Transcript Exploring the Synergies between Commercial and

Exploring the Synergies between
‘Commercial’ and Open Source
Software
Gregg Petrie, Cosmic Project
Lee Miller, Pacific Northwest National
Laboratory
Outline
•
•
•
•
•
•
Introduction
Study Approach
Specific Examples
Conclusions
Action Items
Questions
Introduction : Disclaimer
• Represent Users-Hyperspectral Remote Sensing
NOT claiming to be an expert hacker
Introduction : Motivation
Current Impression
Current Hope: The Whole > Parts
Introduction : Goals
• Identify, develop, test, refine, and
demonstrate synergies using real world
examples
• Demonstrate how open source community
can support users of commercial software
• Promote the support of open source by
commercial vendors
Introduction : More Information
COSMICPROJECT.ORG
Outline
•
•
•
•
•
•
Introduction
Study Approach
Give Some Concrete Examples
Conclusions
Action Items
Questions
Approach
• Survey of Open Source Possibilities
• Create and Experiment With a Loosely
Coupled Environment
• Create and Experiment With a Tightly
Coupled Environment
Approach: Survey
• 238 Open Source Geospatial Packages
• 24 Scientific Open Source Python
Packages
Approach-Survey Observations :
Python
• Provide access to a wide range of additional
tools
• High level structures promote rapid
prototyping
• Self Documenting
• Builds on the Python/ESRI Experience
• Widely used to glue capabilities together
• Well supported (e.g. .NET)
• Python supports both UNIX and WINDOWS
• Use of Python growing in the scientific world
Approach-Survey Observations :
Python Issues
•SPEED
.NET
Moore's Law
More C code
Approach-Loosely Coupled
Environment: Dealing with the
‘Tower of Babel’ file format problem
GDAL
Vendor DLL
Write your own
Loosely Coupled Environment: Actual Solution
for the ‘Tower of Babel’ file format problem
TIFF files
•Flexible
•Supported by many commercial packages
•Stable
•Dealing with large file sizes
•Well Known
Tiff Flexibility :Two Edged Sword
Outline
•
•
•
•
•
•
Introduction
Study Approach
Specific Examples
Conclusions
Action Items
Questions
Examples: Leica Imagine
Imagine provides ‘standard’
remotes sensing tools that
include:
COSMIC
Python
Easy access to FORTRAN and C
software packages
File creation
Registration
Interface to Windows/UNIX AIP
AI tools
Registration
Display
Python tools
IDL like capabilities to write
advanced Hyperspectral
analysis tools
Quick GUI development tools
Access to .NET
Classification
SciPy toolkit
File import/export
WEB tools
SDC Morphology toolbox
Photoshop for presentation
graphics
‘Twisted’ for event-driven
networking
CGAL for computational
geometry
Example: Imagine
Example: Imagine
Capabilities
Professional
Advantage
Essentials
Cost
Example 1: Web Mapping
Examples: Web Mapping
Step 1: Trail Extraction
Step 2: Annotation of Nodes
•Used Open Source to
translate to ASCII files
•Used Python to add
USGS elevation data to
nodes and calculate
distance
•Used the Open Source
PIL software to create gif
file
Example 2: Color Compression
Example 2: Imagine Approach
Example 2: Combined Approach
For each class find
the average color and
update the LUT
Example 2: Results
Top Half=RGB image
Lower Left=
Unsupervised color
Lower Right
=Imagine color
Menu Maze Problem
Example 2: Real Usefulness
Pick the 3 best
BANDS for each
land cover
Image source:http://www.csr.utexas.edu/projects/rs/hrs/hyper.html
Outline
•
•
•
•
•
•
Introduction
Study Approach
Specific Examples
Conclusions
Action Items
Questions
Conclusions
Whole > Sum of the Parts
Capabilities
Professional
Open Source
Advantage
Essentials
Cost
0 Cost
Conclusions: NumPy and Python
can be used to “glue” software
+
And now for
something
completely
different...
Conclusions: Tiff files, with a ‘little’
improvement’ can greatly mitigate
the ‘Tower of Babel’ file format
problem
Conclusions: Vendor Specific
Commercial Advantages
Know QA factors
Professional
Good documentation
Open Source
Capabilities
Not all users programmers
Advantage
Can be faster
Can be cheaper
Can by bureaucratically compatible
Essentials
Cost
Conclusions: Vendor Specific
• Empower Users
• Help current developers reduce both prototyping
costs and time to market
• Increase the rate at which new ideas· developed
by users could be prototyped, tested and moved
into ‘native’ proprietary commercial tool kit
• Provides an alternative to the Menu Maze
• Build on the Python/ARC ESRI Experience
• New pathway for new users
Outline
•
•
•
•
•
•
Introduction
Study Approach
Specific Examples
Conclusions
Action Items
Questions
Action Items For Vendors
• Continue Support BigTiff
• Work together to establish Tiff conventions for
hyperspectral data sets
• Build on the ESRI/Python Model
• Support a Tightly Coupled Environment
Action Items For Open Source
• Support BigTiff
• Work together to establish Tiff conventions for
hyperspectral data sets
• Move toward tightly coupling commercial and open
source data
• Improve Documentation for GeoTiff data
• Couple systems (e.g. GRASS) to Commercial packages
Questions