Object and concept recognition for content-based image retrieval.

摘要

The problem of recognizing classes of objects in images is important for annotation and indexing of image and video databases. Users of commercial CBIR systems prefer to pose their queries in terms of key words. To help automate the indexing process, we represent images as sets of feature vectors of multiple types of abstract regions, which come from various segmentation processes. With this representation, we have developed two new algorithms to recognize classes of objects and concepts in outdoor photographic scenes. The semi-supervised EM-variant algorithm models each abstract region as a mixture of Gaussian distributions over its feature space. The more powerful generative/discriminative learning algorithm is a two-phase method. The generative phase normalizes the description length of images, which can have an arbitrary number of extracted features. In the discriminative phase, a classifier learns which images, as represented by this fixed-length description, contain the target object. We have tested our approaches by experimenting with several different data sets and combinations of features. Our results showed a significant improvement over the published results.