Sea Clutter Statistical Chaotic Characteristic Analysis And Small Target Detection

Sea clutter is the radar echo of the ocean surface. It seriously interferes ship radar’starget-detection capability. With the development of maritime trade, maritime securitybecomes more and more important. Therefore, it’s necessary and significance to studythe characteristics of sea clutter and target-detection on the sea-surface.This dissertation deeply studied sea clutter based on its statistical and chaoscharacteristics. We see sea clutter as a stochastic process model Based on the statisticalproperties. And usually we used ZMNL and SIRP two modeling methods to generate therandom sequence with a certain probability distribution, which is Rayleigh distribution,Log-Normal distribution, Weibull distribution and K distribution and so on. Actually,however, the sea clutter is a natural object. We cannot just see it as a random processmodel. Because the statistical distribution model cannot reflect the essence of sea clutter.Though the calculation of the chaotic invariants of Lyapunov exponent and Kolmogoroventropy of the sea clutter, we can prove that sea clutter has some determine factors. Atleast its can be predicted in a short time, which is a chaotic signal. Based on chaosfeatures we can reveal its intrinsic properties.Based on sea clutter statistical characteristics, this dissertation used constant falsealarm rate(CFAR) method to detect the target on the sea. In this method, we considersea clutter is linear and the target echo amplitude is much larger than sea clutteramplitude. Than we set a detection threshold based on this. So this detection method canonly detect a target which amplitude is much larger than sea clutter.By the reason of chaotic dynamics characteristics can reveal the intrinsic propertiesof sea clutter. In this dissertation, we use the IPIX Radar Sea Clutter Data, which areS.Haykin Pro acquisition. I used GRNN network on target detection at sea for the first time.Firstly, reconstructed phase-space sea-clutter data. Secondly, train the neural network bythe reconstruction of sea-clutter data. At last, we use this neural network to detect smalltarget at sea，and compared with the method of CFAR, according to the intrinsicproperties difference between the echo characteristics of the target and the sea clutter.At the end of this dissertation, according to software modularization, I designed a platform of characteristics simulation of sea clutter and target-detection on the seasurface with matlab software, which has good human-computer interaction feature.