ages, 
»oints 
ed to 
(3) 
Fig.6. TIN 
ED“* '■ £2 
¡EE 
^¡EEE 03 
JEe _ 
l~ 
SS 
iTW 
UBL_ 
£S ^ 
ÛT7E 
itttJH 
lïïli 
id 
id 
Fig.8.feature points 
Fig.9.Original image and combination images 
3. EXPERIMENTS 
We use the panchromatic band of CBERS and Landsat/ETM+ 
to test this method. These two images has different resolution 
and got from different time. First, we abstract the feature points 
at CBERS shows on Fig.8. Then we do the pointing matching 
what’s result shows on Fig.5.After that, we make TIN on the 
images which show on Fig.6.Fig.7 is the result of registration 
image. In order to check the precision of the registration, we 
bond the original image and registration image show on Fig.9, 
from which we can see that the combination image is more 
clear and precise. 
4. CONCLUSION 
In this paper, we have presented a method on High-Precision 
rectification and registration of Multi-source Remote Sensing 
imagery. This method used a lot of new techniques like 
improve hill climbing method, tiny facet Primitive and so on. 
In addition, we proposed a new method to check the 
registration result. 
However, though we have a good result on the registration of 
CBERS images and Landsat ETM+ images, there are job 
should be done on SAR and optical images registration. 
Moreover, using Actualized Algorithm of Retino-cortical 
Mapping to general mach which can abate artificial tie points is 
being studying. 
REFERENCES 
[ 1 ] L. G. Brown, “A survey of image registration techniques’ 
ACM Comput. Surv., vol. 24, no. 4, pp. 325-376, 1992. 
147