Husøy J H, Gjerde T
Rogaland University Centre, Department of Electrical and Computer Engineering, Ullandhaug, Stavanger, Norway.
Med Eng Phys. 1996 Mar;18(2):132-42. doi: 10.1016/1350-4533(95)00028-3.
A data compression technique is presented for the compression of discrete time electrocardiogram (ECG) signals. The compression system is based on sub-band coding, a technique traditionally used for compressing speech and images. The sub-band coder employs quadrature mirror filter banks (QMF) with up to 32 critically sampled sub-bands. Both finite impulse response (FIR) and the more computationally efficient infinite impulse response (IIR) filter banks are considered as candidates in a complete ECG coding system. The sub-bands are threshold, quantized using uniform quantizers and run-length coded. The output of the run-length coder is further compressed by a Huffman coder. Extensive simulations indicate that 16 sub-bands are a suitable choice for this application. Furthermore, IIR filter banks are preferable due to their superiority in terms of computational efficiency. We conclude that the present scheme, which is suitable for real time implementation on a PC, can provide compression ratios between 5 and 15 without loss of clinical information.
提出了一种用于压缩离散时间心电图（ECG）信号的数据压缩技术。该压缩系统基于子带编码，这是一种传统上用于压缩语音和图像的技术。子带编码器采用具有多达32个临界采样子带的正交镜像滤波器组（QMF）。在完整的ECG编码系统中，有限脉冲响应（FIR）滤波器组和计算效率更高的无限脉冲响应（IIR）滤波器组均被视为候选方案。对子带进行阈值处理，使用均匀量化器进行量化并进行游程编码。游程编码器的输出由哈夫曼编码器进一步压缩。大量仿真表明，16个子带是此应用的合适选择。此外，IIR滤波器组因其在计算效率方面的优势而更可取。我们得出结论，本方案适用于在个人计算机上进行实时实现，可在不损失临床信息的情况下提供5到15之间的压缩比。
