Framework for Synthesizing Semantic-Level Indices

摘要: Extraction of the syntactic features is a well-defined problem thereby lending them to be exclusively employed in most content-based retrieval systems. However, semantic-level indices are more appealing user as they closer user's personal space. Most work done at semantic level confined limited domain developed and therein apply satisfactorily only that particular domain. Scaling up such systems would inevitably result large numbers features. Currently, there exists lacuna availability framework can effectively integrate these furnish indices. The objective this paper highlight some issues design report on status its development. In our framework, construction high-level index achieved through synthesis set elemental From collection features, an image/video class characterized by selecting automatically few principal By properly mapping constrained multi-dimensional feature space constituted with semantics data, it feasible construct high indices. The remains, however, identify or meaningful subset This medium Bayesian Network discerns data into cliques training pre-classified data. associates each clique points one classes during later used for evaluating probable which partition belongs. neither requires normalization different aid expert knowledge base. enables stronger coupling between extraction yet sufficiently independent, shown experiments. experiments were conducted over real video consisting seven diverse results show superiority standard classification tools.