Planar Block Adjustment and
Orthorectification of ZY-3 Satellite Images
Taoyang Wang, Guo Zhang, Deren Li, Xinming Tang, Yonghua Jiang, Hongbo Pan, and Xiaoyong Zhu
APPLICATIONS PAPER
Abstract
For the problem of block adjustment for satellite images,
which cannot be solved under conditions of weak geomet-
ric convergence, this paper proposes a strategy that uses a
planar block adjustment method to solve the orientation
parameters of all satellite images, and then each satellite
image is orthorectified. This strategy can ensure both the
uniformity of the positioning accuracy and the strictness of
the relative positions of the adjacent orthoimages. Tests of
139 panchromatic nadir images from the ZY-3 satellite show
that by using only a small number of ground control points
(
GCPs
), whose plane accuracy is 5 m, the plane accuracy of
independent check points (
ICPs
) is better than 7 m after planar
block adjustment. This accuracy meets the requirements for
Chinese 1:50 000 topographic maps. Moreover, the precise-
ness of tie points (
TPs
) in adjacent images is better than 0.5
pixels, so a seamless level in mosaic geometry is attained.
Introduction
With the launch of ZY-3, the first Chinese civilian high-
resolution stereo mapping satellite, Chinese high-resolution
Earth observation capabilities have gradually been enhanced
(Gao
et al
., 2013). Following the integration of its mapping
and resource monitoring functionalities, image data from
ZY-3 will be used primarily for generation of cartographic
maps at 1:50 000 scale and revision of maps at 1:25 000
scale. The satellite carries two types of pushbroom imaging
sensors for acquisition of visible imagery: one multispectral
sensor and three panchromatic sensors that point forward
(
FWD
), backward (
BWD
), and towards the nadir (
NAD
). Detailed
payload information for ZY-3 is presented in Table 1. To meet
the needs of stereo mapping, the
FWD
and
BWD
sensors are
arranged at inclinations of ±23.5° from the nadir to realize a
base-to-height ratio (B/H) of 0.87.
In classical photogrammetric production, the orientation of
the images is a fundamental step, and its accuracy is crucial
in the evaluation of the entire processing system, from the ra-
diometric preprocessing of raw images to the generation of 2
D
and 3
D
georeferenced products (Poli and Toutin, 2012). Block
adjustment of satellite images is an important orientation
method for precise geometric positioning, especially when
there is a lack of
GCPs
. Currently, the block adjustment for
satellite images is based on the principle of light intersection
for the homonymous points in the overlapping area to build a
block network. In other words, the block adjustment of images
is based on the stereo model. At the same time, taking into
account that the satellite image model is a linear array push-
broom, which is quite different from a full frame, the geomet-
ric processing model must change to fit its geometric features.
So far, the precise geometric processing of high-resolution
satellite images (
HRSIs
) has mainly used multi-coverage images
for block adjustment based on various types of sensor models.
In the research on using rigorous models of
HRSI
for block
adjustment, Fritsch and Stallmann (2000) conducted a study
of
MOMS-2P/PRIRODA
images based on the orbital constraints
(Kratky, 1989), which resulted in high accuracy. Poli (2002
and 2005) proposed a piecewise polynomial model (
PPM
) and
its application to block adjustment for Three-Line Sensor
(
TLS
) images and
MOMS
-02 images, which provides sub-pixel
accuracy. Additional parameters (
APs
) modeling of all kinds
of distortion and displacement is used for self-calibration in
block adjustment. Toutin proposed a model for multi-source
remote sensing images combined with adjustment of
VIR
and
SAR images. Toutin’s model (
TM
) proved that slave block im-
ages laid only on elevation joint points can achieve almost the
same accuracy of adjustment as master block images laid on
plane and elevation
GCPs
. This model has been applied for dif-
ferent mainstream high-resolution (
HR
) sensors (Toutin, 2003,
2004a, 2004b, 2004c, 2006a, and 2006b). Hofman
et al
. (1984)
proposed a photogrammetric point determination for space-
borne three-line scanners (Ebner
et al
., 1992). The exterior
orientation parameters are estimated only for so-called orien-
tation images (
OI
), which are introduced at certain time inter-
vals. Weser
et al
. (2008) proposed a generic pushbroom sensor
model for
HRSIs
. The satellite orbit and attitudes are modeled
with splines using the observations provided in the imagery
metadata files. Fraser
et al
. (2011) used the sensor orientation
model of Weser
et al
. (2008) for long-strip adjustment of
ALOS
imagery. Aguilar
et al
. (2012a and 2012b) tested the rigorous
model and empirical model for very high-resolution satellite
images (
VHRSIs
) and analyzed the number of well-distributed
GCPs
in the triangulation process.
In the research on using empirical models of
HRSIs
for block
adjustment, the most representative of the empirical models
is the rational function model (
RFM
) (Tao and Hu, 2001). This
has been introduced by many investigators as a mathemati-
cal model for conventional image-to-ground transformations
of coordinate systems (Fraser
et al
., 2005). During a study of
block adjustment for Ikonos images, Grodecki and Dial (2003)
Taoyang Wang, Guo Zhang, Deren Li, Yonghua Jiang,
and Hongbo Pan are with the State Key Laboratory of
Information Engineering in Surveying, Mapping and Remote
Sensing (LIESMARS), Wuhan University, 129 Luoyu Road,
Wuhan, 430079, P. R. China (
Xinming Tang and Xiaoyong Zhu are with the Satellite Sur-
veying and Mapping Application Center (SASMAC), National
Administration of Surveying, Mapping and Geoinformation,
Airport East Road, Shunyi District, Beijing, 101300, P. R. China.
Photogrammetric Engineering & Remote Sensing
Vol. 80, No. 5, June 2014, pp. 559–570.
0099-1112/14/8006–559
© 2014 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.80.6.559
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
June 2014
559
