Choosing an appropriate image retrieval tool is the primary problem for a multimedia application such as digital image library and online image retrieval. Shape is often regarded as the most important image feature. Fourier descriptors (FDs) are widely used in shape recognition and retrieval. However, as global descriptors, they are often blamed for not being able to describe local shape features[1,2]. Wavelet descriptors (WDs) are proposed to overcome this drawback. Unfortunately, the extra information such as the multi-resolution scheme and local shape features cause much more complicate shape matching algorithms. The efficient or effective use of WD srequires more effort. Experiments are executed to evaluate the retrieval performance of this two descriptors. Some conclusions and suggestions are given according to the experimental results and the literature reviewed.
展开▼