Adaptive vector K-tree partitioning an entropy coder: application to ECG compression

Abstract

The traditional run length coding method is beneficial only in the case of long runs of ones or zeros in the bitstream. There is a need for an alternate encoding method that can achieve compression even if there are frequent changes in the bitstream. This paper presents a new one-dimensional coder named adaptive vector K-tree partitioning (AVKTP). The method emphasises on encoding the binary significance map in a depth-first search manner, also makes the length of sub-vectors bitstream adaptive. The application to ECG compression shows that AVKTP provides higher compression than RLE and VKTP coder retaining the signal quality. The optimum length of sub-vector achieves improved compression. Depth first search method facilitates the encoding and post compression quality control.