On-Orbit Geometric Calibration of the
Panchromatic/Multispectral Camera
of the ZY-1 02C Satellite based on
Public Geographic Data
Pengjie Tao, Luping Lu, Yong Zhang, Biao Xu, and Songbai Zou
Abstract
This paper describes the theory and method of on-orbit
geometric calibration of the panchromatic/multispectral (
P/
MS
) camera of the ZY-1 02C satellite. Instead of using a high
quality test field specifically designed for camera calibration
purpose, this study intends to make a use of readily available,
public geographic data as the reference. Such public data
may include Google Earth™ images, Bing Map images,
SRTM
DEM
, and
ASTER GDEM
. For comparison purpose, the standard
national digital orthophoto maps (
DOM
) and digital elevation
models (
DEM
) are also utilized. To carry out the calibration,
ground control points (
GCPs
) are selected by matching the sat-
ellite images to the reference data. It is shown that the calibra-
tion corrects large amounts of systematic errors caused by sig-
nificant lens distortion and poor direct georeferencing. After
calibration, the interior accuracy of the images is improved
from a range of six to eight pixels to sub-pixel level, and the
direct georeferencing accuracy is increased from 1.8 km to
115 m. The study demonstrates the necessity of camera cali-
bration and the value of public geographic data for this need.
Introduction
The ZY-1 02C satellite was launched by the CZ-4B Y15
rocket from the Taiyuan Satellite Launch Center of China
on 22 December 2011. This satellite fills a gap in Chinese
high-resolution (
HR
) remote-sensing satellites. It carries one
panchromatic/multispectral (
P/MS
) camera and two high-reso-
lution (
HR
) cameras (
HRA
and
HRB
). The
P/MS
camera captures
panchromatic (
PAN
) images at a ground sampling distance
(
GSD
) of 5 m and multispectral (
MSS
) images at a
GSD
of 10 m,
with a swath of about 60 km. The
HR
cameras capture
PAN
images at a
GSD
of 2.36 m, and the joint ground width of these
two cameras is 54 km. The satellite intends to meet the needs
of various applications, such as land resource inventory and
management, surveying and mapping, agriculture, forestry,
water resource conservation, environmental protection, urban
planning, transportation, and disaster management.
However, the direct georeferencing accuracy of the images
is much lower than their ground resolution because of the
systematic errors in the interior orientation parameters (
IOPs
)
and the exterior orientation parameters (
EOPs
) (i.e., orbit and
attitude parameters). An average accuracy of 1.8 km is usually
expected for direct georeferencing, with an additional interior
variation of the images from about six to eight pixels. Such
a poor geometric quality severely hampers the wide applica-
tions of the images. As a result, the on-orbit geometric calibra-
tion of satellite sensors must be carried out through accurate
sensor modeling.
Many sensor models have been proposed in the last 30
years, including empirical, general (e.g., the rational function
model (Fraser
et al
., 2003; Zhang
et al
., 2012)) and rigorous
sensor models (Kim
et al
., 2006; Michalis
et al
., 2008; Teo,
2011). With more
HR
satellites being launched in recent years,
Weser
et al
. (2008) proposed a generic sensor model for push-
broom satellite imagery. It is closely related to the physical
reality of the imaging process and can incorporate vendor-
specific parameters into the general model when metadata are
available. Ikonos was applied to a series of geometric calibra-
tions from 2001 to 2004 by using the calibration fields in Lu-
nar Lake, Railroad Valley, Dark Brooking, and Denver (Helder
et al
., 2003; Dial
et al
., 2003). The calibrations included
IOPs
(i.e., the Field Angle Map (
FAM
)) and elements of the exterior
orientation parameter set (i.e., the interlock angles). The cali-
brations raised the horizontal and vertical accuracies of un-
controlled Ikonos images respectively to better than 12 m and
10 m for the Root Mean Square Error (
RMSE
). Jacobsen (1997)
investigated the on-orbit geometric calibration of the
IRS-1C
satellite through a bundle block adjustment with 15 additional
parameters. The influence of the individual additional param-
eters on the calibration was tested by experiments on
IRS-1C
satellite images over the vicinity of Hannover of Germany.
These experiments showed that selecting appropriate addi-
tional parameters could improve the georeferencing accuracy.
The Panchromatic Remote-sensing Instrument for Ste-
reo Mapping (
PRISM
) on board the Japanese Advanced Land
Observing Satellite (
ALOS
) has three cameras with different
viewing directions (i.e., nadir, forward, and backward) that
can simultaneously capture images along the track. A main
characteristic of the
ALOS
PRISM
sensor is that, each camera
contains a number of charge-coupled device (
CCD
) chips in the
focal plane. Kocaman and Gruen (2007) used ten additional
parameters to correct the systematic errors in the
PRISM IOPs
.
Pengjie Tao, Luping Lu, and Yong Zhang are with the School
of Remote Sensing and Information Engineering, Wuhan
University, No. 129 Luoyu Road, Wuhan, 430079, P. R. China;
and Supresoft Inc, Guanshan 2 Road, Wuhan, 430074, P. R.
China (
).
Biao Xu is with the Chinese Academy of Surveying and Map-
ping, No. 28 Lianhuachi West Road, Beijing, 100830, P. R. China.
Songbai Zou is with the School of Remote Sensing and Infor-
mation Engineering, Wuhan University, No. 129 Luoyu Road,
Wuhan, 430079, P. R. China.
Photogrammetric Engineering & Remote Sensing
Vol. 80, No. 5, June 2014, pp. 505–517.
0099-1112/14/8006–505
© 2014 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.80.6.505
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
June 2014
505
