AeroMetric_waveform_presentation

Report
Full Waveform LiDAR
Understanding full waveform and how it works
Jamie Young
Senior Manager-LiDAR Solutions - AeroMetric







History and background
What is Full Waveform
How does it work
Specifications
Hardware
Process and Software
Application comparison
Benefits
Questions?
TOPICS
Full Waveform
History of full waveform digitization almost as old as LIDAR
1994 Capability exists in bathymetric systems
2002 2d visualization concepts for pseudo-waveform data
2005 Waveform digitizing adapted to terrestrial use
2006 Gaussian waveform decomposition used in Bathymetric systems
2009 Capability available from all major manufacturers
LAS 1.3 file format released
TerraScan processing of waveform data
Glossary -frequently used terms
FWD - Full Waveform Digitization
MRI - Multiple Returns with Intensity
Minimum return separation - minimum range difference for which
independent range/intensity measurements can be made
Sample depth - resolution of the intensity measurement made at each
digitizing interval
Sample interval- time interval (usually in nanoseconds)
between intensity samples
Waveform length -number of samples, or total distance,
digitized within the capture waveform
What is Full waveform?
 The laser pulse is emitted and all the return information of
that pulse is received back to the receiver and stored.
 The system needs to be set up to store the amount of
information desired
 64 samples
 128 samples
 256 samples
 The return information needs to be converted to usable data
 Basically, the full waveform data is converted to discrete
return information
What technology is used?
 The systems used are the same as what is used for
traditional LiDAR sensors with one exception.
 A full waveform digitizer




