Spectral Compression of Multispectral Images using Outlier Modeling and Subspace Clustering

This thesis introduces two different approaches for improving the performance of data reduction and reconstruction of multi-spectral images. First we introduce a new Outlier Modeling (OM) method that detects, clusters and separately models outliers with their own principal bases. In the second part of this research, a sub-space clustering strategy is used for the spectral compression of multi-spectral images. Unlike classic PCA, this approach finds clusters in different subspaces of different dimension. Consequently, instead of representing all spectra in a single low-dimensional sub-space of a fixed dimension, spectral data are assigned to multiple sub-spaces with a range of dimensions from one to eight. As a result, more resources can be allocated to those spectra that need more dimensions for accurate representation and fewer resources to those that can be modeled using fewer dimensions. This initial compression step is followed by JPEG2000 compression in order to remove the spatial redundancy in the data as well.