Investigating the Behavior of Compact Composite Descriptors in Early Fusion, Late Fusion and Distributed Image Retrieval

In Content-Based Image Retrieval (CBIR) sys- tems, the visual content of the images is mapped into a new space named the feature space. The features that are chosen must be discriminative and sufficient for the description of the objects. The key to attaining a successful retrieval sys- tem is to choose the right features that represent the images as unique as possible. A feature is a set of characteristics of the image, such as color, texture, and shape. In addition, a feature can be enriched with information about the spa- tial distribution of the characteristic that it describes. Eval- uation of the performance of low-level features is usually done on homogenous benchmarking databases with a lim- ited number of images. In real-world image retrieval sys- tems, databases have a much larger scale and may be hetero- geneous. This paper investigates the behavior of Compact Composite Descriptors (CCDs) on heterogeneous databases of a larger scale. Early and late fusion techniques are tested and their performance in distributed image retrieval is cal- culated. This study demonstrates that, even if it is not possi- ble to overcome the semantic gap in image retrieval by fea- ture similarity, it is still possible to increase the retrieval ef- fectiveness.