The Sea Clutter Simulation And The Research On Target Detection Algorithms

The target detection technology under the sea clutter is a very popular and important research direction in both military and civilian areas, and has high utility value. It is an important task for bank-based radar and boat-based radar to how to construct and simulate the sea clutter and how to detect the target exactly in the sea clutter. Based on the analyses of the statistical characteristic of the sea clutter, the paper researches some simulation algorithms of sea clutter. Based on the sea clutter model, the paper researches and analyzes the relative detection algorithm. The main works of this thesis are as follows:(1) The paper describes the affection of sea surroundings, the structure relation of scatter of electromagnetism, the statistical characteristic of the sea clutter. Based on the analyses, using the ZMNL algorithm and SIRP algorithm, the paper simulates the K distribution, and using the theory of the scatter of electromagnetism, the paper simulates the scatter of Bragg.(2) The paper researches two algorithms of target detection——SIRP-GLRT and GSW. The sea clutter can be modelled by a compound Gauss serial. Based on the model, the SIRP-GLRT adopts the generalized likelihood ratio to detect the target. Using the simulate data, it gets a result according with the theory through the analyses.On the basis of Analyzing the real sea clutter, we can see the spectrum is the gauss spectrum. The GSW estimates the spectrum width, and regards the spectrum width as a target character.The experiment result proves the validation.(3) The paper analyzes and improves emphatically the ARLPM algorithm. Using the real sea clutter data to fit the AR model, it can find that the pole of the largest magnitude is the target pole, and the spectrum width is the principal target character. So it is the autoregressive largest pole magnitude (ARLPM). Except the spectrum width, the pole's spectrum amplitude and frequency can also be regarded as an assistant target character. Utilizing the pole's spectrum width and spectrum amplitude, we get the two-parameter detection algorithm; utilizing the pole's spectrum width and frequency, we get the detection algorithm based on the LVQ Neural Network. The simulation results show the two modified algorithms are better than the ARLPM.(4) The paper discusses an algorithm based on time-frequency model——Wigner-Hough detection. It has a good foreground.