Target Detection In Sea Clutter Based On Time-frequency Methods

With the development of the technology of high resolution to earth, the target detection and identification at sea has been an increasing hot research. The echo signals received by the radar include not only the scattering echo from sea background,but also the information about the target. In order to improve the radar’s performance in target detection, it is necessary to study the different characteristics between the two scattering echo. The feature on the two kinds of the scattering echo are studied in this paper based on the IPIX radar’s measured data released by the Adaptive Systems Laboratory at McMaster University in Canada. The following topics are addressed in the study:1.At first, two typical sea clutter simulation methods are introduced:zero memory nonlinearity(ZMNL) and spherically invariant random process (SIRP). With ZMNL method we manage to simulate three kinds of related clutter obeying Lognormal,Rayleigh,and Weibull distribution.Also, with SIRP method we simulate the K-distribution clutter sucessfully. In the end of the chapter, both the advantages and disadvantages about the two methods are analyzed.2. Secondly, the powers pectrum characteristics of sea clutter are studied. We include two kinds of power spectrum analysis methods:one is traditional method:periodogram method; and the other is the modern one:autoregressive(AR) model method, with which we analyze the spectrum of real data with target and without target. The results show that the frequency shift of the data with target and without target is existed. Also, the AR model method gets the smoother effect than the periodogram method. In addition we analysis the time and spatial correlation of the sea clutter by using the the correlation function method. The simulation results show that the sea clutter has a common characteristic no matter what kind of they are. which they go through a short related time and then experienced a longer decay period until the correlation function values are almost equeal to zero.3. Thirdly,the probability density function(PDF) of the amplitude of the real data are discussed. They are estimated and compared with some common empirical models, like Lognormal, Weibull, Rayleigh and K-distribution. It is found that the K-distribution could fit the real data well.4.The time-frequency charatitics of the data with and excluding target are analyzed. Short-time Fourier transform (STFT), the Wigner-Ville distribution (Wigner-Ville), Wigner-Hough transform and Fractional Fourier transform (FRFT) method are introduced to explore the time-frequency characteristic of data It is found that the STFT can obtain the good time-frenquency result. In addition from comparison of time-frenquency performance of based on WVD,PWD and SPWVD,it is proved that the SPWVD is more suitable for signal processing and can overcome disadvantages of the cross-term interference. The further research identifies that the slow motion of the target on the sea can be found out based on the WVD method It is obtained a linear banded structure on two-dimensional time-frequency diagram which can be recognized by the the Wigner-Hough transform by changing them into a new coordinate system to get a clearer affection. Finally, It is confimed that the different fractal characters is existed between the sea clutter and target echo over the FRFT in the best transform angle, from this we can detect moving target at sea.