In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
landslides, although the classified objects is extending outside 
of the mapped landslides. The outside part is corresponding to a 
particular lithology in that area, that couldn’t be taken into 
account in the classification process. On the other sites of the 
image it can be seen the overlapping areas between the mapped 
landslides and the classified objects and the degree in which 
they are overlapping. On the right side on the Prahova River, 
presented in the middle of the image, they are big rotational 
slide which cover the entire slope and the identification of these 
big landslides were only partially. There is because of the slope 
gradient values, that here were below the minimum threshold 
value. The slope gradient is the effect of the uncertainties 
propagated from the DEM resolution and the slope algorithm. 
Figure 4 presents the south-eastern part of the study area, where 
the overlap between the classified objects as being landslide and 
the mapped landslides is almost inexistent. Most the features 
classified as being landslides are presented on the upper part of 
the hillslopes where the slope gradient and the profile curvature 
have met the criteria, but not on the grass lands with high 
texture ranges from the center of the image 
4. CONCLUSION 
The use of OBIA for landslides identification and classification 
has been proven as a promising method, but the results obtained 
are still far from obtaining a good and scientifically reliable 
landslide inventory. More attention should be given to the 
classification rules and to the data in the analysis. The lack of 
the DEM with a spatial resolution equal with the orthophotos 
and the lack of a more detailed lithological and pedological map 
have brought to many uncertainties in the classification process. 
Future approaches should take into consideration the quality of 
the data, but also the possibility of using more parameters in the 
segmentation process 
4.1 Acknowledgements and Appendix (optional) 
This study was partial financed from the following research 
projects: 
TD 221, 2006-2007, financed by the National Research 
Council, coordinated by Dr Ionut Sandric; 
AT 147, 2006-2007, financed by the National Research Council, 
coordinated by Dr Mihai Bogdan; 
IDEI 2007-2009 financed by the University of Bucharest, 
coordinated by Dr Comanescu Laura 
Special thanks to ESRI Romania 
5. REFERENCES 
Aheam, S., 2008. Object orientation, In: K.Kemp (ed.) 
Encyclopaedia of Geographic Information Science, SAGE 
Publications, Los Angeles, pp. 323-326. 
Armas I., Damian, R., Sandric I., Osaci- Costache, G., 2003. 
Vulnerabilitatea versantilor la alunecari de teren in sectorul 
subcarpatic al vaii Prahova, Editura Fundatiei Romania de 
Maine, Bucuresti, pp. 10-21. 
Balteanu, D., 1986. The importance of mass movements in the 
Romanian Subcarpathians. Zeitschrift fur Geomorphologie, 
Suppl. Band 58, pp. 173-190. 
Figure 3. Identified features versus mapped landslides (in red 
classified objects and in yellow mapped landslides); terrace 
scarp from eastern part of Breaza de Jos town 
Figure 4. Identified features versus mapped landslides (in red 
classified objects and in yellow mapped landslides); hillslopes 
from the south-western part of the study area 
In the center of the image (Figure 3) can be seen the almost full 
overlap between the classified objects and the mapped