A Test-Bed for Retrieving Images with Extracted Features
通过抽取的特征进行图象检索的算法测试平台

Content based image retrieval is a hot research subject and has been studied widely in recent years. A number of feature based image retrieval algorithms, such as using color feature, texture feature, shape feature and position information etc., have been proposed in the literature. How to manage, compare, evaluate, and combine these algorithms becomes a challenge problem for further research and improvement in feature based image retrieval. This paper describes a specially constructed test bed for algorithms of retrieving images with extracted features. This test bed has both the functions of management, such as algorithm management, image database management, and individual image management, as well as the functions of integrating different feature based algorithms together for performing image retrieval task. On the basis of algorithm gathering, different retrieval stratagems, such as step up retrieval (including average and cumulative retrieval) and combined retrieval (using more than one independent feature based technique) have been realized. Experiments shows that with the help of this test bed, especially for its centralized control and management of algorithms and images, it is ease not only to compare and evaluate different feature based image retrieval techniques, but also to combine their advantages and to develop new algorithms.