Study On Approaches Of Life Detection Radar Signal Processing

Life detection radar is based on Doppler principle to detect life information in non-contact way. It has been a hot research topic at home and abroad because of its important applications in the field of military, medicine, disaster medicine and city counter-terrorism in recent years.Firstly, life detection radar echo signal denoising method is introduced in this paper. Life detection radar echoes are often very weak and submerged in noise. The SNR of radar echo signal is generally very low. Therefore, echo signal denoising is indispensable for extracting life information. First, the effectiveness of frequency domain in accumulation and wavelet transformation in denosing are presented respectively. According to the analysis, the wavelet threshold denoising method has the advantages of small amount of calculation, but demands higher signal SNR. The method of accumulation in frequency domain can handle signal with very low SNR. But it requires a longer time accumulation. According to the characteristics of these two methods, a new method for denoising based on the combination of accumulation in frequency domain and wavelet threshold denoising is proposed in this paper. Simulation results show that the new method has high efficiency in denoising and small calculating load.Secondly, signal processing methods of extracting life information from the radar echo signal are presented. Signals are analyzed in both the time and frequency domain respectively. Autocorrelation analysis and MUSIC power spectrum estimation based on matrix solutions can be used to obtain the features of echo signal both in time and frequency domain, which could be used to judge whether the target is a human being. Although time and frequency domain analysis can roughly say if there are people there, but it is not so good to identify if there is one or more. Therefore further data processing is needed to extract more features to identify target information more accurately. In this paper, spectrum is used to analyze echo signals from people around, and SVD (Singular Value Decomposition) is used to construct the feature vector for identifying single people or multi-people.Finally, pattern recognition can be done by using the Minimum Distance Classifier (MDC). Target recognition methods discussed in this paper are simulated by using the MATLAB platform and results show that the methods to be effective.