In this ongoing work, we propose a Bayesian model that can be used to detect targets in multispectral images when the signals from the materials in the image mix linearly, the noise is Gaussian, and abundance parameters are nonnegative. By using efficient implementations of the Gibbs sampler, the expectation of any measurable functional of the abundance parameters, relative to the posterior distribution, can be computed easily. This general approach can be used to include additional constraints.

This paper presents an efficient content-based image retrieval system that captures users' semantic concepts in clusters. These semantically homogeneous clusters aid in the retrieval system to accurately measure the semantic similarity among images and therefore reduce the semantic gap. They also aid in the retrieval system to find matched images in a few candidate clusters and therefore reduce the search space. The extensive experiments demonstrate that the proposed retrieval system outperforms the peer systems to quickly retrieve the desired images in a few iterations.