The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3b. Beijing 2008 
476 
Figure 4a, b, c. Part of the experimental result which intercepts 
from the first group, a) building models(top), b), c) vertical 
edges extracted from left image and right image 
respectively(middle and bottom). 
Figure 5a, b, c, d. Part of the experimental result which 
intercepts from the second group, a) b) vertical edges extracted 
from left image and right image respectively(left top and right 
top), c) buildings model(left bottom), d) result of change 
detection(right bottom). 
Similar to Figure 4, Figure 5 is part of the experimental result 
which intercepts from the second group. Figure 5d) is the result 
of change detection. In the result, the regions identified by the 
white rectangle are the changes occurred in this area. The 
region identified by a white rectangle alone indicates that a 
building is destroyed, and the region identified by a white 
rectangle as well as some vertical edges indicates that there is a 
building added newly. 
It is found that the change result is more credible for the 
complex buildings. It is shown in the image that the outline of 
herringbone building is constituted by four vertical edges. One 
of them is covered, and one of them can not be extracted as a 
vertical edge because the pixel values have little difference 
around it. As a result, two credible vertical edges can be 
extracted most in the images, and this will certainly influence to 
the result of change detection. However, regarding the complex 
buildings, there are more vertical edges can be extracted for a 
certain building in the images. This is advantageous to describe 
and orientate the buildings more accurately, and the changes for 
the buildings can be detected more exactly. 
5. CONCLUSION 
The methods of image subtraction and image ratio are usually 
adopted by the traditional change detection. This paper 
proposes a method for buildings change detection based on the 
vertical edges and building models. The typical characteristics 
of the objects which are focused on are extracted first. The 
correction of extraction’s errors is carried on for several times, 
and the precision of vertical edges extraction is insured. This 
method is possible to neglect other factors which may influence 
the detection result, and focuses on the objects which is 
concerned. In this way, the accuracy of change detection can be 
enhanced to a great extent. Compared with the traditional 
change detection methods, it is active and purposeful detection. 
It has certain value in theory and application. But its 
widespread serviceability needs to be enhanced further. 
REFERENCES 
Canetti, R., Garay, J., Itkis, G., Micciancio, D., Naor, M., 
Pinkas, B., 1999. Multicast security: a taxonomy and some 
efficient constructions. In: Eighteenth Annual Joint 
Conference of the IEEE Computer and Communications 
Societie, New York, USA, pp. 708-716. 
Di, F. P., C. G. Zhu, L. Ding, 2008. Application of direction- 
vector analysis in change detection of urban land-use. 
Computer Engineering, 34(2), pp. 253-254. 
Hassiba, N., Youcef, C., 2006. Multiple support vector 
machines for land cover change detection: an application for 
mapping urban extensions. ISPRS Journal of Photogrammetry 
and Remote Sensing, 61(2), pp. 125-133. 
Li, D. R., Y. Q. Zhou, W. X. Jin, 2001. Generalities of 
Photogrammetry & Remote Sensing, Surveying and Mapping. 
Survey Publishing House, Beijing, pp. 13-50. 
Li, Q., L. Li, X. Zhao, 2005. Urban change detection using 
landsat TM imagery. Editorial Board of Geomatics and 
Information Science of Wuhan University, 30(4), pp. 351-354.