An elastic band model for Shape retrieval

Abstract

In this paper, we present novel model to find the contour of an image in phases. The model was designed by modelling physical motion of a real elastic band. This is applied to top of any existing shape representation methods in order to offer additional information. The model makes us deal with global and local shape characteristics at once and results in robustness to distortion and noise on the retrieval system as it does not miss the detailed data. Dynamic time warping algorithm is employed to measure the similarity between shapes. We have experimented in MPEG-7-CE-Shape-1 part B image database and the results show that our model is robust to distortion, noise and images having complexities inside of the boundary.