Abstract
This paper proposes a partial shape matching method for functionally approximated shapes. The proposed method consists of a feature extraction procedure and a correspondence detection procedure among convexities. In the feature extraction, we introduced a wedge-shaped tiny figure, called wedge wave, as a basis of convexities. The DP (Dynamic Programming) matching has been applied to the correspondence detection among extracted convexities using a method to reduce the amount of computation.
The proposed method has been applied to several figures for evaluation. The results showed that our method could recognize partial similarity without the restriction of location, size, and orientation. It has been confirmed that, by utilizing the information of an approximating function, a large amount of computation in feature extraction can be reduced.
