International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
The pixels number is selected as the evaluation of the extraction 
precision and the detailed steps are described as followings: 
(1) Calculate semi-automatic extraction area M, pixels and 
actual area M; pixels manually selected of the residential area 
(2) Superpose M, and M» and measure the right extraction 
pixels number, the error extraction pixels number and the miss 
extraction pixels number, calculate correspondence percent. 
Table 3 shows the precision results. 
Table 3 Comparison of the semi-automatic extraction (Unit: Pixel) 
Neural-Network Structure on a Multi-spectral Land-use/Land- 
Cover Classification, Photogrammetry Engineering & Remote 
sensing, 1997, 63(5):535-544 
  
  
  
  
  
  
Image M M Right Error Missing Right Error Missing 
No : 2 Arca Arca Arca Ratio% | Ratio% Ratio% 
4a 22660 | 21246 20670 1990 576 91.2 8.8 2.7 
4b 3221 13465 12825 396 640 97.0 3.0 4.8 
4c 19235 | 19254 17944 1291 1310 93.2 6.8 6.8 
4d 8535 7885 7722 813 163 90.5 9.5 2.1 
4e 55183 | 39670 54171 1012 5499 98.2 1.8 9.2 
  
  
  
  
  
  
  
  
  
  
As Table 3 shown, the right extraction ratios are all higher than 
90%. The average right percent is over than 94.2% and this can 
meet the requirements of the actual applications. 
4. CONCLUSIONS AND DISCUSSIONS 
The semi-automatic extraction results of residential areas front 
different types of high resolution images show that the method 
is simple and efficient and only takes very short time with the 
94.2% correctness to the residential area extraction. And the 
residential areas extracted by this method are available to 
provide information for some application such as planning and 
decision-making. To improve this method to high efficient, the 
seed selecting should be done with the support of the neighbor 
characters of the pixel distribution to obtain the better region 
growing results by the computer. 
ACKNOWLEDGEMENT 
The author would like to thank Photogrammetry research group 
to provide experiment images. 
REFERENCES 
Zhang Z.X, Zhang J.Q. Digital Photogrammetry, Publishing 
House of Wuhan University, 1996, 126-131 
Yang C.J. The Study on the Residential Area Automatic 
Extraction from TM Image Based on t Knowledge. Remote 
Sensing Technique and Application, Mar 2001, Vol. 16, No.1, 
2-5 
Rob J.Dekker, Texture Analysis of Urban Areas in ERS SAR 
Imagery for Map Updating, IEEE/ISPRS Joint Workshop on 
Remote Sensing and Data Fusion over Urban Areas, 2001, 226- 
230 
V Bessettes and J.Desachy, Extraction and classification of 
urban areas on Spot images, IEEE, 1998, 2583-2586 
Vairy M Venkatesh Y V, Debluring Gaussian Blur Using a 
Wavelet Array Transform, Pattern Recognition, 1995, 2 
8(7):965- 976 
Chang YL, Li X B. Adaptive image region-growing. IEEE-IP, 
1994, 3(6):868-872 
V.Bessettes and J.Desachy, Extraction and classification of 
urban areas on Spot images, IEEE, 1998, 2583-2586 
Justin D.Paola and Robert A.Schowengerdt, The Effect of 
1126