Upload
merrick-company
View
138
Download
0
Embed Size (px)
DESCRIPTION
2014 SPAR International conference presentation by Josh Beck, Senior Software Consultant for Merrick & Company's Geospatial Solutions team. Presentation occurred on Tuesday, April 15, 2014 at SPAR International, held at The Broadmoor Hotel & Resort in Colorado Springs, Colorado.
Citation preview
Engineering | Architecture | Design-Build | Surveying | GeoSpatial Solutions
Real-Time 3D Data Visualization and
Analysis of Unlimited Detail
March 27, 2013
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 2
Presenter
Josh Beck, GISP
Software Consultant for Merrick & Company
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 3
Corporate headquarters: Greenwood Village, CO
Founded in 1955; employee-owned
Corporate Overview
Geospatial
LiDAR Acquisition and Processing
MARS© LiDAR Processing Software
Orthophoto and Photogrammetric
MS / HSI / TIR
Multi-Sensor Data Fusion
Satellite Imagery
Application Development
Surveying
UAS Mapping
National Security
Dept. of Defense
Federal
Nuclear
Commercial Nuclear
DOE / NNSA
Energy
Oil & Gas
Utilities
Bioprocessing
Renewables
Nuclear
The Merrick Consultancy
Life Sciences
Federal
University / Private
Commissioning / Validation
Healthcare Facilities
Infrastructure
Water Services
Local Government
Private Development
Federal / DoD
Energy
Sustainable Infrastructure
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 4
Merrick & Company Office Locations
500 employees at 13 national
and 4 international offices
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 5
Merrick’s International Project Experience
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 6
LiDAR Data Management
Traditional Display and Rendering Methods
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 7
Points Loaded into RAM
Pros Cons Conclusion
Potential for fast 2D and
3D point rendering,
depending on data size
Data processing and view
rendering sessions are
limited by RAM –
performance can degrade
even with large amounts
of RAM utilized
Great for small datasets,
commonly used for
creating static videos of
point clouds (must
recreate video from
scratch for alternative
perspectives)
Fast access to data for
processing, depending on
data size
Wait time for points
loading into RAM is
typically slow,
impracticable for massive
LiDAR datasets
Ineffective for very dense
LiDAR data captures, or
extensive geographic
regions. Impossible to
render huge volumes of
point clouds (“Big Data”)
at full resolution
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 8
Points Loaded into RAM Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 9
Point Streaming, Low Resolution Block Technique, and Point Decimation
Pros Cons Conclusion
Special (not typical) point
decimation techniques
can “intelligently” reduce
the data size while
preserving the important
details (i.e. model
keypoint creations)
Data rendering is
decimated during
navigation and then
streamed in when
movement stops, still a
slow rendering processes
as the software fetches
points from disc
This is the current
accepted technique for
rendering large point
clouds. Most geospatial
analysts are trained to
accept the visual
limitations imposed by
software, hardware, and
“Big Data”.
Less RAM is required
than previous technique,
though more RAM does
increase the performance
High risk of missing small
or key features during
navigation due to minimal
point display during
movements
Unsatisfactory for many
applications where
access to all the data at
full resolution should be
mandatory
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 10
Point Streaming, Low Resolution Block Technique, and Point Decimation Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 11
LiDAR Data Management
Breakthrough Technology
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 12
Advanced Spatial Indexing
Pros Cons Conclusion
All points are viewable at
all times, instantly
Initial one-time indexing
or pre-processing can
take time but once it’s
done, the data loads
instantaneously
Best technology for large
point cloud data
management
Minimal RAM utilization
since spatial indexing is
file and software based
Indexing can require high
performance hardware to
process data
Best technology for
massive LiDAR point
cloud rendering
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 13
Unlimited Size and Full 3D Resolution
Access to all LiDAR and/or 3D point data, no matter what the size
The current system has been tested to support approximately 1.4 petabytes of LAS data – this is only a test limit, not a system limit
The system has been tested with synthetic data at a density of 64 points per cubic millimeter across a test area of 1 square kilometer
2D and 3D navigation is instantaneous, no waiting for data to load or streaming of data into view window
Able to see every point in the dataset, only limited by the number of screen pixels on the viewing device
Load times and speed of performance are not related to the size or resolution of the 3D data
Load times are less than one second, no matter how large the datasets
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 14
Additional Advantages
Points are stored in a compressed format
Compression ratios are around 5-20% of the original LAS size
No decompression lag unlike other compressed LiDAR technologies or file compressors like Winzip or WinRAR
Data can be viewed in stereoscopic 3D if desired
This technology is accessible through a converter/viewer or SDK
No dependency on high end computer hardware
Utilizes multi-core and/or multi-threaded CPUs but does not require high end models
Does not require GP-GPU processing for high speed performance
Does not require large amounts of RAM, 2 GB is adequate
Does not even require high speed disc I/O. Rendering speed is similar whether the data is local, on a network, or even accessed through a USB 2.0 port
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 15
Size, Load Time, and Computer Resources
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 16
Next Generation 3D Software
3D data freedom
Points are not loaded into RAM - no more memory limitation
Points are not decimated or pyramided – no more thinned data displays
Real time 3D point access is direct from disc or network utilizing two revolutionary techniques:
Advanced 3D spatial index
Intelligent search algorithm to find and display just the points needed - at most, one point per screen pixel at 30 frames per second
Ability to work with the actual 3D data, not derivative products or a combination of different geospatial data types
No need to grid elevation data, loosing critical 3D information
Movement away from 2D or 2.5D - actual points can be viewed and measured in 3D space, not 3D represented by only colors or shadings
No need to keep features separated by data types (vector, raster, etc.)
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 17
Urban Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 18
Rural Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 19
Mobile Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 20
Terrestrial Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 21
Internet Streaming
3D models can be accessed worldwide across a network or the internet
No need for multiple local storage points of 3D models if an internet connection is available
3D models can be hosted at a single location
Only a single copy of the data needs to be stored which can be served up to unlimited internet users, dramatically decreasing data storage requirements, costs, and complexities
Version control - if changes to the data are required, these are done only one time to the source data
Data can even be streamed over the internet without the need for a browser plug-in
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 22
Colombian Coast Demo
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 23
More Than Just LiDAR
3D data can present imagery or originate from it
RGB fused LiDAR
3D Ortho – interpolated LiDAR at the resolution of imagery
Colorized 3D point clouds derived from stereo pairs using dense matching techniques (i.e. semi-global matching or multi-ray photogrammetry)
3D models from multi-angled oblique imagery
More groundbreaking possibilities in the works
3D Models
3D models can automatically be converted to this format to be rendered and streamed from a single platform in unlimited amounts and detail
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 24
3D Ortho – not an image drape or TIN render
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 25
Semi-Global Matching
Images courtesy of Intergraph
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 26
Oblique Image SGM
Images courtesy of Intergraph
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 27
Multi-Ray Photogrammetry
Images courtesy of Microsoft's UltraMap 3.0
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 28
Oblique Image Texturing Example
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 29
Summary
Real time 3D data rendering at full resolution and detail
Compressed data size
Data loads in 1 second no matter the size
Unlimited amount of data to manage, load, display, etc.
Low computer resource requirements
Preserves true 3D data, not rasterized derivative products
Capable of streaming 3D data across network or internet
Centralized data storage
Version control
Big data is now accessible and even portable
Opportunity for 3D imagery and 3D models
Copyright © 2010 Merrick & Company All rights reserved.
PREXXXX 30
Thank you
Josh Beck
Software Consultant
Merrick & Company
303-353-3715
http://www.merrick.com/Geospatial