The Applications Research Of The Fractional Fourier Transform In The ECG Processing

With the development of medical services in the information technology, people began to study various types of biomedical signals. ECG, as a typical non-stationary random biological signal, is easy to testing and intuitive features, and then, it attracted more and more attention from researchers. The paper researched ECG combined with the theory of fractional Fourier transform and desgined a random signal filtering algorithm which has good real-time processing, quick convergence speed and high calculation accuracy.In the biomedical signal processing, because the biomedical signal is very weak, and vulnerable to interference from the external environment, and what's more, it suffer strong coupling in time and frequnency domain form noise. So the processing method must combine the static charactristics of ECG and remove the coupling in time and frequency domain. The traditional analysis method can't ensure the effectiveness of the filter algorithm, and Fractional Fourier transform, a linear transform in time-frequency domain, is able to map one dimensional time domain or frequency domain function to the two dimensional time-frequency domain function, and accurately reflects the energy distribution of the signal according to the time and frequency, and it is inexist these interference of cross terms. In a word, the fractional Fourier transform was used; the processing result would get better.This paper firstly aimed to analyze a filter method from the Fractional Fourier transform based on single component chirp signal; and then filter single component chirp signal with 5dB Gaussian white noise according to the time-frequency focuses characteristic of Fractional Fourier transform. Experiment result indicated that such algorithm could improve the signal of noise (SNR) to 12.006dB and it owned high convergence speed and good property in anti-noise with analyzing the robustness and astringency of the algorithm. After researching the adaptive filter algorithm, this paper proposed a LMS adaptive filter algorithm based on Fractional Fourier Transform theory, which applying joint process characteristic of time and frequency domains and its order sensitivity feature presented by this theory. The simulation object is ECG data in"Arrhythmia Database"and"Noise stress Test Database"of MIT-BIH, the result revealed that the SNR increased to 28.249dB from 1.05dB and MSE was 0.841, which showed that the algorithm had higher filtering effect, faster convergence speed and strong anti-jamming capability. At last, the paper achieved the filtering algorithm on the S3C2410 platform with processor to make particular feature point and waveform been recovered perfectly.