CS597D: Geometric Analysis of 3D Models Thomas Funkhouser Princeton University CS597D, Fall 2003...

CS597D:Geometric Analysis

of 3D Models

CS597D:Geometric Analysis

of 3D Models

Thomas Funkhouser

Princeton University

CS597D, Fall 2003

Thomas Funkhouser


CS597D, Fall 2003

Introduction

On-line multimedia data is changing the way we get and use information

Call me Ishmael. Some years ago -- never mind how long precisely -- having little or no money in my purse, and nothing particular to interest me on shore, I thought I would sail about a little and see the watery part of the world. It is a way I have of driving off the spleen, and regulating the circulation. Whenever I find myself growing grim about the mouth; whenever it is a damp, drizzly November in my soul; whenever I find myself involuntarily pausing before coffin warehouses, and bringing up the rear of every funeral I meet; and especially whenever my hypos get such an upper hand of me, …

2D Images

Text

Audio

What about 3D data?

Image courtesy ofwww.dreamhorse.com

Introduction

3D data is becoming more commonly available

Someday 3D models will be as common as images are today

Someday 3D models will be as common as images are today

Cheap Scanners World Wide Web3D CafeCyberware

Fast Graphics Cards

ATI

Images courtesy ofCyberware, ATI, & 3Dcafe

Motivation

When 3D data is ubiquitous, there will be a shift in research focus

Future research will ask:“How do we find 3D data?”

Future research will ask:“How do we find 3D data?”

Utah VW Bug Utah Teapot Stanford Bunny

Images courtesy ofStanford & Utah

Previous research has asked:“How do we acquire 3D data?”Previous research has asked:“How do we acquire 3D data?”

Introduction

3D data acquired via the Web will often be void ofstructural and semantic information

Images courtesy ofDe Espona & Utah

Utah VW Bug

Analysis algorithms also are needed to create “useful” 3D models from “raw” 3D data

Analysis algorithms also are needed to create “useful” 3D models from “raw” 3D data

Introduction

Research in retrieval & analysis 3D data is warranted as it has been for other media types

ObjectRecognition

ObjectRetrieval

ObjectClassification

ObjectSynthesis

MatchingObject

SimilarObjects

MatchingClass

NovelObjects

ShapeIndex

ShapeDescriptor

ShapeAnalysis

IndexConstruction

ShapeAnalysis

Clustering &Learning

ClassSpecification

Databaseof

3D Models

GeometricQuery

Introduction

Which is harder to analyze?

2D Image3D Model

Images courtesy ofGeorgia Tech and

www.dreamhorse.com

Lecture Outline

Introduction

Problems

Applications

Course overview• Lectures• Coursework• Resources

Wrap-up

Shape Analysis Problems

Examples:• Feature detection• Segmentation• Labeling• Registration• Matching• Recognition• Classification• Clustering• Retrieval


Examples:Feature detection• Segmentation• Labeling• Registration• Matching• Retrieval• Recognition• Classification• Clustering

“How can we find significant geometric features robustly?”“How can we find significant geometric features robustly?”

Images courtesy of Bill Regli,

Drexel University


Examples:• Feature detectionSegmentation• Labeling• Registration• Matching• Retrieval• Recognition• Classification• Clustering

“How can we decompose a 3D model into its parts?”“How can we decompose a 3D model into its parts?”

Images courtesy of Ayellet Tal, Technion &



Examples:• Feature detection• SegmentationLabeling• Registration• Matching• Retrieval• Recognition• Classification• Clustering

“How can we decompose a 3D model into its parts?”“How can we decompose a 3D model into its parts?”

Images courtesy of Ayellet Tal, Technion &


Handle

Cup


Examples:• Feature detection• Segmentation• LabelingRegistration• Matching• Retrieval• Recognition• Classification• Clustering

“How can we align features of 3D models?”“How can we align features of 3D models?”

Images courtesy of Emil Praun


Examples:• Feature detection• Segmentation• Labeling• RegistrationMatching• Retrieval• Recognition• Classification• Clustering

“How can we compute a measure of geometric similarity?”“How can we compute a measure of geometric similarity?”

Image courtesy of Ilya Vakser, GRAMM


Examples:• Feature detection• Segmentation• Labeling• Registration• MatchingRetrieval• Recognition• Classification• Clustering

“How can we find 3D models best matching a query?”“How can we find 3D models best matching a query?”

1)

2)

3)

4)

Query

Ranked Matches


Examples:• Feature detection• Segmentation• Labeling• Registration• Matching• RetrievalRecognition• Classification• Clustering

“How can we find a given 3D model in a large database?”“How can we find a given 3D model in a large database?”

Images courtesy of Florida State Univ.


Examples:• Feature detection• Segmentation• Labeling• Registration• Matching• Retrieval• RecognitionClassification• Clustering

“How can we determine the class of a 3D model?”“How can we determine the class of a 3D model?”

