Professional Documents
Culture Documents
. --.
. . . .. -
Validat ion of
Data Void and
Data Density
Las0040 10 points 16.82761 582169 514477 75%
Las0041 10 points 21.13412 582169 580270 75%
Las0042 10 points 25.32909 582169 580145 75%
Las0043 10 points 22.2032 582169 580664 75%
Las0044 10 points 22.25405 582169 580559 75%
Las0045 10 points 21.7628 582169 580268 75%
Las0046 10 points 20.79796 582169 580653 75%
Las0047 10 points 29.88647 582169 578288 75%
Las0048 10 points 29.69578 582169 574860 75%
Las0051 10 points 21.05113 582169 577789 75%
Las0052 10 points 20.65534 582169 580461 75%
Las0053 10 points 20.7103 582169 578484 75%
Table 2: Data Density and Data Vold valldaUon results
in each grid. For example, if the require-
ments of the project are the following:
J. The point density should be more
than 10 points per square meter.
2. 75% of the lm X lm grid should
have more than 10 points.
In this case, a grid of 1m X 1m has been
created and it is expect<.'<! that each grid
should have a minimum of 10 point<; per
square meter. For a given .LAS file the total
num))(!r of 1m X lm grid is mmputed at
the beginning. During the program, each
grid will be tested for the number of points
meeting the specific criteria (namely 10
points). At the end, the program calculates
the following percentage:
Percentage of grids that meets the
spec = nurnber of grids meeting
the percentage criteria I total
number of grids in the LAS .files
If the percentage is above the
specified percentage, then the particular
LAS file has met the data void tcsl.
Otherwise, the particular LAS file has
failed the data void test.
The block diagram for validating the
data void and point density is shown in
Figure 1 and the sample of the valida-
tion of data densi ty and data void results
are shown in Table 2.
Dr. Srinivasan "Srini " Dharmapurl has
over 26 years of extensive, wide-ranging
experience within the Geospatiallndustry;
most notably with LiDAR, Photogrammetry,
and GIS. He has worked ln both the private
and public sectors. as well as Internationally.
Currently he ls working as a LiDAR Sclentlst
with Michael Baker Jr .. Inc.
88.37 Pass
99.67 Pass
99.65 Pass
99.74 Pass
99.72 Pass
99.67 Pass
99.74 Pass
99.33 Pass
98.74 Pass
99.25 Pass
99.71 Pass
99.37 Pass
REFERENCES
FEMA's Memorandum for Regional
Risk Analysis Branch Chiefs,
Procedure Memorandum No. 61:
Standards for LiDAR and Other High
Quality Digital Topography, Effective
Date September 27, 2010.
USGS-LiDAR Base Specifications
Version 1.0- Techniques and
Methods ll-B4.
American Society for
Photogrammetry and Remote Sensing
(ASPRS), ASPRS Guidelines, Vertical
Accuracy Reporting for LiDAR Data,
vers. 1.0, May 24, 2004.
2013 VOL. 3 NO. 4 Li DAR 57
INDUSTRY PIONEERS
Doug Flint
Roots
uestion-when you were
growing up how often did
you take advice from your
big sister? Never, in my case, I didn't
have a big sister, but in Doug Flint's
case it started him on a LiDAR industry
pioneering journey and eventually
to his current position as the Vice
President-Business Development at the
Geospatial Solutions Division of Leica
Geosystems. Doug is based in Westford,
MAin the same facil ity that produced
the company's first high performance
airborne LiDAR 15 years ago. Let's see
how he got there.
Doug grew up in an engineering
environment; his father was an electri-
cal/ mechanical engineer. As a kid
Doug was always fixing stuff around
the house-from bicycles to washing
machines- and working hard at math
and science in school! Doug remembers,
"Somehow I was looking for something
beyond the boundaries of electrical and
mechanical engineering when my sister
happened to interview at the University
of
He conti nues, "TI1ey asked her if she
had any siblings considering colleges
and what their interests were. She told
them she had a brother interested in
engineering and was presented with
a course catalog that included their
unique Institute of Optics, Optics
Engineering degree program. I was
hooked, applied and was accepted "early
decision': I stayed on for an extra year
58 Li DAR 2013 VOL. 3 NO. 4
BYGENEROE
of graduate school after earning my
B.S. and picked up an M.S. in Optics
Engineering with a specialization in
optical thin-film coatings. "
Early Career
From there Doug headed for the high
tech capital of its time, Route 128 and his
first position as, you guessed it, an Optical
Coating Engineer at the ltek Corporation
in Lexington, Massachusetts. ftek special-
ized in large and/or complex optical
imaging systems such as reconnaissance
cameras, satellite payloads and even the
Mars Rover Imager. Doug recalls, "One
of my projects there was a 6-ft
protected aluminum reflective coating for
imaging down to the Lyman Alpha (121.6
nm) This was a proof-of-
concept coating for the now famous
Hubble Space Telescope.
