Simultaneous Classification of TimeVarying Volume Data Based on the Time Histogram - PowerPoint PPT Presentation

Loading...

PPT – Simultaneous Classification of TimeVarying Volume Data Based on the Time Histogram PowerPoint presentation | free to download - id: 13f490-NDlmZ



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Simultaneous Classification of TimeVarying Volume Data Based on the Time Histogram

Description:

Time Histogram based Classification. Constrained Freeform Intervals ... for classification. Time Histogram ... Semi-Automatic Classification. Motivation ... – PowerPoint PPT presentation

Number of Views:123
Avg rating:3.0/5.0
Slides: 23
Provided by: aki56
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Simultaneous Classification of TimeVarying Volume Data Based on the Time Histogram


1
Simultaneous Classification of Time-Varying
Volume Data Based on the Time Histogram
  • University of California, Davis
  • Hiroshi Akiba, Nathan Fout, Kwan-Liu Ma

EuroVis06
2
NERSC's IBM SP, Seaborg
Time Consuming
5
120
35
60
90
3
Outline
  • Problem Statement
  • Time Histogram
  • Time Histogram Display
  • Time Histogram based Classification
  • Constrained Freeform Intervals
  • Semi-automatic with Equivalent Classes
  • Visual Feedbacks
  • Applications
  • Demo
  • Conclusion Future Work

4
Problem Statement
  • Challenge
  • Specification of transfer functions for all
    timesteps of time-varying data
  • ? Single Transfer function is not sufficient
  • Time consuming to classify hundreds of thousands
    of timesteps
  • Approach
  • Use Time Histogram as a guiding information to
    simultaneously classify all timesteps
  • Visual Feedbacks with multiple display of
    keytimesteps

5
Time Histogram
  • Histogram frequency counts for each scalar value
  • Often shown as the background of the 1D transfer
    function
  • Time Histogram Set of histograms from several
    timesteps
  • global statistical view of the data
  • Widely used
  • Display and interaction methods Kosara et.al.04
  • Differential time-histogram table Younesy
    et.al.04
  • Dynamic behavior of time-Varying data Doleish,
    04
  • Our Approach
  • Further investigated the display method for large
    dataset
  • Utilize time histogram for classification

6
Time Histogram Display
  • Show on a 2D map by mapping counts to color
  • Linear mapping does not always work
  • A large number of counts are aggregated within a
    small data range.
  • Allow the user to control the mapping function

7
Tjet
Argon Bubble
data value
Combustion (Y_OH)
time
8
(No Transcript)
9
Time-Histogram based Classification
  • Time histogram as a guide
  • ? Constrained Freeform Interval
  • Time-histogram as a selection tool
  • ? Semi-automatic with Equivalent Classes

10
Semi-Automatic Classification
  • Motivation
  • trial-and-error process of finding a good
    transfer function is time consuming and
    nontrivial
  • Method
  • Group connected regions based on the range of
    counts
  • Clicking a segmented region assigns the transfer
    function

11
Semi-Automatic Classification
  • Motivation
  • trial-and-error process of finding a good
    transfer function is time consuming and
    nontrivial
  • Method
  • Group connected regions based on the range of
    counts
  • Clicking a segmented region assigns the transfer
    function

12
(No Transcript)
13
(No Transcript)
14
Visual Feedback
  • Provide Quick overview of time-varying behavior
  • Visual Feedback for classification
  • Temporal Reduction

15
Temporal Reduction Rendering
  • Given n timesteps
  • Find distance between two consecutive histograms
  • Merger two closest histograms
  • Repeat 1.2 untill there are only t timesteps
  • Draw interval representative volume
  • Pick one timestep
  • Find average volume
  • 4D projection Woodring Shen, 03

16
(No Transcript)
17
Summary Future Work
  • UI for time-varying data is proposed
  • Multi-display widgets
  • Time histogram map
  • Conveys time-varying features
  • Classification
  • Direct manipulation widgets
  • Conditional classification
  • Experiments on semi-automatic classification
  • Evaluation of useful statistical features

18
Demo
19

20
Direct Volume Rendering
  • For each pixel, trace ray through data and sample
    along the ray.
  • Map sample value to color and opacity and
    composite onto screen.
  • Accelerated by texture based approach

Volume Boundary
Image Plane
21
Direct Volume Rendering
  • For each pixel, trace ray through data and sample
    along the ray.
  • Map sample value to color and opacity and
    composite onto screen.
  • Accelerated by texture based approach

Volume Boundary
Image Plane
22
Time-Varying Data Visualization
  • Data Reduction
  • Static Encoding
  • the whole data is encoded together by exploiting
    coherency of the data
  • EX VQ, DCT, FFT, run length coding
  • Multi-resolution
  • input data is converted into a hierarchical
    representation of varying resolutions
  • EX Wavelet transformation, space filling curve
  • Feature Extraction
  • the important or distinct structures in data set
    are first extracted based on certain criteria
  • EX Iso-value, topological analysis, volume
    segmentation, machine learning
  • Rendering and User interface
  • Animation
  • Project volumes on a plane
About PowerShow.com