Images courtesy of Darpa E3D Project

Query

Classes


Examples:• Feature detection• Segmentation• Labeling• Registration• Matching• Retrieval• Recognition• ClassificationClustering

“How can we learn classes of 3D models automatically?”“How can we learn classes of 3D models automatically?”

Images courtesy of Viewpoint

Lecture Outline

Introduction

Problems

Applications


Wrap-up

Shape Analysis Applications

Examples:• Virtual worlds• Animation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bio• Medicine• Forensics• Art


Examples:Virtual worlds• Animation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bio• Medicine• Forensics• Art

vp41620.wrl


Examples:• Virtual worldsAnimation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bio• Medicine• Forensics• Art

Image courtesy of Ayellet Tal, Technion &



Examples:• Virtual worldsAnimation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bio• Medicine• Forensics• Art

Movie courtesy of Ayellet Tal, Technion &



Examples:• Virtual worlds• AnimationMechanical CAD• Chemistry• Military• Paleontology• Molecular bio• Medicine• Forensics• Art

Images courtesy of Bill Regli,

Drexel University


Examples:• Virtual worlds• Animation• Mechanical CADChemistry• Military• Paleontology• Molecular bio• Medicine• Forensics• Art

Morphine


Examples:• Virtual worlds• Animation• Mechanical CAD• ChemistryMilitary• Paleontology• Molecular bio• Medicine• Forensics• Art

Images courtesy of Darpa E3D Project


Examples:• Virtual worlds• Animation• Mechanical CAD• Chemistry• MilitaryPaleontology• Molecular bio• Medicine• Forensics• Art

Images courtesy of Delson & Freiss


Examples:• Virtual worlds• Animation• Mechanical CAD• Chemistry• Military• PaleontologyMolecular bio• Medicine• Forensics• Art

Image courtesy of Ilya Vakser, GRAMM


Examples:• Virtual worlds• Animation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bioMedicine• Forensics• Art

Image courtesy of Polina Golland, MIT

Hippocampus-amygdala study in schizophrenia


Examples:• Virtual worlds• Animation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bio• MedicineForensics• Art

Images courtesy of Boeing


Examples:• Virtual worlds• Animation• Mechanical CAD• Chemistry• Military• Paleontology• Molecular bio• Medicine• ForensicsArt

Images courtesy of Stanford University

Lecture Outline

Introduction

Problems

Applications


Wrap-up

Lectures

Topics:• Methods (80%)• Applications (20%)

Speakers:• Professors• Students• Guests

Coursework

In class:• Present papers• Lead discussions

Out of class:• Two course projects (~6 weeks each)• Proposal talks, written reports, presentations• Any topic(s) related to course

Course Projects

Sample topics:• New representations• New algorithms• Compare methods• Use methods• Other attributes• New applications• Non-CS applications

Course Projects

Sample topics:New representations• New algorithms• Compare methods• Use methods• Other attributes• New applications• Non-CS applications Reflective symmetry descriptors

Images courtesy of Misha Kazhdan

Course Projects

Sample topics:• New representationsNew algorithms• Compare methods• Use methods• Other attributes• New applications• Non-CS applications

Images courtesy of Katz & Tal

Hierarchical Mesh Decomposition using Fuzzy Clustering and Cuts

[Katz & Tal, 2003]

Course Projects

Sample topics:• New representations• New algorithmsCompare methods• Use methods• Other attributes• New applications• Non-CS applications

Harmonic Descriptor

Spin Image

Shape Context

Images courtesy of Kazhdan, Johnson, & Belongie

Course Projects

Sample topics:• New representations• New algorithms• Compare methodsUse methods• Other attributes• New applications• Non-CS applications

Reflective symmetry descriptor

Mesh simplification

Images courtesy of Hoppe & Kazhdan

Course Projects

Sample topics:• New representations• New algorithms• Compare methods• Use methodsOther attributes• New applications• Non-CS applications

Text

Shape

Appearance

Course Projects

Sample topics:• New representations• New algorithms• Compare methods• Use methods• Other attributesNew applications• Non-CS applications

Modeling by Example

Course Projects

Sample topics:• New representations• New algorithms• Compare methods• Use methods• Other attributes• New applicationsNon-CS applications

Mechanical CAD1

Paleontology

Chemistry

MolecularBiology

Ilya Vakser (GRAMM)

National Design Repository

Resources

Data sets• Princeton shape benchmark• Protein data bank• CAD databases• CAT and MRI scans• Range scans

Software• Ringlet• Other useful tools

Papers

Wrap Up

Students’ to do list:• Sign up for in-class presentations• Start thinking about project topics

CS597D: Geometric Analysis of 3D Models Thomas Funkhouser Princeton University CS597D, Fall 2003...

Documents

Transcript of CS597D: Geometric Analysis of 3D Models Thomas Funkhouser Princeton University CS597D, Fall 2003...