Object-based Image Analysis
2009
With the availability of high-resolution satellite imagery, the use of remote sensing data has become very important for the verification of the nuclear safeguards agreement between a NPT Member State and the IAEA. Items and activities which can be monitored within nuclear facilities are, for instance, construction of buildings, plant expansion, changes of the operational status, and planning of underground activities. When applied to high-resolution imagery, traditional pixel-based image algorithms often yield limited results. Especially, if small structural objects are to be detected, object-based procedures are more appropriate. In comparison to solely spectral features applied within pixel-based approaches, utilization of object features, such as size or orientation of objects, their shape or texture and the relations between the objects in dierent scales, considerably extends the possibilities of image analysis. Analyzing satellite image data in an object-based method also oers the possibility to involve specific knowledge in the image classification or recognition process. This specific knowledge is here given through the “interpretation key”, presented in the previous chapters. This part deals with the computerbased interpretation of satellite and aerial images. Thereby the applicability of the “key” in the computer-based process is investigated. Figure 1 shows the advantage of a wide feature basis within an object-based classification. Here a subset of an Iranian Nuclear Power Plant is shown. With only the spectral values it is very dicult for a pixel-based classification to detect the heterogenous building in the top of the image and one will hardly be able the find out what kind of a building it is. Using objects provides us with a very wide feature base as mentioned above. So, a possible classification rule for this building could be: all objects within the image which are rectangular and have a special size (shape features), which have a given distance to the reactor dome (position feature) and have smaller rectangular objects on the roof (object-relation feature) are “Generator Halls”. The parameters for
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
7
References
4
Citations
NaN
KQI