PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
October 2020
593
Question:
I came across one of your articles while searching for
some answers about breaklines, so I figured I’d contact you for
more information. The article I found is from September 2017 and
it is about using breaklines for lidar datasets. I’m working with a
surveyor and a drone pilot. The pilot flies, the surveyor does the
ground control setup, I process the data and the surveyor confirms
whether the product is accurate. Recently, the surveyor asked
me if I can provide him directly with a TIN (triangulated irregular
network). The problem that I am having is with breaklines. I can’t
find an industry standard or guidelines onwhen or where breaklines
are needed. I’m starting to wonder, aside from using them for
obvious things like gutters and roads and the sides of a cliff, if this
is as much of an art as a science. Would you be able to point me
in the right direction so I can learn more about what features need
breaklines and when we should and should not use them?
Nathan Mangsen, Mangsen Mapping
Dr. Abdullah:
First of all, it is always good to hear from a for-
mer student. As for your question, whether there is a need
for breaklines is a controversial issue among users within
the geospatial community. Breaklines were originally devel-
oped to ensure the accurate modeling of a terrain surface
where sparsely compiled mass points (3D points collected by
a stereo compiler using photogrammetric mapping princi-
ples) may result in the inaccurate representation of abrupt
changes in terrain. Before lidar and digital photogramme-
try, all topographic maps were created through a manual
process using stereo photogrammetry. This process was
time-consuming, expensive and did not always result in an
accurate terrain model, since it depended on the thorough-
ness of the compilers and the quality of their stereo vision.
The introduction of the breaklines concept was a genius
approach at the time because we could not afford to model
the terrain with a dense enough network of mass points, i.e.
collecting a mass point every one meter along the terrain,
to accurately depict all details on the ground. Over the last
two decades, lidar has alleviated the geospatial mapping
community’s main concern about the density of mass points
and its ability to accurately model the terrain. These days,
aerial lidar is collected at densities ranging from 2 points
per square meter (USGS QL2) to hundreds of points per
square meter, while a terrestrial or mobile lidar system
can acquire data with a density of thousands of points per
square meter. However, 30 years after the introduction
of lidar, some factions of the industry are still hooked on
the idea of collecting breaklines to augment dense lidar
data. This outdated practice continued even when the data
acquired using mobile lidar resulted in a point cloud with a
density of thousands of points per square meter. Breaklines
are used today in several other geospatial applications,
including hydro enforcement, transportation engineering
and to avoid some anomalies during the ortho-rectification
process. The need for breaklines in those applications can
be eliminated if software companies would devise solutions
based on artificial intelligence (AI), machine learning and
deep learning algorithms that utilize the computational
power of the processing machine, AI-based algorithms and
the richness of lidar data (that lidar includes points cloud,
“The introduction of the breaklines concept
was a genius approach at the time because we
could not afford to model the terrain with a dense
enough network of mass points.”
“The need for breaklines in those applications
can be eliminated if software companies would
devise solutions based on artificial intelligence
(AI), machine learning and deep learning
algorithms that utilize the computational power of
the processing machine, AI-based algorithms and
the richness of lidar data”
Photogrammetric Engineering & Remote Sensing
Vol. 86, No. 10, October 2020, pp. 593–594.
0099-1112/20/593–594
© 2020 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.86.10.593
