| Modern radar signal processing is mainly focus on the signal detection, target tracking and recognition and threat judgment. The CFAR detection under the complex background has always been a hot and difficult problem in the radar signal processing. It is currently the main goal of the CFAR research, to enchance the effectiveness and stability of various clutter processing and reduce the CFAR loss.The clutter interference faced by modern radar signal processing is more complex. Because of its obvious shortcomings, the traditional single model CFAR detector is difficult to meet detection requirements. For example, CA-CFAR in homogeneous clutter has the superior detection performance, but has a defect of poor adaptability in the clutter edge and multi-targets environment. GO-CFAR improves the detection performance of the clutter edge, but it is nuable to adapt to the multi-target environment. SO-CFAR improves the detection performance of the multi-target interference, but its detection performance is poor in the clutter edge. So it is very difficult to achieve the desired detection performance by a certain fixed CFAR detector. Aiming at this problem, this paper designs the intelligent multi-model CFAR detector to improve the detection capability of the system.When designing the algorithm of the multi-model CFAR, the paper analyzes the clutter statistical model and the recognition algorithms. Because the paper mainly aims at the low-resolution marine radar, the Rayleigh distribution was selected as clutter statistical distribution model. Through the analysis of all kinds of clutter model identification algorithm, the paper uses the test of the goodness of fit based on Anderson-Darling to estimation the clutter edge, and then estimates the undulateness of clutter environment again according to statistics VI and MR, so it can select the appropriate processing algorithms for processing signal from a variety of CFAR processing models.The paper uses the advantages of the FPGA for logical operations and assignment of the large number of variables, and gives a concrete realization of the circuit for designed multi-modle CFAR algorithm. It can be effectively achieved the three CFAR processing of CA-CFAR, GO-CFAR, SO-CFAR and the related processing. The processor has good detection results in a variety of clutter and the resource consumption of the system is relatively reasonable. |
