Highly Nonlinear Approximations for Sparse Signal Representation
A previously introduced method for compression of 1D ECG signals is extended to operarate in 2D. Once the 1D to 2D conversion is realized, a transform lossy compression strategy is implemented. The particularity of the proposed 2D transformation consists in applying a Discrete Wavelet Transform along one of the dimensions and the Discrete Cosine Transform along the other dimension. The performance of the methods is demonstrated on the MIT-BIH Arrhythmia database. On the whole, the 2D approach yields very significant improvements in compression power.
by Johan Chagnon and Laura Rebollo-Neira