Approximation of polygonal curves can be solved by dynamic programming. These methods provide an optimal solution but are slow for a large number of vertices. We propose a new optimization approach based on dynamic programming with a reduced-search in the state space. The proposed algorithm is simple, fast, and has low space complexity. The time and quality trade-off can be controlled by selecting an appropriate corridor width and number of iterations in the algorithm.
Published in:
Pattern Recognition, 2002. Proceedings. 16th International Conference on
(Volume:4
)
Date of Conference: 2002