TREES: Computer vision based tree-ring analysis and measurement system.

24
27 July, 2000 J. Engle, University of Arizona, ECE- DIAL/LTRR 1 A C A C OMPUTER OMPUTER -A -A SSISTED SSISTED T T REE REE -R -R ING ING C C HRONOLOGY HRONOLOGY C C OMPOSITION OMPOSITION S S YSTEM YSTEM System Overview System Overview B B y: James B. Engle y: James B. Engle Department of Electrical and Computer Engineering Department of Electrical and Computer Engineering at at The University of Arizona The University of Arizona

description

A C OMPUTER -A SSISTED T REE -R ING C HRONOLOGY C OMPOSITION S YSTEM System Overview B y: James B. Engle Department of Electrical and Computer Engineering at The University of Arizona. - PowerPoint PPT Presentation

Transcript of TREES: Computer vision based tree-ring analysis and measurement system.

Page 1: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July, 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 1

A CA COMPUTEROMPUTER-A-ASSISTEDSSISTED

TTREEREE-R-RINGING C CHRONOLOGYHRONOLOGY

CCOMPOSITIONOMPOSITION S SYSTEMYSTEM

System OverviewSystem Overview

BBy: James B. Engley: James B. Engle Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering

atatThe University of ArizonaThe University of Arizona

A CA COMPUTEROMPUTER-A-ASSISTEDSSISTED

TTREEREE-R-RINGING C CHRONOLOGYHRONOLOGY

CCOMPOSITIONOMPOSITION S SYSTEMYSTEM

System OverviewSystem Overview

BBy: James B. Engley: James B. Engle Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering

atatThe University of ArizonaThe University of Arizona

Page 2: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 2

A Computer-Assisted Tree-RingChronology Composition System

Table of Contents

SectionSlides1. Background - TREES program overview........................ 3-62. Introducing the CROSSDATE program........................... 7-103. Correlation Tool............................................................... 12-144. Composition Tool............................................................. 15-185. Check Tool....................................................................... 19-226. User Help......................................................................... 237. Contribution to Dendrochronology................................... 24

Reference: Engle, James B., A Computer-Assisted Tree-Ring Chronology Composition System, Master’s Thesis, University of Arizona, 2000.

Page 3: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 3

A Computer-Assisted Tree-RingChronology Composition System

TREES: Computer vision based tree-ringTREES: Computer vision based tree-ring analysis and measurement system. analysis and measurement system.

Tree-ring Sample Image Acquisition Tree-ring Detection and Measurement Tree-ring Width Standardization

Focus is on a single tree-ring sample.

Page 4: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 4

A Computer-Assisted Tree-RingChronology Composition System

TREESTREESSystem System ComponentsComponents

Page 5: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 5

A Computer-Assisted Tree-RingChronology Composition SystemA Computer-Assisted Tree-RingChronology Composition System

TREESTREESGraphic UserGraphic UserInterfaceInterface

Tree-Ring ViewsSkeleton Plot

SequentialProcess Steps:1. Capture Frames2. Create Mosaic3. Find Rings4. Measure Rings5. Create Skeleton Plot

Page 6: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 6

A Computer-Assisted Tree-RingChronology Composition System

TREES Program Output:TREES Program Output:Index Value Time SeriesIndex Value Time Series Standardization is the process of

modeling and removing non-climatic variation to reduce absolute tree-ring width measurements to a common relative scale.

Raw tree-ring width measurements are detrended to produce index values.

Index values represent relative variations in growth and are the basis of comparing one sample’s growth to another.

Page 7: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 7

A Computer-Assisted Tree-RingChronology Composition System

CROSSDATE: A Computer-Assisted Tree-RingCROSSDATE: A Computer-Assisted Tree-Ring Chronology Composition System Chronology Composition System

Separate application from TREES. Output of TREES is input to CROSSDATE. Works simultaneously with TREES. Focus is on multiple tree-ring samples. A tool, not an expert system.

Page 8: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 8

A Computer-Assisted Tree-RingChronology Composition System

ObjectivesObjectives

CROSSDATE is a software tool to assist dendrochronologists to:

– Determine the dating of a tree-ring sample.Determine the dating of a tree-ring sample.– Compose a master chronology for a site from Compose a master chronology for a site from

individual samples.individual samples.– Perform a QC Check to ensure the validity of a master Perform a QC Check to ensure the validity of a master

chronology.chronology.

– Determine the dating of a tree-ring sample.Determine the dating of a tree-ring sample.– Compose a master chronology for a site from Compose a master chronology for a site from

individual samples.individual samples.– Perform a QC Check to ensure the validity of a master Perform a QC Check to ensure the validity of a master

chronology.chronology.

Page 9: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 9

A Computer-Assisted Tree-RingChronology Composition System

RequirementsRequirements Ability to determine, manually and with computer

assistance, the point where two tree ring samples match in time.

Ability to compose master chronologies from individual samples and other master chronologies.

Ability to rigorously check a tentative chronology for correctness.

Ability to determine and review the specific samples from which a master chronology is composed for purposes of returning to the original wood sample.

