基于最大值参考单元的双剔除门限恒虚警目标检测算法

针对传统恒虚警目标检测算法在非均匀噪声环境下虚警率过高的问题,提出一种基于最大值参考单元的双剔除门限恒虚警(dual censoring threshold based on maximal reference cell-constant false alarm rate,DCT-MRC-CFAR)目标检测算法.基于参考窗最大值参考单元得到剔除比较门限,通过比较,剔除极大值参考单元,根据剩余参考单元的数量选择参考窗中剔除后剩余参考单元或者参考窗中全部参考单元来估计背景噪声功率,并得到功率检测门限.与其他算法进行仿真对比,DCT-MRC-CFAR算法在均匀噪声环境下,检测性能接近于单元平均恒虚警(cell averaging-constant false alarm rate,CA-CFAR)算法;在非均匀噪声环境下,检测性能较稳定,且优于自动删除单元平均恒虚警(automatic censored cell averaging-constant false alarm rate,ACCA-CFAR)和自动双删除单元平均恒虚警(automatic dual censored cell averaging-constant false alarm rate,ADCCA-CFAR)算法.结果表明,提出的DCT-MRC-CFAR目标检测算法在均匀和非均匀噪声环境下,均具有较优和较稳定的检测性能.

Dual censoring threshold CFAR target detection algorithm based on maximal reference cell

For the problem that the conventional constant false alarm rate (CFAR) detection algorithms suffer from excessive false alarms in non-homogenous environment,a dual censoring threshold based on maximal reference cell-constant false alarm rate (DCT-MRC-CFAR) target detection algorithm is proposed.Compared with the censoring threshold which is generated based on the amplitude of maximum reference cell,the higher amplitude reference cells are censored from the reference window.Base on the number of the remaining reference cells in the reference window,the proper reference cells are selected from the remaining reference cells or all reference cells of the reference window to computer the background noise power level and obtain the adaptive detection threshold.The performance of the DCT-MRC-CFAR detection algorithm is evaluated in different simulation environments.Compared to the other detectors,the DCT-MRC-CFAR detection algorithm performs like the CA-CFAR detection algorithm in homogeneous environment and better than the automatic censored cell averaging-constant false alarm rate (ACCA-CFAR) and automatic dual censored cell averaging-constant false alarm rate (ADCCA-CFAR) detection algorithms in non-homogeneous environment.The simulation results show that the proposed DCT-MRC-CFAR detection algorithm has excellent and robust performance both in homogenous and non-homogeneous environments.