 |
 |
   |
|
|
|
The principle is to use fusion of information issued from several extraction
algorithms and to provide a more reliable result, and then fit the detected
linear features using prior information (vectors from the geographic database).
The whole process can be divided in four steps:
