BLOCK ADJUSTMENT BASED ON NEW STRICT GEOMETRIC MODEL OF
SATELLITE IMAGES WITH HIGH RESOLUTION
Jianqing Zhang*, Yong Zhang, Ying Cheng
School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, China, 430079 -
(jqzhang,yongzhang)@supresoft.com.cn, cheng99ying@ 163.com
Commission VI, WG IIL/1
KEY WORDS: Remote Sensing, Photogrammetry, High resolution, Block, Adjustment, Transformation, Exterior, Parameters
ABSTRACT:
The satellite image with high resolution has been commercially available since early 21st century, and the generation of digital
elevation model and Ortho-image with lower costs and shorter period from these high-resolution images has become practical. But at
first one must calculate the parameters of these high-resolution images. Recently, A new strict geometric model, based on affine
transformation, for RSIHR was proposed. In the new model, there are eight affine coefficients and one slantwise angle for each
image. However, in order to calculate the nine parameters, more than five control points are needed. Therefore, many control points
should be used in a block for the new model. It is difficult to acquire many control points in a block, and block adjustment with the
new model based few control points is necessary for computing the parameters of every image in the block. After the new strict
geometric model, based on affine transformation, for RSIHR is briefly presented, the method of the block adjustment with the new
model is introduced in this paper. At last, the method of the block adjustment with the new model has been tested for IKONOS and
other HRSI. The block adjustment based on RPC parameters with IKONOS images has been tested also and compared with the new
strict geometric model. All the tested result shows that the accuracy of the new model can reach the level corresponding to the
ground resolution of the images.
1. INTRODUCTION
High-resolution satellite imagery (HRSI) has been
commercially available since early 21* century, and the
generation of digital elevation model and Ortho-image with
lower costs and short period from these HRSI has become
practical. But at first one must calculate the parameters of these
high-resolution images.
High-resolution satellite images are acquired with Linear Array
Push-Broom Scanners. Because of the strong relativity of the
traditional parameters of remote sensing imagery, these
parameters could not be acquired sometimes. So it is necessary
to find some new way for sensor orientation for high-resolution
satellite imagery.
Recently, The Rational Polynomial camera (RPC) model has
consequently gained a considerable degree of popularity for 3D
object feature positioning from HRSI, especially given that they
have been shown to yield accuracies commensurate with
rigorous photogrammetric models (Grodecki, 2001). For
example, IKONOS and QUICKBIRD imagery have RPC or
RPB parameters that provided by the image suppliers. But
RPC/RPB parameters could not be used for directly geo-
positioning, block adjustment with control points is necessary.
It has recently been demonstrated that a bundle adjustment
approach can be employed with IKONOS imagery to yield
bias-corrected RPCs that enable sub-pixel geo-positioning and,
subsequently, high-accuracy DTM extraction and ortho-image
generation (e.g. Fraser and Hanley, 2003; Godecki and Dial,
2003; Jianging Zhang and Shunyi Zheng, 2003).
* Corresponding author.
But unfortunately, not all high-resolution satellite images are
supported by RPC or RPB parameters. Therefore, a new strict
geometric model, based on affine transformation, for the high-
resolution images without RPC or RPB parameters, was
proposed (Jianging Zhang and Zuxun Zhang, 2002). In this new
model, there are eight affine coefficients and one slantwise
angle for each image. Recent research has revealed this new
model has yielded geo-positioning accuracy no less than that
obtained via bias-corrected RPC block adjustment (Jianqing
Zhang and Shunyi Zheng, 2003). However in order to calculate
the nine parameters, more than five control points are needed.
Therefore, many control points should be used in a block. It is
difficult to acquire many control points in a block, and block
adjustment with the new model based few control points is
necessary for computing the parameters of every image in the
block.
In this paper, we will concentrate upon the approach of block
adjustment based on the new model with few control points. In
section 2, the method of block adjustment based on the new
model is introduced. The calculation of the initial value of
terrain coordinates of the tie points and orientation parameters
(eight affine coefficients and one slantwise angle) of each
image will be discussed in section 3. At last, the method of
block adjustment with the new model has been tested for
IKONOS images and other HRSI respectively. The block
adjustment based on RPC parameters with IKONOS images has
been tested also and compared with the new model. All test
results shows that accuracy of the new model can reach the
level corresponding to the ground resolution of the images.
International Arc
2. BLOCK
ST
Jianqing Zhang
three-step image
projection. In th
transformation is
2D image space,
affine projection,
the actual incline
coordinates of the
is described as the
N
mcosc
/ CR (x FE Xo ) ta
For the purpose
y - y, Will be sub
This model comp
and one slantwise
can be derived as t
Ax x sé
f cose
where AZ is the t
Fraser pointed out
equation (1) could
image is less thar
always the truth fo
[n the second test «
slantwise angles o
and the correction
AZ =300 meter.
If the orientation |
terrain coordinates
As follows:
(A5 P4). Y — 4" p
where:
[
ay dp 5 c
b, b,
94 a, a,+
A=
ba b,
A A
a, Ap Ga
b b
nl n2
The equation (His
