Vectorcardiographic Data Compression via Walsh And Cosine Transforms

In this study the use of two-dimensional transforms for compressing human vectorcardiographic (VCG) data is investigated. The VCG signal is two-dimensional in nature, one dimension consisting of the spatial axes and the other consisting of the samples in time along a particular spatial axis. The discrete cosine transform (DCT) and the Walsh-Hadamard transform (WHT) were used. The variance criterion was employed for selecting components to be retained. The training set was formed from 225 VCG records from three different diagnostic classes. The DCT yielded compression ratios from 3:1 to 5:1, while compression ratios of around 2:1 were obtained with the WHT.