After only a couple of years, Doug
was made manager of the group. He
notes, "1 had to go from designing and
fabricating the coati ngs to generating
proposals in response to RFP's, calculat-
ing costs and prices, and managing
our R&D projects:' Gradually Doug
found himself enjoying the challenges
of management, sales and marketing a
little more than engineering. In fact, he
earned an MBA part-time (nights) from
Bost on University as evidence of his
new interests.
ln 1979 Doug joined Honeywell in
their Electro-Optics Division where
he worked on business development
for NASA and DoD projects. This is
where he first met Ron Roth, a coUeague
that would eventually become quite
influential in Doug's career.
Entrepreneurial Period
In 1983, Doug began the entreprenewial
'start-up company' phase of his c:u'eer.
This is where he had to put all of his
past experience to work- engineering,
production, sales, support, people skills,
management and even accounting-in
"survival" rather than "paycheck" mode.
Doug recalls, "1 went from nmning the
U.S. sales and support operation of a suc-
cessful Israeli high-tech printed circuit
board inspection equipment start-up, to
a thermal properties analytical instru-
mentation venture, to buying out a small
group (2 engineers) specializing in long
distance laser range finders that could
measure distances out to 18 tniles."
This became Azimuth Corporation,
founded in l991 and soon included
co-pioneers Bob Eaton (electronic), Ron
Roth (systems I business), Scott Bender
(mechanical) and Paul Galla (software
<tnd eventual creator of the LAS format)
who are all still his colleagues today.
They are the technology heroes of this
story. Doug commented, "1l1e best
compliment that Azimuth routinely
received from various organizations was
that they were shocked to find out how
smalJ we were. Our business practi ces
system didn't work Azimuth would
blame their workflow and they would
blame Azimuth's sensor! Ron Roth hit
the textbooks to brush up on WGS84
geoids, atmospheric corrections and
other trajectory and system factors to
figure out where a scanned laser pulse
from a moving aircraft is hitting the
swface (not an easy feat). ''Eight mouths
later, we had the first test flight-anc:l
the first successful data collection from
the original AeroScan airborne LiDAR
system;' Doug proudly recalls. What was
What we are seeing now in 'napping, topo.
constructlon, Infrastructure, plant and mobile
scanning advances is the tip of the Iceberg. '
and products reflected those of much
larger companies:
Without any external funding,
Azimuth developed laser rangers and
profilers that measured ice deformations
on the order of 2-in. caused by volcanoes
4,000 feel under Antarctic ice, fired
braking rockets on parachute-dropped
military cargo pallets and have helped
track countless airborne and ground-
based vehicles on military test ranges. ln
1997, Earth Data (now a part of Fugro)
asked A:z.imuth to develop a high-altitude
LiDAR system to fly with imaging
systems taking USGS DOQQ images at
a 6000m flying height and 75 deg FOV.
Doug n(ltes, "No other vendor had that
capability, so we won the contract:'
Doug's team assumed that they
would do the sensor and Earth Data
would do the software. 1l1ey assumed
\Vrong; EarthData feared that if the
then a 15 kilopulse-per-second system
has developed into the 500 kiJopulse-per-
second ALS70 LiDAR product family
some 15 years later.
By the year 2000, Azimuth was
manufacturing four systems per year
and selli ng them worldwide. They began
to get visits from companies interested
in a "partne1ship" and they selected LH
Systems (nowLeica Geosysterns) as
t heir new owner in 2001. "We even had
a brush with terrestrial laser scanner
development via a visit from Ben Kacyra
(Cyra Technology) looking to select a
laser ranging technology; recalls Doug.
