Compressing High-Res Images Correlated in Multiple Dimensions

Eigenspace methods take advantage of the fact that a set of highly correlated images can be approximately represented by a small set of eigenimages. However, all known eigendecomposition algorithms are directly proportional to the resolution of original images. With the ability to generate consistently better resolution images using state-of-the-art equipment, these techniques still show performance issues and exponential increase in memory requirements. Also, these algorithms exploit one-dimensional temporal correlation between successive images and hence, cannot be efficiently applied to images correlated in multiple dimensions. This book analyzes the effect of spatial resolution reduction of images on their resulting eigendecompositions and proposes computationally efficient technique that reduces spatio-temporal correlations in those images. This work also explains effective parametrization of images correlated in multiple dimensions and proposes optimum ordering of those images in frequency domain. This optimum ordering allows the proposed algorithm to uncorrelate the data as efficiently as possible without introducing too large an error in resulting eigenspace approximation.