinas, Brazil
TPDL: Sept 25-29, 2011, Berlin, Germany
Network Dynamics andSimulation Science Laboratory
Digital Library
Research
Laboratory (DLRL) @ Virginia
TechSlide2
ContentsIntroduction
Fingerprint Image CollectionsAlgorithms, Analyses, and Experiments ServicesFramework and PrototypeRelated WorkConclusion & Future WorkNetwork Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide3
IntroductionLack of
a fingerprint digital libraryFocus:human expert training: DOJ, FBIthe developing, testing, and training of fingerprint identification algorithms: VT, CampinasFingerprint DL services managecollectionsimage processing and matching algorithmsexperiment resultsexperiment analysesThe goal of this workend-to-end image-based experimentation and analysis services, framework, and implementation
Network Dynamics andSimulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia
TechSlide4
Experimentation Workflow
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide5
Fingerprint Image Collections
Fingerprint featuresMinutiaeRidgesClassificationsHumidityPressureDistortionSkin distortionRollingAnalysis challengesRidges mergedPressured impressionsHumidity on fingertipsPartial printsSimultaneous prints
Network Dynamics andSimulation Science LaboratoryDigital Library Research
Laboratory (DLRL) @ Virginia TechSlide6
Fingerprint Minutiae Features
TerminationBifurcation
RidgeSlide7
Ridge Tracing Classifications
Proper
DryWet
Network Dynamics andSimulation Science Laboratory
Digital Library
Research
Laboratory (DLRL) @ Virginia
TechSlide8
Physical Distortions
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide9
Rotation and Displacement Distortions
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide10
Analysis and Experiment Services in DL Framework
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide11
Basic Notation – 5S Formalisms
Term DefinitionTermDefinitionDOi;DOjdigital objects
i, j CVVertexCa collection Coll
Stmi ij.DomColla set of collections
ij
.Dom
V
Streams
stm
j
a stream
S
3
Streams Structures Spaces
st
j
a structure
tfr
S
3
Spaces
V
Streams
(N
N)
sp
j
a space
j
St
2
a set of functions
Network Dynamics and
Simulation Science Laboratory
Digital Library
Research
Laboratory (DLRL) @ Virginia
TechSlide12
Distortion Generation & Image Processing
FunctionGenerate modified images based on a distortion function based on:streams,structures, or structured streams as defined in the 5S frameworkInputa function f and a digital object (DO) doiProducta distorted version of the DO dojPre-condition and post-condition C Coll :
doi C and C Coll : doj CDefinition f : doi doj , given a digital object doi
Network Dynamics andSimulation Science Laboratory
Digital Library
Research
Laboratory (DLRL) @ Virginia
TechSlide13
Function
identify the locations and quality of major featurese.g., ridge bifurcation and terminationInput stmi Productstj ; ijPre-condition and post-condition stmi Streams and stj Structs;
ij St2; stmiij.Dom; stj.V ij.Dom, respectivelyDefinition given a digital object (stmi) produce a descriptor from the object (stj ; ij)
that represents the digital objectNetwork Dynamics andSimulation Science Laboratory
Digital Library
Research
Laboratory (DLRL) @ Virginia
Tech
Ridge Tracing & Minutiae ExtractionSlide14
Matching Algorithms & Searching
Functionidentify matches between two images as groups of minutiaeuse 3, 6, or 9-point triangles of high-quality minutiae locations less susceptible to distortionsreduce the effects of small distortions on the identification of minutiae location and qualityInputtwo images, doi; dojProductsimilarity score k based on minutiae matchesDefinition binary operation service f(doi;
doj) = k; kR, unary services (e.g., rating and measuring)f(doi) = k; k R, where a real number k is a similarity scoreNetwork Dynamics and
Simulation Science LaboratoryDigital Library Research
Laboratory (DLRL) @ Virginia
Tech
MatchSlide15
Service Specific Evaluating (Sufficiency)
Functiongiven an image, determine if there is sufficient data for a matchInputdoi Outputdoi;wiPre-condition C Coll : doi CPost-conditionwi [a; b] RDefinitiongiven a digital object
an evaluating service produces an evaluation (i.e., a real number) for itNetwork Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia Tech
49,234 / 51,294Slide16
Visualizing & Plotting
Functionprojection of information into measurable spacescharts, histograms, plots, or meshesvisualization techniques: analyze the appearance and disappearance of minutiae over distortion degreesInput a collection C and a transformation kOutputa space jPre-conditions and post-conditionsC Coll and tfr k(C) = spj MetricDefinition given a collection C produce visualizations in a space j
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide17
Example DL Experiment ScenariosMatching score accuracy experiment
How are minutiae relocated after distortions?Minutiae count and reliabilityAre minutiae still identifiable after distortions?How confidently can minutiae be matched after distortions?Minutiae plotting on fingerprintWhat can we learn from minutiae analysis?Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide18
Matching Score Accuracy Experiment
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide19
Minutia Count Experiment
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide20
Minutiae Reliability Experiment
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide21
Minutiae Plotting on a Fingerprint
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide22
Experimentation, Workflow, and Analysis Framework