Cyra and ERDAS were also acquired by
Leica Geosystems all within a three-
month period. Doug notes, "Now after
U years with Leica Geosyslems, the
original, small core team is still together
in Westford, MA-still setting the pace
in airborne LiDAR technology:'
Ret rospective
Doug o t s ~ T have had the great
fortune of having many bosses that were
unrelentingly driven to achieve, succeed
and generally not let failure get in the
way of their goals. From them l have
learned good marketing skills in high-
tech, high-dollar optical t:apital equip-
ment markets-both commercial and
aerospace. Most impor.tantly, I learned
from each of these mentors how to think
strategically and tactically for product
development and market exploitation.
1l1ere are sti ll not many organizations
that consistently define their corporate
vision and then devise the strategy,
roadmaps, goals & objectives, tactics
and operations to successfully pursue
that vision:'
Concerning the current state of the
industry Doug bel ieves that airborne
and terrestrial laser scanning technology
doubles its hardware performance
approximately every two years. That
now the real race is in integrated work-
flows, fused sensor data and information
management and distribution. It's all
about end-to-end 3D-reality capture,
reality modeling and visualization.
BIM, indoor and mobile mapping, and
photogrammetry I machine vision join
LiDAR as the hot technologies and
markets chased by the industry today.
"The problem with cutting-edge,
high-tech systems is they create a
constant stream of development
challenges. Anyone in this industry
occasionally longs to be a bartender,
or some less stressful profession. I love
laser scanning, but competing in lower
complexity product markets could add
years to my life;' comments Doug.
Doug notes that Google Earth and Bing
have brought 3D and maps to everyone.
This has greatly accelerated the demand
2013 VOL. 3 NO. 4 LADAR 59
for indoor and outdoor 3D solutions and
content. Cheap memory and servers I
distributed processing I cloud computing
I and internet distribution are just some
of the computing capability advances that
have led to the industry's integrated work-
flows and supercomputer-like perfor-
mance on yom laptop. They have brought
incredible data acquisition and processing
capabilities that work for and with LiDAR
technologies- think wavetorm digitiza-
tion and photogrammetrically-derived
info douds, for example.
Doug reminds us, "One of the
original drivers of the ai.Jborne UDAR
market was the fact that elevation model
production from photogrammetrlc
point measurements was too slow and
costly with the computer and software
BENTLEY SYSTEMS 27
wwvt.bentley comllldornews
CARDINAL SYSTEMS 51
www.cmdtnj!Jsystems netl
CERTAINTY 3D 35
www.rerlaln1y3d.f!)fl1
ESRI 37
www.esr <omlllrtarn;ws
EN VI 29
www.exeltsvts.comlflsrt-UC
GEOCUE 3
www.geocue.lom
GEODIGITAL 41
'Om
60 UDAR 2013 VOL. 3 NO. 4
technologies available a decade ago.
1he cost of IT capability has dropped so
radically that so-called dense matching
has made real inroads:
On the issue of pressing needs Doug
comments, "'When younger generations,
who are brought up on 30 visualization,
begin to take charge, this will spur
acceptance of the innovations that are
occurring now and in the future. The
major barrier to our industry is 'resistance
to change' and a slow process to adopt the
new capabitities coming on line:'
Looking to the future, Doug notes,
"Depending how you count, we are
about 17 years into commercial airborne
LiDAR and 13 years into HDS scanning.
What we are seeing now in mapping,
topo, construction, infrastructme, plant
GEOINT2013 17
www.geolnl 2013.com
!CUBED 13
www .. J.com/litlm
INTERGEO 2013 31
www.lntergeo de
INTERGRAPH 23
geospa!ial rntergraph.com/20 13
LEICA GEOSYSTEMS 43
hds.Jerca-geostslems.corn/p20
MAPTEK 7
www.maptek.com
MICROSOFT 25
www !JIItaCamEagJe com
and mobile sc:anning advances is the
lip of the i<.:eberg. We havt.' sensors,
workflow, visualization and distribution
to work on, to name just a few challenges.
We will be amazed by what we will see
and do on/with our screens, laptops,
tablets. wrist tops and Gough:! Glass
by 2020. And that is just regaJding the
applications we have thought of aheady!"
As you can imagine Doug does not
have a lot of free time, but when he does
he loves to ski-both snow and water.
