Near-lossless ECG Data Compression Method For Mobile Tele-cardio-monitoring

New de-noise, characteristic waveforms detection and ROI (Region of Interest) based near-lossless compression methods for ECG signal based on wavelet transform have been proposed in this paper. De-noise and characteristic waveforms detection which are the necessary signal processing procedures of mobile cardio monitoring terminal, provide necessary information for ROI based near-lossless compression. So this thesis has involved the main signal processing steps of tele-cardio- monitoring.The de-noise algorithm exploits time-frequency distribution of ECG signal's wavelet transform to construct space domain filters for multi scales. By combining these filters with 3σfiltering method good noise remove effect has obtained.The characteristic waveforms detection algorithm, which exploits multiresolution analysis of wavelet transform and some anti-noise methods, gets the positions of characteristic points by means of detecting max module pairs and the points that pass zeros.The near-lossless ECG compression algorithm, based on the ROI technique, has provided lossless or near-lossless quality for the region of interest and adequate compression ratio that satisfies the application. It preserves the coefficients that relate to the reconstruction of region of interest and compress comparatively highly the coefficients that have no relation to region of interest. Diagnostic information can be preserved by such means.The de-noise, characteristic waveforms detection and near-lossless compression algorithms based on integer wavelet transform (IWT) have been proposed at chapter V. The most benefit of IWT is its low computation cost, thus other algorithms based on it can be real-time realized. But IWT does not belong to linear transform, so the loss version reconstruction of original signal is not smooth. Four-point filtering algorithm is adopt to make acceptable reconstruction quality available for de-noise and compression. The ECG data codec based on completely embedded SPECK (Set Partitioned Embedded bloCK) algorithm has the features of excellent performance and low complexity, along with others such as embeddedness and progressive transmission. The signal can be lossless compressed and transmitted, and acceptable quality can be got even under a low bit rate.