SINGLE AND STEREO BASED 3D METROLOGY FROM HIGH-RESOLUTION
IMAGERY: METHODOLOGIES AND ACCURACIES
Arie Croitoru?, Yong Hu*, Vincent Tao*, Zhizhong Xu *, Feng Wang *, Peter Lenson "
? GeoICT Lab, York University, 4700 Keele Street, Toronto M3J 1P3 - (arie, yhu, tao, zxu)@yorku.ca
^ GeoTango International Corp., 4850 Keele Street, Toronto, M3J 3K1 - (fwang, plenson)@geotango.com
KEY WORDS: Three-dimensional, Metrology, Sensor, High-resolution, Imagery, Accuracy
ABSTRACT:
With the introduction of the Rational Function sensor Model (RFM) and its adoption as a primary sensor model for a verity of high-
resolution sensors, the volume of imagery for which REM data is made available is on the rise. This emerging trend is posing new
opportunities for fast and reliable 2D and 3D mapping and metrology using various high-resolution sensors, while offering users a
greater level of usability and interoperability. This paper examines and reviews the scheme for direct 3D information extraction from
high resolution satellite imagery based on the RFM, both in single and stereo environments, and will address the issue of the
attainable accuracy in each of these methods. Additionally, the methodology of 3D information extraction form high-resolution
single and stereo imagery (aerial and satellite) will be presented and discussed. This paper will also examine the application of the
RFM based 3D metrology scheme in a photogrammetric software environment and will provide real-world examples to the
applications of 3D metrology from high-resolution imagery.
1. INTRODUCTION
Detailed 3D information is a cornerstone in a variety of
applications. Nowadays, it is hard to imagine how military
simulations, urban planning, navigation, emergency response,
cellular communication, news and entertainment, or computer
games would maintain the same level of efficiency and
functionality without 3D spatial data. With the increased demand
and utilization for 3D information, a need for timely and accurate
metrology is also emerging. As various infrastructure systems
become dependent on 3D information, timely and accurate
measurement of a variety of dimensions such as heights, 3D
distances, slopes or angles, becomes an essential information
component. Yet, in spite this growing demand, the ability to
accommodate it is still highly impaired by the availability of
adequate processing methods, sensor models, calibration
information, GCPs and usability. The need to explore new data
sources and to provide direct 3D metrology is therefore evident.
One such data source that is becoming a dominant resource of
detailed 3D metrology in various applications, are high-resolution
satellite imagery (HRSI). Only a few years ago the HRSI was
available to a limited number of government and defense
agencies, which were utilizing such imagery using highly
sophisticated software and hardware tools. The notion of widely
available high-resolution satellite imagery that can be easily
exploited seemed to be an unlikely reality that would not be
realized in the near future. With the turn of the century this dream
is rapidly becoming a reality as the world of HRSI is evolving in
an unprecedented rate. Sub-meter satellite imagery is already
available from a variety of commercial satellite sensors, such as
the IKONOS?M and QuickBird™, in a range of formats and
processing levels and at an affordable price. These types of
sensors and their growing availability are revolutionizing the role
of HRSI in numerous applications ranging from intelligence to
insurance, media, marketing, agriculture, utilities, urban planning,
forestry, environmental monitoring, transportation and real estate.
The satellite data and services can be applied to almost any
industry. It is estimated by some analysts that the commercial
market for high-resolution satellite imagery will reach at least 30
to 40 Billion (USD) by 2005. These estimates are already being
revisited with the recent emphasis on homeland security in the US
and with the additional funds that are being allocated to NIMA for
the purpose of purchasing high-resolution imagery.
While the satellite imagery industry has made a quantum leap in
terms of resolution, data availability and quality, the available
tools and paradigms to process such data into valuable user
oriented information are lacking behind. Many of the available
tools for processing satellite imagery are beyond the reach of most
users: they still require a high level of technical expertise and are
usually complex. Furthermore the challenge of providing users
with robust and easy to use tools for extracting information from
satellite imagery is a challenge yet to be met in many of the
commercially available software tools. Although the high-
resolution satellite imagery industry has progressed considerably.
over the last decade in terms of resolution, quality and
availability, the available tools and methodologies for fast and
easy processing of high-resolution imagery still pose a major
barrier for most users.
One of the primary barriers to a wider adaptation and utilization
of satellite imagery was the sensor model. Sensor models are a
key component in restituting the functional relationships between
image space and object space, and are essential in image ortho-
rectification and stereo intersection. Physical sensor models are
rigorous and highly suitable for adjustment by analytical
triangulation and normally yield a high modeling accuracy (a
fraction of one pixel) Furthermore, in physical models,
parameters are statistically uncorrelated as each parameter has a
physical significance. Yet, from the user's point of view, the
utilization of a physical sensor model poses some difficulties. One
of the primary drawbacks of the physical sensor model is that its
application requires explicit understanding of each of the physical
parameters and a high level of expertise. Moreover, even with
complete understanding of the physical sensor model, users are
still faced with the challenging task of recovering the exterior
orientation of the sensor using a set of Ground Control Points
1022
Inter
(GCI
avail
sensi
colle
With
situa
parti
requi
requi
mod
impl
the a
Cons
map]
imag
resol
led t
the
inter:
RFM
block
mapr
Inspi
provi
the c
how
mode
view
auxil
econ
21
The |
betwe
point
Vice \
allow
absen
to-im;
the fo
where
(samp
norms
and p
jii, I
coeffi
term
permt
order
Imagi
standa
scenaı
depen
and H