Image-based experimentation stepsUser selects a collection of images, algorithms, and inputsAlgorithm-specific analysis scripts identify and extract the phenomenon being tested from the algorithm outputExperimentation workflowExecute each algorithm with a specific collectionVisualization services display the results based on distortion parametersFramework consists of building workflows or compositionsCollections, algorithms, and analysesNetwork Dynamics and
Simulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia TechSlide23
Prototype Overview
Image-based DL servicesManage a real and distorted image collectionAutomated generation of distorted images from real fingerprintsSelect and execute image-based algorithmsMatch automated analysesPrototype and web-interfaceOnline collection of original and distorted images System for selecting and composing service workflowsGoogle chart API presents the results of completed analysis tasksImages: 137,785 printsFVC 2000/02: 3520, 3520SD27: 516 Self-collected: 629Distorted: 129,600 (<1 sec generation)
Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide24
Prototype Training
A web-interfaceBrowse the image collection, image information, distortion parameters used to generate specific images, extracted minutiae, and ridge informationSuccessful minutia extraction visualizationsHumidityx-translationsy-translationsRotations Skin plasticityNetwork Dynamics and
Simulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia TechSlide25
Related Work – Existing Fingerprint Databases
FBI's Integrated Automated Fingerprint Identification System (IAFIS) Large fingerprint management systemTens of millions of imagesSearch capabilities against both latent and ten printsDigitized imagesLacks:training expertsexperiment settingdistortingplottingvisualizingThe Universal Latent Workstation (ULW) First latent workstation Supports interoperabilityShares latent identification services with local and state authorities, and with the FBI IAFIS, all with a single encoding
Network Dynamics andSimulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia
TechSlide26
Related Work – Fingerprint ExperimentationExperiment Database & Collaboration Framework
Penatti et al. [9] proposed an experiment management tool - Evaevaluates descriptors in content-base image retrievalprovides image descriptors image management runs comparative experimentsstimulated the development of our holistic DL experiment frameworkPrevious work also supported scientific communities in a web-based integration framework [10]Workflow systems: Kepler, Pegasus, Traverna, TrianaSimulation system models and analysesNetwork Dynamics and
Simulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia TechSlide27
Related Work – Fingerprint Analysis
The Analysis, Comparison, Evaluation and Verification (ACE-V) Scientific Working Group on Friction Ridge Analysis, Study and Technology (SWGFAST) groups Oliveira et al. [8] Novel tools for reconnecting broken ridges in fingerprint imagesHuang et al. [1]Singular point detectionKozievitch et al. [4] Compound object (CO) scheme based on the 5S framework to integrate four different very-large fingerprint digital librariesAllows uniform use in an integrated DL Our work: DL framework design from a services perspectiveDelivers experimentation and analytical results Integrates related services designed by different researchersNetwork Dynamics and
Simulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia TechSlide28
Conclusion & Future Work
ContributionDL supports collaborative research for DOJ/FBI trainers and researchers Servicesgenerating distorted image datasetstesting different algorithms (e.g., for minutia detection and matching)managing and work-flowing scientific research datasets, algorithms, and analysis resultsridge tracing: improve poor images, sharpen, predict distortion events based on profile, train existing algorithms and people, predict failuresStatus & Future WorkAlgorithm development and analysisIncorporate (training and development) algorithms from other types of fingerprint DLsExperiment e.g., Identify the distortion chain between two imagesTeach the effect of distortions on minutiae points
Other ApplicationsAstronomy and geo-location identification image processingUseful for cross-domain generalizationNetwork Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide29
Jonathan Leidig - leidig@vt.edu
Q & ANetwork Dynamics and
Simulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia TechSlide30
Analysis and Experiment Services
Fingerprint-specific servicesAnalysis and experiment settingDistortion generation & image processingMinutiae extraction & ridge tracingMatching & searchingEvaluatingVisualizing & plottingNetwork Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide31
Analysis and Experiment Setting
Algorithms in experiments require an algorithm-specific descriptionDistortion generation algorithmMinutiae extraction algorithm Ridge tracing algorithm Matching algorithm Network Dynamics andSimulation Science Laboratory
Digital Library Research Laboratory (DLRL) @ Virginia TechSlide32
Example WorkflowMinutiae extraction algorithm
# of minutiae located by distortion parametersThe assigned quality score (0.0 to 1.0) for each minutiae Executing this algorithm On the entire set of distorted images From a base imageWith respect to distortion parametersStatistical significance testIdentify factors hindering the identification of minutiaePre-requisiteThe distortion generation algorithm prior to forming a workflow involving algorithmic executions and subsequent analysisNetwork Dynamics and
Simulation Science LaboratoryDigital Library Research Laboratory (DLRL) @ Virginia Tech