GPS
IMU
Laser
Scanner
LIDAR waveform
how is it created?
 Multiple return pulses are
generated as the laser pulse
hits various levels in the forest
canopy, creating in total a
complete return waveform
 Waveform measurement is a
natural extension of the
conventional “discrete-return +
intensity” measurement process
Footprint
Return
waveform is
generated by
all reflective
surfaces
within the
laser footprint
Laser
Footprint
Full Waveform Digitization (FWD)
basic concept
Start Pulse
Tn ,
In
T1 ,
I1
Detector
Signal
LIDAR waveform visualization
PULSE 4
PULSE 5
Pulse 3
 Each output laser
pulse will hit a unique
combination of
surfaces on the terrain
below:
- Different elevation
- Different percent of
footprint intercepted
at each foliage level
- Different reflectivity
of intercepted
surfaces
 Each output laser
pulse will result in a
unique waveform
shape
PULSE 1
PULSE 2
What is Full Waveform Digitization?
capturing the complete return, not just the peaks
 Conventional discrete return
electronics capture only the
exact time of the peaks of
independently-recognized
return pulses
 Peak intensity is also measured
 In FWD systems, the entire
return signal is measured,
allowing capture of subtle
deviations in the shape of the
reflected pulse as compared to
the shape of the outbound laser
pulse
Discrete Returns
Waveform
Exploiting individual waveforms
Gaussian decomposition for finding “buried” data
 Gaussian decomposition
 Return signal from ground
 Return signal digitized at user-selected
interval (typically 1 ns; equivalent to ~15
cm height)
 Fitting of first return Gaussian component
 Fitting of second Gaussian component
 Fitting of third Gaussian component
 Note: Each Gaussian component
must be fitted for:
 Time of occurrence
 Peak amplitude
 Pulse width
 Benefit: any “stretching” of pulse
detected can be used to indicate
vegetation height on ground (and
automatic adjustment of range) or
inclined surfaces
AeroMetric (Leica ALS-70) Specifications
WDM65 Waveform Digitizer Module
Specification
Maximum waveform rate
(waveforms captured for
every other laser shot if
pulse rate>120 kHz)
Sample depth,
Sample interval
Integration
Weight
Power
Typical Altitude
Storage
Value
120 kHz
256 samples @ 1.0 ns
128 samples @ 1.0 ns
256 samples @ 2.0 ns
128 samples @ 2.0 ns
64 samples @ 2.0 ns
256 samples @ 4.0 ns
New internal DLM
1.0 kg
77 W
2400 AGL
512 GB SSD
Equivalent tree height
38.4 m
19.2 m
76.8 m
38.4 m
19.2 m
153.6 m
Operating envelope
max waveform rate versus slant range
 At pulse rates below 120 kHz,
waveforms captured at laser
pulse rate
 At pulse rates above 120 kHz,
waveforms capture for every
other pulse, up to 200 kHz (150
kHz for ALS50-II)
Hardware configuration
full waveform digitizing for ALS
 Upcoming release – announce 17 Nov
2009
 Core is new “FWD-ready” Data Logger
Module (DLM65) – installed in all new
ALS60 systems
 Existing HDD replaced by 160 GB SATA
SSD (MM60) and allows missions up to
7620 m AMSL equivalent cabin pressure
 3 variations
 Option on new ALS60 (771706)
 Upgrade on fielded ALS60 (771708)
 Upgrade on fielded ALS50-II (773668 +
771707)
 Waveform viewer software + ALSPP data
output in LAS 1.3 format
Hardware details
DLM65 / digitizer kit
DLM65 chassis
Waveform
Interface PCB
added to System
Controller tray
(signal splitter)
Double-wide
CPU replaced
with faster singlewide CPU with
on-board SATA
driver (releases 2
slots)
Slot Blocker
Existing 32-bit
DIO PCB for
System
Controller data
logging remains
FWD kit
2x waveform
digitizer PCB (60
kHz max each)
1x time
synchronizer
PCB
Firmware
license
DLM Power supply hard
mounted to card cage (releases
2 slots)
Baffle to direct airflow
Some points about FWD
 Intensities must be digitized at <2 ns intervals to minimize
aliasing, though 1 ns more common
 1 ns in time represents 0.15 m in range (i.e., elevation)
 Signal amplitude at each interval typically digitized at 8-bit
resolution (i.e., one byte)
 Therefore, 256 additional bytes of waveform data needed to
digitize the return waveform from a 38.4 meter-tall object @
1 ns intervals
 Range data is still be measured independently to achieve
typical 1.5 cm (i.e., 100 ps) range resolution
Using waveform data for classification concept
 Assumes that waveform shape/content, as opposed to mere
extraction of equivalent discrete returns, is used to classify
object which reflected laser pulse
 Assumes that waveform is compared to a “catalog” of
“typical” waveforms for different target classes
Software
FWD-equipped systems release summary
Software Release Comments
Number
TracGUI
2.73 #2
Datalogger 7.1.0.23
FCMS
3.20
Intel SSD
2CV102HA
ALSPP
2.70 #7
Executable version only – works OK
Still need installable version
WaveViewe 1.0.0.4
r
Executable version only – works OK
Still need installable version
TerraScan
Released by Terrasolid November 2009
9.15
Officially released beta to be used only for FWD-equipped;
otherwise use 3.15 even for FWD-ready systems
Finished testing 10 November 2009
FWD post processing Overview
 ALS Post Processor
 Support is available. v2.73 #2 (or greater)
 Outputs LAS 1.3 type 4 files
 Wave Viewer Utility. v1.0.0.4 (or greater)
 Simple LAS 1.3 Waveform file viewer
 TerraScan (See other CSS workflow documentation for
details)
 Support for LAS 1.3 released Oct ’09. v9.14 (on www.terrasolid.fi)
 The following features are included:
 View waveform data for a selected point in the point cloud
 Scan the waveform for returns that were too close together for the discreteranging electronics to detect or because the returns were below the threshold
discriminator (i.e., creates a new “discrete” return)
 No other calibration needed
FWD processing directory structure
FWD - flight planning
 No support for waveform data collection in FPES initially (i.e.,
waveform capture settings must be manually entered in FCMS)
 Four settings control system configuration for waveform capture




