The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3b. Beijing 2008 
513 
Figure 13. Extraction Result of Single-shape road 
Figure 14. Result of Crossing Road Figure 15. Result of Cruciform 
Cross 
Figure 16. Extraction Result of Elevate Urban Highway 
From the results, we can get the conclusions that the method 
proposed 
1) There is a great accuracy (Figure 13)in extracting road with 
single shape(Figure 5). Edge line and the centrelines of the road 
can be showed exactly. 
2) To the roads whose surface has many branches (Figure 6), 
the extraction result is not very good (Figure 14). One branch 
has not been extracted accurately. The reason for it is that in the 
same area, different branches have different grey value, which 
will be unable to extract all of the branches totally. 
3) Cruciform cross (Figure 7) is one of the most popular kinds 
of road, which consists of two simple crossing lines. Although 
the extracting result is not better than 1), cruciform cross has 
been marked obviously. (Figure 15). 
4) To the elevate urban highways (Figure 8), this proposed 
approach has great advantage in extracting work. Although the 
shape of them is complex, its edge has also been detected, 
especially the overlapping part(Figure 16). 
5. CONCLUDING REMARKS 
The results of research indicate that, the method mentioned in 
this paper has a good effect on the extraction work of the road 
information in modem cities, especially on the elevate 
highways that are obvious in modem cities. The appearances of 
elevate highways now are getting more and more complex, the 
centre section of this which is composed of many surrounding 
roads overlapping each other whose curvature are larger than 
general ones. Considering the aspect of the uniformity of grey 
value on road, we group the edge points of road again in the 
same supported region, which can realize the elevate highways 
exactly. With the rapid development of city and the complex of 
the road, traditional methods for extraction haven’t satisfied for 
all the demands of the urban development, so an improved one 
in this paper has been proposed in view of the existed problems. 
But some roads can’t be extracted correctly for the deficiency 
of the image quality and algorithm. Moreover, a few distances 
from the original roads appeared in some road crossings, which 
should be do more researches on it. 
ACKNOWLEDGMENTS 
This work was financially supported by: National 863 High- 
tech Project “ Complicated Features’ Extraction and Analysis 
from Central Districts in Metropolis” PRC (2007AA12Z178). 
And the paper is also substantially supported by Chang Jiang 
Scholars Program, Ministry of Education, People's Republic of 
China. The author would like to thank ChangJiang Scholars 
Professor Li RongXing for his help. 
REFERENCES 
Chen Y., 2003 Digital Photogrammetry Surveying of Remote 
Sensing Image, Tongji University Press, China, pp. 100-129 
Gruen A., Li H. Road Extraction from Aerial and Satellite 
Images by Dynamic Programming[J], ISPRS Journal of 
Photogrammetry and Remote Sensing, 1995, 50(4), pp.11-20 
Li Y.X., Li H.Y., A Semi-automated Approach to Road 
Extraction From High-resolution Satellite Imagery[J],Journal of 
Xiangnan University Oct., 2004 Vol.25 No.5, P75-P76. (in 
Chinese with English abstract) 
Rafael C. Gonzalez , Richard E. Woods Digital Image 
Processing (Version II) ,2004, Electronics Industry Press, 
China, 2004.6, pp.463-474 
Rao H., Zhang H., Cai Z.Y.. Recognition of Welding Line 
based on Phase Classification[J]. Modem Electronic Technique, 
2004, 27(16), pp 66-69 
Shi W.Z., Zhu C.Q., Wang Y. The Rreview and Prospect of 
Extraction Road Features from Remote Sensing ImagesfJ], Acta 
Geodatica et Cartograghica Sinica, 2001, 30(3), pp.257-261.(in 
Chinese with English abstract) 
Song F.L., Liu R. Road Extraction from Remotely Sensed 
Image : Review and Prospects [J], Water Conservancy Science 
and Technology and Economy Vol. 11 No. 10 Oct.2005, 
pp.636-638. (in Chinese with English abstract) 
Trinder J.C., Wang Y.D., Sowmya A., et al. Artificial 
Intelligence in 3D Feature Extraction[A], Automatic Extraction