The map of fusion is then used to compute a dynamic programming with the FISCHLER and al. Algorithm [M. A. Fischler, 1981]. The main advantages of this algorithm is the robustness and the ability to allow the extraction of non monotonic curves using a two-way scanning of the search area, for instance, S-like shapes in any direction.

As any dynamic programming algorithm, it can find a path even if the linear feature is partially occluded. This is useful when linear feature disappears under vegetal cover or tunnel. But the path in the occluded parts is generally erratic due to noise; it is necessary to perform a smoothing.

This algorithm is very fast, this characteristic is very important because some vector databases are very imprecise and it is necessary to scan very large areas around the given vectors. Curvature information integration in the dynamic programming, as suggest MERLET and ZERUBIA [N. Merlet, 1996], slows down the process. So, the algorithm of FISHLER has been chosen and an independent smoothing is performed after optimization.

A slight modification of the original algorithm allows to select and fix one best initial point in the source set, and avoids short cuts close to initial end of the path.