Using LiDAR Data in ERDAS IMAGINE

Using LiDAR Data in ERDAS IMAGINE
Jay Pongonis
Senior Technical Instructor
LiDAR Basics
Introduction
Recent developments in
Photogrammetric terrain extraction
show great promise but…
– Airborne LiDAR still better at
extracting forest floor
– LiDAR still better at certain
linear features (e.g., power
transmission lines)
– LiDAR offers 24-hour operating
envelope
What is LiDAR?
– LiDAR is a technology that measures distances by sending pulses of light at an
object.
– System measures the time it takes for a pulse to return.
– Time is then converted to distance which in turn is converted to geo-referenced
data – in close-to real time.
– LiDAR is an active sensor; therefore, data can be acquired day or night (as long as
the atmosphere is clear).
– Generates large datasets (up to 500,000 points/second).
– Despite the size, data can be post-processed to provide accurate and detailed
Digital Elevation Models (DEM).
– Variations of LiDAR include airborne, terrestrial and mobile systems.
5
Generic airborne LiDAR: hardware
Capture
Generic LiDAR: the point cloud
 Not a raster
 Represents XYZ points measured
 Can be more than one point per outbound laser shot
 Typically color coded by
– Elevation
– Return number
– Intensity
– Class
– Flight line
LiDAR Viewing basics
by elevation
by class
10
by intensity
by return
What are LAS Formatted LiDAR Files?

The LAS file format is a public file format for the interchange of LiDAR data between vendors
and customers.

Developed by American Society for Photogrammetry and Remote Sensing (Version 1.2
released April 29, 2008).

Committee was made up of:
–
stellar photogrammetric and remote sensing professionals, academic institutions, government
agencies and the private sector

Important attributions include but are not limited to:
–
Classification and Classification Flag
–
Intensity and Elevation
–
Color (Version 1.2)
–
Return signal waveform (Version 1.3)
–
Further expansions in Version 1.4
11
LiDAR is the “Third” Type of Data
Vector
Point Measurements and Contours have been used historically to
represent terrain surfaces. These are combined with break lines to create
Triangulated Irregular Networks (TINs) from which surface points can be
interpolated.
The data representation is a sparse set of highly irregularly space {X,Y,Z}
values.
Raster
They have been converted to gridded formats using various techniques
to produce Raster datasets. Delivered as Digital Elevation Models
(DEMs).
The representation is a dense set of regularly spaced {Z} values.
Point Cloud
LiDAR data is a collection of points with attributes.
The representation is a dense set of semi-regularly spaced {X,Y,Z,
Attribute..} values.
12
Background: conventional airborne LiDAR

Swath and point density depend on flying height,
FOV, scan rate, pulse rate, aircraft speed

Direct measurement – no further processing
required to create point cloud

ALS70 (shown at right) is typical “full-capability”
system

Point clouds used for both modeling and “bare
earth” terrain extraction
Background: mobile mapping LiDAR

Custom-integrated system (shown at right)
is common

Several permutations used in industry
– LIDAR-only (such as system at right,
which uses Leica HDS 6100 scanner)
– Camera-only
– LIDAR + camera
LiDAR technology development

Historical improvements
–
–
–
–
–

Accuracy
Pulse rate
Minimum vertical separation distance
Full waveform digitization (FWD) acquisition and exploitation
Scan pattern control (pattern shape and scan rate)
Areas with greatest improvement
–
–
Accuracy
Pulse rate
Accuracy leveling over recent years

~ 50% reduction every 5 years
– Rapid improvement in late 1990s
– Slowing absolute rate of improvement
in recent years

Limiting factors
– Airborne GNSS accuracy
– Availability of high accuracy ground
control over large job sites
Pulse rate improvement still steady
Pulse rate as indicator of productivity
– Intuitive: more points per hour  less
hours flying
– Less intuitive #1: more points per hour
 wider swath in each flight line 
less side overlap (%) to overcome
navigation error  fewer flight lines
– Less intuitive #2: same point density
from higher altitude  reduced swath
width variation due to terrain elevation
changes  less side overlap (%) to
overcome swath width variation 
fewer flight lines
Recent developments: growth in pulse rate