Pre-trigger samples (configured in FCMS hardware configuration)
Number of samples
Sample interval
Maximum pulse rate
 Options can be preconfigured or changed during flight execution
FWD - flight execution using FCMS
FWD -flight execution with FCMS
FWD - TracGUI real-time waveform display
Waveforms are shown in this
display, in flight, while online using F4 F4 W command
Classification flowchart
Load sample
waveform
Sample
waveform
Target
waveform
1
Load next
target
waveform
Compare sample
waveform to target
waveform
Target
waveform
2
Match?
No
Target
waveform
3
Yes
Any target
samples left?
Assign
Yes class
No
No
match
Wave Viewer screen
Data output
waveform viewing and access
 Wave Viewer
 Main statistics (sample rate, waveform depth,
waveform sequence number, timing data)
 Allows scrolling through entire captured series of
waveforms
 Displays waveform
 Displays time/intensity indicator (2 black boxes
near waveform peaks) from discrete-return data
collection that operates in parallel with
waveform capture
 TerraScan
 Waveform display associated with any given
discrete-return point
 Future expansion allows possible improvement by
deriving additional parameter (pulse stretch) in
pre-processing and passing these on to TerraScan
for more efficient//accurate filtering
FWD - ALS Post Processor/ing
 Select “Process Waveform
Data” option.
 Software looks for “RawWfd”
folder with matching mission ID.
 Automatically output LAS 1.3
file for each flight line
FWD - Wave Viewer
 X-axis: sample (number), time
(ns) or range (m)
 Y–axis: signal strength (volts or
counts)
 Used to review data and
confirm that waveform data is
correctly correlated with
discrete-return data
FWD - post processing: Wave Viewer
 In this example, two returns
were missed by the discretereturn ALS range electronics
 The first missed pulse was too
close to the first pulse to be seen
 The second missed pulse was too
small to be seen.
FWD - post processing: intensity scale
 Note: Y scale factor of the raw intensity values and the waveform
digitized values are not the same.
 Intensity PCB in the SCM is designed to use the full 8-bits for the typical signal
amplitudes
 The waveform digitizer PCB has limited configuration options so the the option closest to
the Intensity PCB scale factor is used
 The intensity board uses the following scale (latest revision only,
earlier versions may have different scale factors):
 0.110V signal = 10 counts
 3.7V signal = 250 counts
 The waveform digitizers use the following scale:
 0 V = 0 counts
 4.1V = 255 counts
FWD - post processing: sample files
 Sample LAS 1.3 files will be located on FTP site
 Contents
 Wave Viewer
 LAS 1.3 file format specification (can be released to public)
 Data set
 Note: Contact PM for FTP information
FWD –post processing
TerraScan displays
FWD - post processing
TerraScan capabilities
 Viewing of waveform profile
when clicking on discrete points
in the point cloud
 Mensurating more discrete
returns from WF data (“extract
echos” feature)
FWD -TerraScan V10.17+ “Extract Echoes” feature
 Allows extraction of small signals via user-adjustable “ambient noise”
threshold
 Allows extraction of returns at less than the minimum separation
dictated by discrete-return electronics (see cyan-colored points
below)
MRI versus FWD: Applications
Application or surface type
MRI LIDAR
FWD LIDAR
Yes
Yes, but overkill
Maybe, species dependent
Yes, but may be overkill
Yes, depending on floor
cover
Yes, but may be overkill
No
Yes
Yes
Maybe, if lines separated
by less than minimum
separation distance
Maybe, species dependent
Yes
Sloped surface detection
No
Yes, but height difference
over laser footprint must be
~1 sample interval
Bathymetry
No
Yes
Tree height
Forest canopy structure
Forest floor
Tall dense grass
Power line profiling
Tree species identification
Using waveform data for classification caveats
 Each return waveform highly dependent on
 Specific geometry (i.e., portion of laser footprint intercepted at different heights
above ground)
 Reflectivity of each surface intercepting a portion of the laser footprint
 Therefore
 Waveforms returned from a group of nearby laser shots may have to be
averaged to arrive at a more consistent “typical” waveform for comparison to the
target waveform library
 Waveforms to be averaged must be referenced to some consistent key point
(e.g., ground level) before averaging, in order to generate a meaningful
comparison
MRI and FWD provide some similar functionality
 FWD exploitation techniques can extract returns that are “buried”
 FWD waveforms from a group of adjacent laser shots may need to be
averaged to make automated classification feasible
 Averaging MRI “pseudo-waveforms” from a group of adjacent laser
shots may be an effective alternative to FWD, especially if
 LIDAR system can measure intensity for a large enough number of returns for
each outbound laser pulse (e.g., 3+)
 Inter-return minimum separation distance is small enough (target dependent)
 High point density in MRI systems helps to overcome the additional
per-pulse information supplied by FWD systems
 Some applications may be accomplished by both methods, but may
be better suited to one or the other (see table)
Benefits of FWD
getting more from a single flight
 Present
 Extraction of points below the discrimination threshold of discrete-return electronics (weak
returns)
 Extraction of points with smaller vertical separation than detectable by discrete-return
electronics (close, but not overlapping pulses)
 Future
 Detection of pulse stretching (return pulse wider than laser pulse) indicating
 Potentially sloped surfaces
 Low vegetation on ground, indicating need to adjust point elevation downward
 Improved classification by using combination of return pulse width and spatial context
 Indication of biomass by evaluating area contained under the pulse shape
WILDER LiDAR Blog
http://bloglidar.wordpress.com
Thank You
Thank You to Leica for Contributions to this presentation

similar documents