He dreams of what it might be Like to
be an artist or a musician. Imagine if his
sister had tried to steer him in one of
those directions ...
Gene Roe is the Managing Editor and
Co-Founder of UDAR Magazine.
NEPTEC
www neptec.com
NOR PIX
ww.,, norptx.com
OPTECH
www opterh.<om
RIEGL USA
WWW rieQillSil COfll
TEXTRON SYSTEMS
TOPCON
www UlpconooSitlonlng.comlmapplng
TRIMBLE
wvM rlmCIIP.tom/3D :cilnnlng
IBC
47
9
21
5
IFC
BC
Generalized Data
I
n the last Random Points arlkk, I
discussed the overview concepts of
data and sensor fusion. I promised
at the conclusion to continue with a
fusion technique I call "Just in Time"
(JIT) product generation. However, I
thought it would a useful exercise to first
discuss data generalization.
In the mid-1980's 1 had the privilege
of participating in a massive Defense
Mapping Agency (DMA) development
called the Mark 90 program. TI1e DMA
was producing a wide range of mapping
products (e.g. Nautical Charts, Harbor
and Approach Charts, Topographic
Line Maps and more) essentially as
one-off products. Data were collected
from a wide variety of sources to
create thematic map separates. These
separates were transferred to film at a
one-to-one {relative to the printed map
size) scale and used in printing physical
maps; there were virtually no digital
maps at this time. You can imagine
how difficult and time consumi ng this
process could be. Additional ly, the
DMA had to stockpile literally pallets of
maps in warehouses for dissemination
to the user communi ty. If a change was
deemed necessary, the map had to either
be reprinted or a "red-line" separate
produced for overprinting of the map.
The DMA embarked on a huge series
of modernization programs aimed at
creati ng a "map factory'' that could
produce maps in a fiT fashion. The idea
was to build a world-wide database of
base information and use this database
Product
Request
1! 11 Mop P100Ud GeoeaUOil
in a series of software processes to
create map products. The database
was called the Mapping, Charting and
Geodesy (MC&G) database, reflecting
its wide purpose.
Imagine the machinery of such a
system. Given a multitude of data
sources, the mapping algorithms must
pull together those sources needed to
create a specific product, accurately
combine them, create labels, render
color schemes, and so forth. Just a few
of the technologies involved include
conflation, conflict detection and resolu-
tion, label placements, names resolution
and generalization. The general idea is
depicted in Figure 1.
1his is an immensely complex task that
even w this day (nearly 30 years later) has
Products
not been fully accomplished. However,
the various attempts (beginning with the
DMA programs) have created thousands
of algorithms that take us ever closer to
the generalized solution.
All of this is directly related to some
difficult problems in representing
the real world in point clouds. \Y/e
are patticularly interested in the
"generalization" problem-that is, how
do we accurately represent data as we
"zoom out" on that data? Consider a
Large scale map where a highway runs
very close to a parallel railroad. As we
zoom out on the map, the road and
railroad would normally merge into a
single line. However, if the desired map
product specified that these two featwes
must remain distinct, a cartographic
2013 VOL. 3 NO. 4 UcAR 61
displacement must occur as smaller
scale products are produced- this is the
process of generalization.
Consider the point doud "zoom out"
problem. To represent features in a point
cloud, the data must be at least twice
as dense as tht! smaUt!St dt!sired feature
(this is the Nyquist criteria to which
T have referred in p1'evious articles).
Fot example, if you want to be able to
measure a 4 em diameter walk button
at a crosswalk, the LIDAR data point
spacing must be 2 em o1closer. As one
zooms out on a view, more and more
data must be accessed to render the view.
For example, if the view volume contains
an area 5 meters high, 20 meters wide
and 10 meters deep, the view volume
encompassed is 1,000 m
3
! At a point den-
sity of 1 point per 8 cm
1
(one poi nt per
cube measuring 2 em on a side), we have
125.000 points per cubic meter. Thus we
would need to access 125 million points
to render the scene of our view volume.
For t his viewing to scale, the data must
be decimated. We routinely do this in
image processing by low pass filtering
the dat a and then decimating. This is a
natural way of decimating images since
this is what happenS in an optical system
such as o ur eyes.
