1129 
RESEARCH ON RELATION BETWEEN CONTROL POINTS PRECISION AND 
GEOMETRIC CORRECTION ACCURACY 
Hong Youtang 
Department of Land Science and Technology, China University of Geosciences (Beijing) 
KEY WORDS: Geometric, Accuracy, Correction, Image, Analysis, Precision 
ABSTRACT: 
Through ASTER, ETM, SPOT5 and IKONOS four kinds of remote-sensing images (a coarse correction) , this paper makes a 
geometric correction for these four remote sensing images, using the control points which are collected by calculating with 
topographic map, surveying in field with portable GPS, and GPS (technique of RTK), calculates the image coordinate of the 
corrected test points. At last, researching the relation between control points precision and geometric correction accuracy by 
statistical calculating. By means of comparison and analysis in a set of test, discovers that the more higher accuracy of control points, 
the more higher geometric accuracy of corrected images. But for low-mid resolution remote sensing images (such as ASTER. ETM 
etc.), as the accuracy of control points improving, the geometric accuracy of corrected images elevates a little; however, for high 
resolution images (such as SPOTS5. IKONOS, etc.), as the accuracy of control points improving, the geometric accuracy of 
corrected images elevates a lot. 
1. INTRODUCTION 
There are several methods to get control points which are 
needed in geometric correction of remote sensing images. The 
first is surveying in field by portable GPS. Secondly, calculating 
with topographic map. The last one is obtaining through GPS 
(technique of RTK) or other high precision surveying 
instruments. Over the three ways, the portable GPS operation is 
easy and convenience but a low precision. Although survey by 
GPS (technique of RTK) or other high precision survey 
instruments has a high precision, it is a lot of fieldwork and not 
suitable to work in large area. The precision is relate to the scale 
for calculating with map, generally, it’s easy to buy topographic 
map with the 1:50000 (1:10000 in same area) scale, even less. 
However the larger scale maps are difficult to get and most of 
them are in old period. It’s an influence for selecting control 
points. Practically, most of control points which are needed in 
geometry adjustment of remote sensing images can be 
calculated through topographic maps. So how about the 
precision of geometric correction with the control points are got 
by other collecting means? Also, whether we can adopt other 
appropriate means of collecting control points or not for the 
different spatial resolution remote sensing images in geometric 
correction. This paper makes a geometric correction with 
control points collected by the above-mentioned three kinds of 
methods, through ASTER, ETM, SPOT and IKONOS four 
kinds of remote-sensing images (a coarse correction) for the 
study. After comparative analysis we get the conclusions. 2 
2. METHODS 
In this research, we select a certain amount of control points 
which can be identified from the topographic map, the remote 
sensing and the field. Calculate coordinates of collecting control 
points respectively on 1:50000 and 1:10000 scale topographic 
maps (correct by 1:10000 scale one for IKONOS). At the same 
time, get control point coordinates with portable GPS and GPS 
(RTK) in the wild. And then accurately correcting ASTER, 
ETM, SPOT and IKONOS four kinds of remote 
sensing(quadratic polynomial method for correction, most of the 
neighboring pixels of resample) through the three sets of data 
obtained, and calculating image coordinates of control points 
after accurate geometric correction. Finally ,by means of 
formula (1), we calculate and obtain point accuracy with 
accurate geometric correction for four kinds of remote sensing 
images, get the conclusions by comparative analysis. 
S r =±. 
\öö +SÔ] 
L X X y y J 
N 
(i) 
N is the number of control points 
2.1 Research on precision of geometric correction of remote 
sensing images by topographic map 
Select obvious points of ground feature, for example, make 
junction, comers of buildings as control points of correction. 
The control points in ASTER, ETM and SPOT5 image are 
selected by 1:50000 topographic map of Beijing, 16 control 
points are in ASTER image, 13 are in ETM image, 16 are in 
SPOT image; the control points in IKONOS image are selected 
by 1:10000 topographic map of Daanshan area in Beijing, the 
number of control points is 16. The detailed distribution of GCP 
reference Figure 1 —Figure 4. 
Figure 1. ETM image and control points for correction