ABSTRACT

 

More and more images have been created in digital form around the world. There is growing interest in finding images from a large collection or from remote databases. In order to find an image, the image has to be described or represented by certain feature. Shape is an important visual feature of image. Searching images using shape features has attracted much attention. There are many shape representation and description techniques in the literature.

In this paper, important shape representation techniques are classified and reviewed. The implementation procedures of each technique are examined and its advantage and disadvantage are discussed. Some recent research results have been included and discussed in the paper. In the end, promising techniques for image retrieval will be identified according to standard principles.

Generally, there are two classes of approaches in shape representation and description: contour-based versus region-based. Under each class, the methods can be divided into structural methods and global methods. The different methods can be further distinguished between methods working in space domain and methods working in transform domain.

The review has found that structural approaches are useful in applications where partial matching is needed; methods based on Hausdorff distance are useful for locating objects in an image or sub-image matching. Both types of the methods have limited applications. For general shape applications, methods based on complex moments and spectral transform, such as Zernike moments and GFD, are the best choices. They satisfy the six principle set by MPEG-7: good retrieval accuracy, compact features, general application, low computation complexity, robust retrieval performance and hierarchical coarse to fine representation. If storage is a concern, FD can be considered.