However, we have adopted a system
of tagging points within a point cloud,
anointing each point with "i ntelligent"
content. For exampl e, a single point may
be tagged as a "Vertical Obstruction"
such as the top of a radio tower. lf we
were to simply treat the point data as we
do an image. low pass filtering and deci
mating, we most li kely would lose this
vert ical obstruction point. Tn most uses
of the data, this would be completely
unacceptable. Enter Cartographic
Generalization (or a similar technique
that we apply to point clouds).
62 UDAR 2013 VOL. 3 NO. 4
FiSh- 2: Tw n ultl \J I ur 1 rr
A search of the literatwe will reveal
very tittle on the topic of point cloud
generalization. 11ms we seem to be at
the forefront of this technology that
will prove critical as point douds begin
t o replace other data sources for base
intelligence sources.
There has been some initial work on
a very specific case of generalization-
t errain thinning. Consider a ground
model comprising point cloud data
with a nominal point spacing (NPS) of
50 em. This would yield an average of 8
triangles per ~ when the point cloud
of the ground surface were rendered
as a triangulated irregular network
(TIN)- see Figure2.
Most CAD software would be
overwhel med ifit were faced with load-
ing a TIN of such high density. For this
reason, thinning algorithms are included
with most LIDAR processing software.
'Ibis thinning operation for ground data
is typically called a "Model Key Point"
algorithm. Ln fact. the LAS point doud
specification includes a special bit flag to
mark points as "MKP" points.
This thinning method functions as a
simple error bracketing algorithm. 1l1e
user specifies the maximum vertical
error that is permissible for the model
(treating the original data as "truth").
The o riginal data are modeled as a TIN
and points are iteratively removed. As
each poi nt is removed, the modified
TIN is generated and compared (vert i-
cally) to the original TfN. lf the error
i.s below the user specified threshold,
the point i s left out of the model and
the next point ls removed. If, on the
other hand, the model deviates from
"truth" by more than the user specified
vertical limit, the test point is restored
to Lhe model.
This process repeats until no further
points can be removed from the model
without violating the accuracy criteria.
The remaining points are fl agged
as Model Key Points. It should be
intuitively obvious that very few MKPs
will be requi red in flat terrain whereas
a fairly dense coUection of MKPs
will be required where the terrain
sharply changes. This is illustrated in
figure 3 where the nodes of the TIN
represent the model key points and all
points (orange points and TIN nodes)
represent the ground.
So now the question is "how do
we extend this MKP idea to other
features?" Can we flag important points
(li terall y) in the data set and preserve
these poi nt features as we generali ze
the data? Obviously this requires
a new class of algorithm. Much
can be learned from the old DMA
generalization efforts as well as other
developments in these areas.
This is a very important preamble
to the idea of JIT product generation
from LIDAR data. Lfwe are to store
LIDAR data at full resolution and
then generate ad hoc products (using
server-side processing, for example),
we will have to develop intelligent (and
fast!) algorithms that can perform the
requisite processing. We will pick up
on this theme in the next edition of
Random D
Lewis Graham Is the President and CTO
of GeoCue Corporation. GeoCue is North
America's largest supplier of LIDAR production
and workflow tools and consulting services for
airborne and mobile laser scanning.
Figure 3: MCJdel Key Poln :; (nodes of TI N) Ground Points (noces of Tl"' ana all other points)
SmerzJ continued from page 64
TED Talks
tfyou're not familiar with Ted Talks
let me introduce you to one of the
best video platforms on the planet for
inspiring thought and action. lt's a col-
lection of mostly common individuals
havi ng completely uncommon lectures
on an incredibly wide variety of topics,
including our 30 technology. One of
the most basic reasons this site is so
immensely popular and videos are so
often viewed is because all talk
about \XTHY. Each video has a message
that is designed to somehow enrich the
viewer and provide you with a message
about WHY. And as such, their videos
are viewed worldwide. [Check out
Simon Sinek's
WHY should you care?
So how does this translate into yow-
business? lt is going to be easier for you
to recruit and retain the best employees
if they understand your company's
WHY. There will be a much larger
blue ocean of market opportuni ty in
different market segments if you can
explain your WHY. It will be far
to set goals, strategies, and tactical
plans in place if you know your WHY.
