ISPRS Commission II, Vol.34, Part 3A „Photogrammetric Computer Vision“, Graz, 2002
NEAR REAL-TIME ROAD CENTERLINE EXTRACTION
C. K. Toth! and D. A. Grejner-Brzezinska?
Center for Mapping, The Ohio State University, 1216 Kinnear Road, Columbus, OH 43212-1154, USA
Department of Civil and Environmental Engineering and Geodetic Science’, The Ohio State University
toth@cfm.ohio-state.edu
Commission III
KEY WORDS: Mobile Mapping, Image Sequence Processing, Direct Sensor Orientation, Real-Time Systems
ABSTRACT:
In this paper a new GPS/INS/CCD integrated system for precise monitoring of highway center and edge lines is presented. The
system has been developed at The Ohio State University (OSU) for the Ohio Department of Transportation (ODOT). The positioning
component of the system is based on tightly integrated GPS/INS (dual frequency GPS receiver and a high-accuracy strapdown INS),
and the imaging component comprises a fast, color digital camera from Pulnix (TMC-6700, based on 644 by 482 CCD, with the
acquisition rate up to 30 Hz), installed in a down-looking position. The high image rate provides sufficient overlap of the subsequent
images at reduced highway speed. The stereo image data processing is supported in near real-time by on-the-fly navigation solution.
In this paper, we discuss the design, algorithmic solution and operational aspects, as well as the calibration and performance analysis
of the developed system. Feasibility of the application of real-time navigation data to on-the-fly image processing is also presented.
In particular, a performance analysis of the integrated system, based on reference ground control, is discussed.
1. INTRODUCTION
Mobile Mapping Systems (MMS) have been developed since
the early 1990s with a primary focus on the acquisition of the
street environment data, i.e., man-made features and their
attributes along the road corridor, as well as the topography.
Over the years, MMS has evolved from a rather simple, low
to modest accuracy mapping system, to modern state-of-the-
art multisensor systems, incorporating an increasing amount
of real-time operations. Mobile computing and wireless
communication are considered two of the strongest trends in
the modern computer industry. The proliferation of mobile
computer and wireless technology used in modern MMS,
combined with multiple, high resolution digital imaging
sensors, bring fundamental changes to the ways the
geoinformation data are acquired and analyzed: the data can
be analyzed on-the-fly, and transferred to the data centers,
where they can be transformed to the intelligent
georeferenced information, and subsequently distributed to
the users.
The MMS presented in this paper, although classified as real
time, does not fully follow the paradigm of mobile computing
outlined above. The data are not transferred to the data
analysis center, but rather part of the data processing is
performed during the data collection, in real-time, by the
onboard computer. Since the system is designed for mapping
of center and edge lines of the highways, the instantaneous
data transfer is not crucial. The real-time image processing is
designed to limit the amount of data stored for further
processing. In particular, the linear features can be extracted
and tracked from the imagery on-the-fly, using real-time
navigation information, which can effectively support the
formation of stereo-pairs. Therefore, the real-time part of the
image processing is only concerned with the relative
orientation (RO). Tthe final processing can be done in post-
A - 362
mission mode when more precise navigation data become
available. In this paper, a discussion related to the recently
performed tests demonstrating the achievable accuracy in real
time is included, while more details on the system design and
the concept of real-time processing can be found in (Toth and
Grejner-Brzezinska, 2001a and 2001b; Grejner-Brzezinska,
Toth and Yi, 2001; Grejner-Brzezinska, Yi and Toth, 2001;
Grejner-Brzezinska and Toth, 2002).
2. SYSTEM DESIGN AND IMPLEMENTATION
The positioning module of this system is based on a tight
integration of dual frequency differential GPS phases and raw
IMU data provided by a medium-accuracy and
high-reliability strapdown Litton LN-100 inertial navigation
system. LN-100 is based on Zero-lock™ Laser Gyro (ZLG™)
and A-4 accelerometer triad (0.8 nmi/h CEP, gyro bias —
0.003°/h, accelerometer bias — 25ug). An optimal 21-state
centralized Kalman filter estimates errors in position, velocity,
and attitude, as well as the errors in the inertial sensors. In
addition, the basic 21-state vector can be augmented by the
additional states representing GPS differential ionospheric
correction terms, which are estimated (per satellite pair, as
double difference mode is used) when the base-rover
separation exceeds 10 km distance. The primary filter design
follows the concept of AIMS™ (Grejner-Brzezinska et al.,
1998; Toth and Grejner-Brzezinska, 1998), developed earlier,
which has been modified and extended to accommodate
needs of precision navigation in urban environments. These
augmentations primarily include the implementation of the
static INS calibration (ZUPT mode) and the extension of the
measurement update module to include the pseudolite data
(Grejner-Brzezinska et al, 2002), as well as further
processing optimization. Under favorable GPS constellation
(minimum of 5-6 satellites), the estimated standard deviations