A REtrieval Technique for Similiar Shapes

We propose an organization for a database of objects that permits the efficient retrieval of all objects in the database with a shape similar to a search template. The retrieval technique we propose is robust in the presence of noise, and can handle several different notions of similarity including changes in scale, position, and even relative sizes of components. We can thus have the computer reporduce, on a large database of images, the process performed by a human in 'riffiling' through a book, using an index structure to retrieve likely candidates quickly.