Abstract
The performance of content-based image retrieval (CBIR) systems is largely limited by the gap between the low-level features and high-level semantic concepts. In this paper, a probabilistic semantic network-based image retrieval framework using relevance feedback is proposed to bridge this gap, which not only takes into consideration the low-level image content features, but also learns high-level concepts from a set of training data, such as access frequencies and access patterns of the images. One of the distinct properties of our framework is that it exploits the structured description of visual contents as well as the relative affinity measurements among the images. Consequently, it provides the capability to bridge the gap between the low-level features and high-level concepts. Moreover, such high-level concepts can be learned off-line, and can be utilized and refined based on the user’s specific interest during the on-line retrieval process. Our experimental results demonstrate that the proposed framework can effectively assist in retrieving more accurate results for user queries.
