Signal compression is an important problem encountered in many applications
. Various techniques have been proposed over the years for addressing the p
roblem. In this paper, we present a time domain algorithm based on the codi
ng of line segments which are used to approximate the signal. These segment
s are fit in a way that is optimal in the rate distortion sense. Although t
he approach is applicable to any type of signal, we focus, in this paper, o
n the compression of electrocardiogram (ECG) signals. ECG signal compressio
n has traditionally been tackled by heuristic approaches. However, it has b
een demonstrated [1] that exact optimization algorithms outperform these he
uristic approaches by a wide margin with respect to reconstruction error By
formulating the compression problem as a graph theory problem, known optim
ization theory can be applied in order to yield optimal compression. In thi
s paper, we present an algorithm that will guarantee the smallest possible
distortion among all methods applying linear interpolation given an upper b
ound on the available number of bits,
Using a varied signal test set, extensive coding experiments are presented.
We compare the results from our coding method to traditional time domain E
CG compression methods, as well as, to more recently developed frequency do
main methods. Evaluation is based both on percentage root-mean-square diffe
rence (PRD) performance measure and visual inspection of the reconstructed
signals. The results demonstrate that the ex-act optimization methods have
superior performance compared to both traditional ECG compression methods a
nd the frequency domain methods.