Shape indexing by dynamic programming

With the vast amount of publicly available high-quality images, image databases start to emerge, and image database management methods are becoming essential. An important manner of finding your way around an image database is to look for images similar to a desired reference image. There are many indicators for content similarity of images: color, texture, composition etc. In this paper we focus on shape similarity, and present a new dynamic algorithm for shape indexing ("DASI") based on curve-signatures and dynamic-programming.