| With development of long-distance medical treatment system , many kinds of medicine digital signal's acquirement,processing and transmission do in the embedded system. We have encountered some problems induced by the increasing quantity of signal data. To solve these problems, the key is to find out a quick and effective data compression algorithm. So it is very valuable to study medicine data compression algorithm applied in the embedded system.From many references to still image compression standard---JPEG2000, this paper studied new compression algorithms adopted by JPEG2000, then selected the best one which is appropriate to apply in the embedded system, finally degraded it to one dimension. In this algorithm, lifting wavelet is used to eliminate correlation of electrocardiograph (AB. ECG). Lifting wavelet is a development of classical wavelet. It removes the restriction of Fourier Transform theory and runs through three basic steps of split,predict and update. Compared to classical wavelet transform, the lifting wavelet can do integer-to-integer and in-place transform which reduce computing time and demand to memory. Attracted by characters of successive-approximation quantization (AB. SAQ) and gradual data transmission, We used Embedded Zerotree Wavelet (AB. EZW) to encode wavelet coefficients acquired from lifting wavelet transform. In the end, a quick and lossless compression algorithm have been realized.Then our algorithm was recomposed and optimized based on characters of digital signal processor (AB. DSP) and downloaded into DSP hardware system. Test with ECG data, we drew a conclusion following:(1) This algorithm can lossless compress ECG data and yield a high compression ratio than Huffman algorithm.(2) The waveform can reconstruct through imperfect datum, which carried out a gradual data transmission when the code stream was broken.(3) It reduces the computing time and demand to memory compared to Mallat algorithm.
|
PDF Full Text Request