Compressed representation of a backscatter ionogram database using Karhunen-Loeve techniques

An algorithm for the compression of a large database of 6*10/sup 5/ backscatter ionograms is presented, based on an iterative implementation of the Karhunen-Loeve transform. The algorithm removes database redundancy by representing images as vectors in a low dimensional "feature space" where image feature dynamics may be efficiently studied. The iterative algorithm ensures that the feature space spans the image variation manifest in the database, guaranteeing that every image may be reconstructed from its feature vector to within a preset accuracy threshold.<
>