Up until ~2004: limited by
– Max pulse rates of available lasers
– Relatively high end-of-cycle timing
overhead
– Fly lower to pulse faster

2006 – 2009: Multiple Pulses in Air (MPiA,
a.k.a., “CMP”, “MTA”)
– Allowed laser to be fired before
reflection from previous pulse is
received
– Doubles the pulse rate for a given
flying height
– Practical to achieve high pulse rates at
reasonable altitudes
– Limitations
 Pulse consistency
 Adequate pulse energy

2009: first dual-output scanners announced
Single-output limits along-track spacing

better spacing is in cross-track direction only with greater pulse rate
Note along-track spacing twice as large at FOV edge as at nadir!
Dual-output scanning doubles scan rate, pulse rate

doubles effective scan rate and pulse rate
Note that along-track spacing is same at FOV edge as at nadir!
Why LiDAR?
applications and advantages
Summary of applications
broad categories and limitations

Applications – anywhere surface data is desired

DSMs/DEMs for
–
–
–
–
–

Orthorectification of image data
Modeling
Visualization
Change detection
Metrics (timber stand volume, biomass, stockpile volume)
Applications are limited only by
–
–
Sensor spatial resolution (up to 30 points/m2 achieved in fixed-wing aircraft)
Accuracy (typically 5-15 cm, but 3 cm achieved with care)
Wide-area mapping
MPiA technology at work

MPiA is now a mainstream
technology

Huge projects being undertaken
w/ MPiA systems (Example –
NWG has collected 750,000
km² collection @ 1 point/m²)
Image courtesy of North West Geomatics
Hydrology
flood plain mapping and simulation
Hydrology
erosion studies
Forestry
tree height and biomass estimation

Top View – Color Coded by elevation

Section view color coded by class
–
–
–
Brown = Ground
Green = Vegetation
Red = Model Key Points
Forestry
accurate ground profiling during leaf-on conditions
Urban modeling
photorealistic rendering for visualization
Urban modeling
building extraction
Urban modeling
detailed “as-built” data
Mining and construction
accurate volumetric calculations
Corridor mapping
power line position and vegetation clearance
Corridor mapping
highway corridor mapping
Archeology
35
Advantages of airborne LiDAR
where and why to use it

Direct measurement - no image matching or stereo pairs needed

Works in non-ideal lighting – LIDAR is self-illuminated and offers more flying hours per
day
– Works at night
– Works under cloud cover

Ideal for featureless and ambiguously-featured terrain
– Forest floor extraction (100-1000 times denser data than DEMs extracted from
stereo photography)
– Snow
– Sand
Conclusions

LiDAR technology is
productive in many
applications relevant to
government at all levels