Simply put, it is going to be much easier
for you to enj oy a bottom line profit if
you are starting with WHY. Offer your
clients this clear understanding. An
inspired client is far less price sensitive
and much more loyal over time.
So, what's your WHY?
Ken Smerz Is the President of Precision 3D
('Afl{ol't!' a service provider
that travels throughout the nation working
with AIE/C and forensic clients. He can be
reached at kennpreclslon3d.com with any
questions or comments you might have.
2013 VOL. 3 NO. 4 Li DAR 63
Something is Missing
Talking Heads
I
've been to a variety of conferences
lately, both within the 30 imaging
industry as well as other business
evt:nts. I've had the privilege of listening
to some very bright minds and overaJI
successful people share their vision
and opinions on a variety of subjects.
And aside from the predictable self-
aggrandizing that commonly occurs,
especially in the world of 3d, there have
been some valuable messages. But I find
too often there is something missing.
ln conunon with the majority of these
presentations is the lack of WHY in the
message. \.Vhy were they successful and
more importantly WHY should I care? So
let me introduct: you to Simon Sinek who's
hook (Start With Why: How Great Leaders
Inspire Everyone to Take Action {2000))
delves into what he says is a naturaJly
occurring pattern that explains why we
are inspired by some people, leaders,
messages and organizations versus others.
And why some simply fail even though
they have a good plan -a benevolent
objective. To quote Sinek, " ... people don't
buy what you do, they buy why you do it:'
I strongly recommend his book
for a variety of reasons for anyone in
business. One of the examples he uses
to illustrate how important the \VHY
is comes from the story of the fa mous
explorer Edmund Shackleton who in
1914led an expedition to Antarctica
against truly death defying odds in a
64 UDAR 2013 VOL. 3 NO. 4
wooden sailing ship. The story tells of
the shipwrecked crew who together
survived months floating on ice caps.
They were successful and nobody died
because of the composition of the crew
and their personalities. Shackleton ran
an ad in the paper before the mission to
recruit the crew with the proper mind-
set. The ad specifically didn't say what
the mission was, but stressed WHY they
wert: doing it. Tt read in part, " ... honour
and recognition in case of succes..,:'
3D imaging world? J.n that last sentence
substitute the word "measurement" in
place of 'cancer treatment' and it's the
common speech you've heard.]
The doctor continues about how
frustrated he is because while they're one
of the leading organizations in the world,
they fight desperately to compete for
medical research dollars and recognition
with the highly promoted Banner MD
Anderson Cancer Center. Immediately
in my head l envision the signage
'' ... people don't buy what you do,
they buy why you do it.''
Lost Doctor
As an example, I'm sitting in a presenta-
tion by a research scientLo;t from the
University of Arizona who spends
the next 37 minutes explaining to the
audience what they are doing to find a
cure for cancer. Incredibly knowledge-
able, the doctor cites statistics and
shows case study data on their success
rate; discusses cutting edge technology;
and paints a great vision for the future
with his methods that will revolutionize
cancer treatment. [Sound a little like the
that Ba1mer has posted on bill boatds
throughout Arizona that shows their
logo, that has a giant red stripe through
the word c a n c e r ~ Banner is effectively
telling me WHY I should care and
support them even in their logo, while
this doctor is telling me how great he is at
1esearch. At no time during the pre-
sentation I sat through did the scientist
indicate WHY 1 should care about his
cause. Think how much better it COltld
have been if he'd told me WHY I should
support him over Banner Medical.
continued on page 63
Technologies Corp
Visit us at AUVSI's Unmanned Systems 2013
www.neptec.com
Increase Productivity While Capturing High Resolution Data
fiilJDt, ltimblt litl'll lhn\bL ur """"""' 'b ofrnmbl. ,_._,Vl,J.-IblAllrnlf,d
tt:t )'St .. md'lncrtl" II'!UUlllt'\ AJI1 J 0G( l'.l(ll)({l'i.11i)
When creating high quality deliverables
for engineers, architects or facility owners,
you need speed and confidence in the
data you are capturing. Trimble 30
scanning solutions provide your business
with the tools needed to exceed client
expectations and maintain a competitive
advantage. Trimble 30 scanning
technology allows you to reduce field time
and costs while improving the quality of
deliverables you provide to your clients.
learn more at:
www.trimble.com/3 D-laser-scanning