Ku Cheng-Tung, Hung King-Chu, Wang Huan-Sheng, Hung Yao-Shan
Institute of Engineering Science and Technology, National Kaohsiung First University of Science and Technology, Taiwan; Department of Information Management, Tzu Hui Institute of Technology, Taiwan.
Med Eng Phys. 2007 Dec;29(10):1149-66. doi: 10.1016/j.medengphy.2006.12.003. Epub 2007 Feb 16.
Error propagation and word-length-growth are two intrinsic effects influencing the performance of wavelet-based ECG data compression methods. To overcome these influences, a non-recursive 1-D discrete periodized wavelet transform (1-D NRDPWT) and a reversible round-off linear transformation (RROLT) theorem are developed. The 1-D NRDPWT can resist truncation error propagation in decomposition processes. By suppressing the word- length-growth effect, RROLT theorem enables the 1-D NRDPWT process to obtain reversible octave coefficients with minimum dynamic range (MDR). A non-linear quantization algorithm with high compression ratio (CR) is also developed. This algorithm supplies high and low octave coefficients with small and large decimal quantization scales, respectively. Evaluation is based on the percentage root-mean-square difference (PRD) performance measure, the maximum amplitude error (MAE), and visual inspection of the reconstructed signals. By using the MIT-BIH arrhythmia database, the experimental results show that this new approach can obtain a superior compression performance, particularly in high CR situations.
误差传播和字长增长是影响基于小波的心电图数据压缩方法性能的两个内在因素。为了克服这些影响，开发了一种非递归一维离散周期小波变换（1-D NRDPWT）和一个可逆舍入线性变换（RROLT）定理。1-D NRDPWT可以抵抗分解过程中的截断误差传播。通过抑制字长增长效应，RROLT定理使1-D NRDPWT过程能够以最小动态范围（MDR）获得可逆倍频系数。还开发了一种具有高压缩率（CR）的非线性量化算法。该算法分别为高低倍频系数提供小和大的十进制量化尺度。评估基于均方根差百分比（PRD）性能度量、最大幅度误差（MAE）以及对重建信号的视觉检查。通过使用麻省理工学院-贝斯以色列女执事医疗中心心律失常数据库，实验结果表明，这种新方法可以获得优异的压缩性能，特别是在高压缩率情况下。