Page 10: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 10

A Computer-Assisted Tree-RingChronology Composition System

RequirementsRequirements

Easy to use, intuitive graphic user interface (GUI) that mimics the steps a tree-ring scientist follows when performing manual crossdating.

Ability to exchange data (Import/Export) in common formats to other tree-ring application programs.

Ability to operate on PC and UNIX workstation hardware platforms.

Ability to operate seamlessly with the TREES program.

Page 11: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 11

A Computer-Assisted Tree-RingChronology Composition System

CROSSDATE CROSSDATE Graphic User Graphic User

InterfaceInterface

Page 12: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 12

A Computer-Assisted Tree-RingChronology Composition System

Correlation ToolCorrelation ToolMatch two (2) standardized tree-ring samples in time based on relative patterns of variation in growth.

Procedure:1. Calculate correlation coefficient (r) for each possible match point.2. Convert r to t-statistic for hypothesis tests.3. Evaluate the significance.4. Rank and display the results.

Page 13: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 13

A Computer-Assisted Tree-RingChronology Composition System

Correlation CoefficientCorrelation CoefficientEfficiently computed using Fast Fourier Transform (FFT) and “One-Pass” statistical accumulation.

r k Sxy

Sxx Syy

Sxy k 2 x(n)y(n k)

n0

L 1

L 1

x y

y

x

Y

( L 1)

xy

X

Y*

FFT(.)

FFT-1(.) XY*

FFT(.)

Sxy

CONJ(.)

Correlation Coefficient:

Sxx k 2 L x(n)2 x(n)

n0

L 1

2

n0

L 1

L(L 1)

Statistical Accumulation forStandard Deviation:(Sxx shown, Syy similar)

Cross-Correlation via FFT:

Page 14: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 14

A Computer-Assisted Tree-RingChronology Composition System

CROSSDATE Correlation GraphCROSSDATE Correlation Graph

Page 15: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 15

A Computer-Assisted Tree-RingChronology Composition System

Composition ToolComposition ToolCreates a master chronology from individual samples and other master chronologies after performing pattern matching. Definition: A master

chronology is a composite of individual samples. Composite means the arithmetic average of component samples index values for each year of the entire time period range.

Page 16: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 16

A Computer-Assisted Tree-RingChronology Composition System

Master ChronologyMaster Chronology Useful for recording the growth pattern variation for a site over

some time period. Subsequently useful for dating new samples using correlation

pattern matching. International Tree Ring Data Bank (ITRDB) provides a simple

format for data exchange. CROSSDATE imports and exports ITRDB format chronology files.

The annual average index value and number of component samples are exchanged with other programs.

Average (year ) Component IndexValues (year )

Number of Component Samples (year )

Page 17: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 17

A Computer-Assisted Tree-RingChronology Composition System

CROSSDATE Master ChronologiesCROSSDATE Master Chronologies

Composedhierarchically

usingintermediate

chronologies.

Page 18: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 18

A Computer-Assisted Tree-RingChronology Composition System

CROSSDATE CROSSDATE Master Master

Chronology Chronology Display Display ScreenScreen

Sample Depth

Components

Page 19: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 19

A Computer-Assisted Tree-RingChronology Composition System

Check ToolCheck ToolPerforms extensive checking on each sample component of a master chronology. Checks performed:

1. Date Alignment2. File Date Check3. Statistical Outliers4. Overall Fit5. Segment Fit

Results are available on Graphic Display and as a Text Report.

Page 20: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 20

A Computer-Assisted Tree-RingChronology Composition System

Component QC Check Panel Component QC Check Panel (Graphic Display)(Graphic Display)

Page 21: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 21

A Computer-Assisted Tree-RingChronology Composition System

Component QC Check Report Component QC Check Report (Text Report)(Text Report)

Page 22: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 22

A Computer-Assisted Tree-RingChronology Composition System

Component QC Check Report Component QC Check Report (Text Report)(Text Report)

Page 23: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 23

A Computer-Assisted Tree-RingChronology Composition System

CROSSDATE CROSSDATE User HelpUser Help

– HTML User Help Files are distributed with CROSSDATE.

– Menu button invokes Netscape browser with CROSSDATE help index.

– User Help Files may be viewed with any HTML viewer.

Page 24: TREES: Computer vision based tree-ring               analysis and measurement system.

27 July 2000 J. Engle, University of Arizona, ECE-DIAL/LTRR 24

A Computer-Assisted Tree-RingChronology Composition System

ConclusionConclusion

Contributions to DendrochronologyContributions to Dendrochronology

• Graphic framework for performing tree-ring sample Graphic framework for performing tree-ring sample comparison activities.comparison activities.

• Uses familiar methods.Uses familiar methods.• Completes the TREES system.Completes the TREES system.• May be extended as new methods are developed.May be extended as new methods are developed.

Contributions to DendrochronologyContributions to Dendrochronology

• Graphic framework for performing tree-ring sample Graphic framework for performing tree-ring sample comparison activities.comparison activities.

• Uses familiar methods.Uses familiar methods.• Completes the TREES system.Completes the TREES system.• May be extended as new methods are developed.May be extended as new methods are developed.