COMPARATIVE ANALYSIS OF GEOREFERENCING
PROCEDURES USING VARIOUS SOURCES OF CONTROL DATA
A. Habib 2 , A. Jarvis 2 , A. P. Kersting 2 , Y. Alghamdi 2
d Department of Geomatics Engineering, University of Calgary, Calgary, Alberta, Canada T2N 1N4
habib@geomatics.ucalgary.ca - (amyjarvi, ana.kersting, yamalgha)@ucalgary.ca
Commission IV, WG IV/9
KEY WORDS: Digital Camera, Georeferencing, Orthoimage Generation, Object Reconstruction, Performance Analysis, Sensor
Fusion
ABSTRACT:
The increasing efficiency and affordability of modem spatial acquisition systems, such as LiDAR, have provided reliable alternative
sources of control data for image georeferencing. Indirect georeferencing using LiDAR control features to derive the exterior
orientation parameters (EOP) allows for a straight-forward integration of LiDAR and photogrammetric data which can be used, for
example, for more effective orthoimage generation and 3-D city modeling. This paper presents a comparative analysis of indirect
georeferencing by investigating the use of three sources of control data, namely, ground control points, LiDAR patches, and LiDAR
lines. In this regard, ground control points are obtained through GPS surveying and are used to derive the EOPs of the involved
imagery. Point features cannot be distinguished in LiDAR data however, and thus when using LiDAR as the source of control for
photogrammetric georeferencing, linear and areal features are used. Moreover, a semi-automated approach for the extraction of lines
and patches from LiDAR data is presented. Linear and areal features are identified in the imagery while conjugate LiDAR features
are extracted through planar patch segmentation and intersection. Following the extraction procedure, the mathematical models for
incorporating the image and LiDAR features for georeferencing are discussed. RMSE analysis and orthoimage generation using the
three sets of EOP - obtained from GCPs, LiDAR patches, and LiDAR lines - are used to quantitatively and qualitatively compare
the object reconstruction results using the three sources of control data. Finally, the practicality issues of these methods are
compared in terms of the effort involved in the derivation of control information and the underlying procedures.
1. INTRODUCTION
Photogrammetric reconstruction procedures aim at deriving
three-dimensional information from two-dimensional images.
These procedures create surfaces that are rich in semantic
information, which can be clearly recognized in the captured
imagery, and provide highly accurate results due to the inherent
redundancy associated with photogrammetric reconstruction.
Many photogrammetric applications, such as mapping and
orthoimage production, require the interior and exterior
orientation parameters of the imaging sensor. The interior
orientation parameters (IOP), which include the principal point
coordinates, focal length, and distortion parameters, can be
derived from a camera calibration procedure or obtained from
the camera manufacturer. The exterior orientation parameters
(EOP), which include the position and attitude of the camera at
the time of exposure with respect to object space coordinates
frame, can be derived either through indirect georeferencing
using ground control information or direct georeferencing when
GPS/INS is available onboard the imaging platform. The
quality of the reconstructed surface is affected by the accuracy
of the IOP and EOP, and thus it is essential that these
parameters be determined to a high degree of accuracy.
In direct georeferencing, the IMU body frame attitude and the
GPS phase centre position are directly measured using onboard
GPS/INS systems. In addition, the IMU boresighting angles and
the GPS antenna offsets relative to the camera perspective
centre are computed to determine the position (Xo, Yo, Zo) and
attitude (co, 9, k) of the camera at the time of exposure with
respect to object space coordinates frame. This approach is
computationally efficient since only an intersection procedure is
required, as well as economical in the long run since no ground
control is required. Nonetheless, indirect georeferencing is
traditionally favourable due to its accuracy and robustness
against IOP biases (Cramer et al., 2000). In indirect
georeferencing, the EOPs are determined indirectly using
ground control, where the most common type of control
involves the use of ground control points obtained through field
surveying procedures. This form of ground control can be
costly in terms of the required time and effort but has proven to
be an accurate source of control for the georeferencing
procedure.
With the recent developments in spatial acquisition systems,
however, some reliable alternative forms of control are
becoming available. In this regard, LiDAR scanning is rapidly
taking its place in the mapping industry as a fast and cost-
effective 3-D data acquisition technology for capturing accurate
positional information from physical surfaces. LiDAR
georeferencing is directly established through the GPS/INS
components of the LiDAR system. The increased accuracy and
affordability of GPS/INS systems are the main reasons behind
the expanding adoption of LiDAR systems. The use of LiDAR
derived control has two main advantages. First, it allows for a
straight-forward integration of LiDAR and photogrammetric
data, which has numerous benefits such as effective production
of 3-D city modeling and orthoimage generation. In addition,
the use of LiDAR control features can eliminate the need for
ground control points which simplifies the indirect
georeferencing procedure and makes it more affordable.
1147