Description

Inspired by the fact that the human visual perception system tends to tackle ambiguous information in a principled way with little effort in degradation, we model a system based on three strategies: extraction of essential signatures captured from a global context, simulating the global pathway; highlight detection based on local conspicuous features of the reconstructed image, simulating the local pathway; and hierarchical classification of extracted features using probabilistic techniques. The techniques involved in hierarchical classification use input from both the local and global pathways. We aim at exploiting the visual context through a combinatorial approach. Here, the visual context refers to a low-dimensional representation of the whole visual content, i.e., the “essential” of the scene or global essential signature of the image. Subsequently, the local conspicuous features and the global essential signature are combined and clustered by a Monte Carlo approach. Finally, clustered features are fed to a self-organizing tree algorithm (SOTA) to generate the desired hierarchical classification results. As an enhancement we reduce the dimension of feature vector using an optimization algorithm and classify the data using a probabilistic Bayesian classifier. This produces an enhanced classification result in terms of accuracy, time and storage capacity.