Airborne LiDAR provides a
highly detailed “big picture” in
3+ dimensions

The combination of 3-D and
temporal aspect of LiDAR
data make it an effective tool
for change management
Exercise 1: View LiDAR Data in IMAGINE
The ERDAS IMAGINE eWorkspace
File Button
Ribbon
Contents pane
Main
workspace
help
Dock or
undock views
2D View
Retriever pane
Status bar
40
Terminology: The Ribbon
Allows you to perform general tasks using
Groups: collapsed
Groups: expanded
Tabs
Commands
Can also
have menus
Menus
Terminology: Selecting Files
Tabs
Directory navigation
File selection
Type of file
Viewing LiDAR Images in IMAGINE
IMAGINE creates a raster of the point cloud,
Displays elevation in dark and light pixels
Grayscale
Relief
Exercise 2: Fill NoData Regions in the LiDAR
Grow AOI
Determine areas of continuous
spectral response
Select a sample pixel, a seed
Software evaluates DN values
of surrounding pixels
 Similar: included in region
 Dissimilar: not included
Graphic polygon drawn around
similar pixels
Similar pixel
values
Dissimilar values
Grow Properties
Controls which pixel values should be included in the
region
Constrains the search
region by area or distance
Controls which neighboring pixels
are considered contiguous
Controls the similarity. Set to 0 and
only pixels of same value as the
selected pixel are included in region
Use to regrow region if any
properties are changed
Exercise 3: Terrain Preparation Tools
Merging Terrain Files – Traditional Mosaic
Mosaicking terrain files with Mosaic Pro loses Vertical Datum
information
Vertical Datum is the how you are measuring the elevation; i.e.,
What is “Sea Level”
This information must be manually added back in
49
Terrain Prep Tool
The Terrain Prep Tool can be used to quickly Merge DTMs
Designed for merging terrain files
LAS, DTED, DWG, DXF, SRTM, LAS, 3D ASCII, IMG, 3D
Shapefiles all supported.
Maintains the ElevationInfo in the output file
50
Preprocess Settings
Thin Points: Used to remove redundant data points from the
terrain
Filter Points: Used to remove duplicate points from the
overlapping areas of data to be merged
51
Surface DTM
Creates 3D raster data out of vector, TIN, or point cloud datasets
Applies a surfacing method (interpolation of the data) across
NoData areas within the input dataset
52
Exercise 4: Change Detection with Model Maker
IMAGINE Model Maker
Graphical User Interface for creating and editing GIS and Image
Processing operations
Model is a set of instructions to create new images or data from
existing data sets
Uses the Spatial Modeler Language (SML)
Provides a graphical modeling tool known as the Model Maker
Spatial Modeler Language (SML)
Used by Model Maker to
execute operations
Used to write your own script
models
An SML script is temporarily
created during model
execution
Model Maker
Enables the user to “draw”
models using a palette of
tools
Inputs, Functions, Outputs
A model is essentially a flowchart defining:
 Inputs
 Functions
 Outputs





Image
Vector
Matrix
Table
Scalar
 Calculation
 Function
 Operation
LiDAR Change Detection
Uses Temporal Data (LiDAR from 2 different years) to calculate
the probability of change from one date to the next.
Probability that height increased (construction)
Probability that height decreased (demolition)
Exercise 5: Topographic Analysis Tools
Slope
The change in elevation over a certain distance
90
0
Aspect
NORTH
0 degrees
WEST
EAST
90 degrees
270 degrees
SOUTH
180 degrees
Painted Relief Images
Exercise 6: Dynamic 3D Viewing in VirtualGIS
Creating a Water Layer
Allows you to flood the entire terrain to a specified
elevation or flood individual areas of the terrain
Navigating a Flooded Area
Navigating around and under the flooded area provides an
understanding of the flood’s extent
Water Display Styles
Surface Style
 Solid
 Rippled
 Texture
Water Color
Reflections
Technology Preview
New - Support for Point Clouds
2012 Point Cloud Tools
69
Point Cloud Tools





View in 3D
See all point
attributes or Lidar
file attributes
Color by elevation,
class, file, returns,
RGB, intensity and
correlation (?)
Switch points on
and off by return
and/or class.
Perform Area and
point edits such as
Bias, delete or set
constant z.
2012 Point Cloud Tool



Auto Roam with pause
skip and speed control
along a corridor defined
by a vector
Simultaneously view the
profile across the
corridor and down the
corridor
With a single line
measure the linear
distance, slope, vertical
and horizontal length of
the segment
3D with roam by polyline with 2 profiles
– colored by class and elevation
2D, 3D and Profile
Measure tool in profile
58 Million RGB encoded points from XPro
Digitize areas and Change or assign a new class
View as Footprint, hill shade, painted relief or points
77
Viewed as elevation and intensity and linked
78
Editing Workflow
Coming Soon - New LAS modeling capabilities
QUESTIONS
81