Download Visualization

CS591: Introduction
Mengxia Zhu
Fall 2007
Class objective
To study visualization principles,
techniques and algorithms which are used
for exploring, transforming and viewing
data as computer images to gain
understanding and insight into the data.
 Introduction to basics of parallel computing
and MPI for large scale scientific datasets.

Course materials

No textbook required

Lecture notes

Posted on
http://www.cs.siu.edu/~mengxia/Teaching.htm

Research papers


Distributed/referred in class
Web sources

Referenced in lectures
My expectation
Experience in C programming
 Basic Algebra and calculus
 Basic understanding of computer graphics
and OpenGL
 A little deprivation of sleep…

Grading Policy


Midterm and final exam
Grading items:

Homework: 20%
 Mid term and final exam: 30%
 Lab projects: 40%
 Paper presentation: 10%

Grading Scale:







A = 85% or more
B = 75% to 84%
C = 65% to 74%
D = 50% to 64%
F = below 50%
Late submission will be punished.
Academic dishonesty will be treated seriously
Office Hours

Regular Hours
 M,

W, F: 12:00PM — 12:50PM
Special Hours
 Any

time by appointment
Contact Info
 Office:
Faner 2142
 Email: [email protected]
 Phone: (618)453-6057
Computer Graphics for Visualization

OpenGL

Drawing geometric objects

Viewing

Interception and Culling

Lighting and Shading
Scientific Visualization

Isosurface rendering

Volume rendering
 Splatting
 Raycasting

Vector and tensor visualization
What Visualization?
Process of making a computer image or
graph for giving an insight on
data/information
 Transforming
abstract, physical
data/information to a form that can be seen
 Interpreting in visual terms or putting into
visual forms (i.e., into pictures)

Cognitive process
 Form
a mental image of something
 Internalize an understanding
Visualization Process
Computation:
Measured/Scanned
Data:
-CT, MRI, ultrasound
Financial data:
-transactions per day
-simulation/modeling
Data
Transform
Map
Display
Viz vs. Graphics vs.. Imaging

Imaging - Enhance, analyze, manipulate
images

Graphics - Make pictures! geometric
data is stored in the computer for the
purposes of performing calculations and
rendering 2D images

Visualization - Exploration,
transformation, viewing data as images
Relation To Other Fields
Illumination
Signal/Image
Engineering
Processing
Optics
Computational
Vision
Geometry
Visualization
Applied
Psychology
Mathematics
Cognition
User
Hardware
Interfaces
Why?

Extends our vision
 Removes
limits of human vision in space,
time, frequency and complexity
 Creates images or pictures of things that
otherwise can not be seen
See an object’s internal structure (visible man)
 See things that are far away or slow in
evolution (stars and nebulas)
 See microscopic world (crystal structure)
 See things that move very fast (molecular
dynamics)

Human Inner Organs

Visible (voxel) man
 Reconstruction
of
human body from
tomographic datasets
of dissected real
body
www.uke.uni-hamburg.de
Stars and Emission Nebulas

Visualizing
Orion
Nebula:
Nadeau et al.,
Computer
Graphs Forum, 20:
27
(2001)
Crystal Structure

MgSiO3
perovskite

An orthorhombic
unit cell

Atomic
coordination
Types of Visualization

Scientific Visualization
 Scientific

data
Information Visualization
 abstract
data has no inherent spatial structure
thus it does not allow for a straightforward
mapping to any geometry with arbitrary
relationship

Data Visualization
 A more
general term
 data sources beyond science such as financial,
marketing, or business data
 Broad enough to encompass both scientific and
information visualization
Scientific Visualization

Relates to and represents something
physical or geometric
 Images
of human brain
 Air flow over a wing

Data come from scientific computing
and measurements
Scientific Computing

Real materials simulation/modeling
 Electronic
calculations
 Atomistic MD (molecular dynamics)
modeling
 Finite element (continuum) modeling

Solving differential equations
 Computational
fluid dynamics
 Temperature distribution
 Electromagnetic field
Example: Air Flow over Windshield

Air flow
coming from a
dashboard vent
and striking the
windshield of
an automobile

http://wwwfp.mcs.anl.gov/
fl
Measurement: Medical Imaging
Standard brain CT image
Volume rendered brain image
Ultrasound
http://www.gemedicalsystems.com
Challenges?

Scale

Dimensionality

Data types

Presentation

Interactivity
Data Explosion

How to make sense out of the datasets
when they become very large

Scientific data
 A million-atom
simulation: 7 GB/step
 Satellite or space station: TB/day
 MRI dataset: 2563 = 16 MB/slice
 Laser scanning: 2 million points/minute
Dimensionality

Three dimension (trivariate data)
 We
are in 3D world
 Volume visualization (mapping 3D data to
2D screen)

Multidimension (hypervariate data)
 Car
attributes: Make, model, year, miles
per gallon, cost, no. of cylinders, size,
weight
 How to display relationships between many
variables
Data Types

Structured versus unstructured data
 Unstructured
(irregular) data are less compact and
efficient
 Preprocessing of data

Scalar, vector and tensor data
 Density,
temperature
 Data from flow dynamics
 Stress-strain data

Non-numerical data
 Ordinal:
days of the week
 Categorical data: names of animals
Presentation Problem

Display without ambiguity
 Colors,
lighting, translucent, animation,
texture mapping

Too much data for too little display area
(screen)
 Too
many cases
 Too many variables

Need to highlight particular cases or
variables
Interactivity

Visualization is naturally interactive
 Real-time
interactions, i.e, virtual
environments
 Show
data
multiple different perspectives on the