HP Labs Technical Reports
An Adaptive Compression Scheme for Waveforms
Bhaskaran, Vasudev; Natarajan, Balas K.
Abstract: Compression schemes have traditionally been developed from an information-theoretic viewpoint, often achieving compression by exploiting statistical redundancies in the data. In this paper, we present a compression scheme for waveforms that is developed from a geometric viewpoint, involving the construction of a sparse piecewise linear approximation to the waveform. In particular, the schemes hinges on an algorithm for the following problem: given a piecewise linear function F and error tolerance (epsilon), construct a piecewise linear function G that is within (epsilon) of F and consists of the fewest number of segments over all such functions. F and (epsilon) may be vector valued and of arbitrary dimension. The tolerance (epsilon) may also be independently specified at each sample point of F. G is said to be within (epsilon) of F, if for all , |F(x)-G(x)| less then or equal to epsilon. We report the application of this algorithm in a variety of settings including electrocardiograms (ECG), speech, and digitized grayscale images.
Back to Index