海面小目标检测的自适应背景感知研究
|
摘要
海面小目标检测是舰载雷达的重要使命,为了优化海杂波下的检测环境,并提高小目标检测的信杂比,引入背景估计思想,建立了基于径向基神经网络的海杂波背景感知模型。通过研究雷达实际测量数据,提出了简单有效的数据预处理方法,通过构造适合处理一维回波数据的神经网络模型,并采用正交最小二乘学习算法对模型结构进行自适应调整,减小了模型复杂度并提升了模型的性能,从而实现海杂波背景的良好感知。最后,基于实测数据对模型进行性能验证,针对杂波对消前后的数据计算信杂比,得到信杂比改善因子均值达到了2 dB。结果显示本方法优化了海杂波下的检测环境,并能够在一定程度上改善小目标检测的信杂比,表明了本方法的有效性。
Small target detection in sea clutter is an important mission of shipborne radar. In order to optimize the detection environment under sea clutter and improve signal-to-clutter ratio for small target, this paper introduces the background estimation idea, and establishes a background perception model based on radial basis function neural network. Through analyzing the actual data of radar, a simple and effective data preprocessing method is proposed. By constructing a neural network model suitable for processing one-dimensional echo data, and using the orthogonal least squares algorithm to adaptively adjust the model, the complexity of model is reduced and the performance is improved, so as to achieve a good perception of the background. Finally, based on actual data, the performance is verified. Signal-to-clutter ratio is calculated before and after clutter cancellation, and average value of improvement factor reaches 2 dB. Results show that this method optimizes detection environment under sea clutter, and can improve signal-to-clutter ratio for small target to a certain extent, indicating the effectiveness of the method.
Small target detection in sea clutter is an important mission of shipborne radar. In order to optimize the detection environment under sea clutter and improve signal-to-clutter ratio for small target, this paper introduces the background estimation idea, and establishes a background perception model based on radial basis function neural network. Through analyzing the actual data of radar, a simple and effective data preprocessing method is proposed. By constructing a neural network model suitable for processing one-dimensional echo data, and using the orthogonal least squares algorithm to adaptively adjust the model, the complexity of model is reduced and the performance is improved, so as to achieve a good perception of the background. Finally, based on actual data, the performance is verified. Signal-to-clutter ratio is calculated before and after clutter cancellation, and average value of improvement factor reaches 2 dB. Results show that this method optimizes detection environment under sea clutter, and can improve signal-to-clutter ratio for small target to a certain extent, indicating the effectiveness of the method.
引文
[1] 杨文浩, 李小曼. 融合子块梯度与线性预测的单高斯背景建模[J]. 计算机应用, 2016, 36(5): 1383-1386.
[2] 王福忠, 尹凯凯. 一种基于中值滤波的局部阈值分割算法[J]. 电子测量技术, 2017, 40(4): 162-166.
[3] 王铎. 基于形态学和邻域差值的红外小目标检测算法[J]. 光电技术应用, 2016, 31(2): 19-21,30.
[4] 刘浩然, 赵翠香, 李轩, 等. 一种基于改进遗传算法的神经网络优化算法研究[J]. 仪器仪表学报, 2016, 37(7): 1573-1580.
[5] 李伟,张旭东.基于卷积神经网络的深度图像超分辨率重建方法[J].电子测量与仪器学报,2017,31(12):1918-1928.
[6] BOUWMANS T. Traditional and recent approaches in background modeling for foreground detection: An overview[J]. Computer Science Review, 2014(11): 31-66.
[7] 王志明,张丽,包宏.基于混合结构神经网络的自适应背景模型[J].电子学报,2011,39(5):1053-1058.
[8] 张焱, 沈振康, 王平. 基于 BP 神经网络的红外小目标检测[J]. 系统工程与电子技术, 2004, 26(12): 1901-1904.
[9] 柴杰,江青茵,曹志凯.RBF神经网络的函数逼近能力及其算法[J].模式识别与人工智能,2002,15(3):310-316.
[10] HORNIK K, STINCHCOMBE M, WHITE H. Multilayer feedforward networks are universal approximators[J]. Neural Networks, 1989, 2(5): 359-366.
[11] PARK J, SANDBERG I W. Approximation and radial-basis-function networks[J]. Neural Computation, 1993, 5(2): 305-316.
[12] ZHANG N, DING S, ZHANG J. Multi layer ELM-RBF for multi-label learning[J]. Applied Soft Computing, 2016(43): 535-545.
[13] KHAN S, NASEEM I, TOGNERI R, et al. A novel adaptive kernel for the RBF neural networks[J]. Circuits Systems and Signal Processing, 2017, 36(4): 1639-1653.
[14] 潘立登, 吴宁川. 径向基函数神经网络正交最小二乘改进算法的实现[J]. 北京化工大学学报 (自然科学版), 2002, 29(4): 82-84.
[15] 丁昊, 董云龙, 刘宁波, 等. 海杂波特性认知研究进展与展望[J]. 雷达学报, 2016, 5(5): 499-516.
[2] 王福忠, 尹凯凯. 一种基于中值滤波的局部阈值分割算法[J]. 电子测量技术, 2017, 40(4): 162-166.
[3] 王铎. 基于形态学和邻域差值的红外小目标检测算法[J]. 光电技术应用, 2016, 31(2): 19-21,30.
[4] 刘浩然, 赵翠香, 李轩, 等. 一种基于改进遗传算法的神经网络优化算法研究[J]. 仪器仪表学报, 2016, 37(7): 1573-1580.
[5] 李伟,张旭东.基于卷积神经网络的深度图像超分辨率重建方法[J].电子测量与仪器学报,2017,31(12):1918-1928.
[6] BOUWMANS T. Traditional and recent approaches in background modeling for foreground detection: An overview[J]. Computer Science Review, 2014(11): 31-66.
[7] 王志明,张丽,包宏.基于混合结构神经网络的自适应背景模型[J].电子学报,2011,39(5):1053-1058.
[8] 张焱, 沈振康, 王平. 基于 BP 神经网络的红外小目标检测[J]. 系统工程与电子技术, 2004, 26(12): 1901-1904.
[9] 柴杰,江青茵,曹志凯.RBF神经网络的函数逼近能力及其算法[J].模式识别与人工智能,2002,15(3):310-316.
[10] HORNIK K, STINCHCOMBE M, WHITE H. Multilayer feedforward networks are universal approximators[J]. Neural Networks, 1989, 2(5): 359-366.
[11] PARK J, SANDBERG I W. Approximation and radial-basis-function networks[J]. Neural Computation, 1993, 5(2): 305-316.
[12] ZHANG N, DING S, ZHANG J. Multi layer ELM-RBF for multi-label learning[J]. Applied Soft Computing, 2016(43): 535-545.
[13] KHAN S, NASEEM I, TOGNERI R, et al. A novel adaptive kernel for the RBF neural networks[J]. Circuits Systems and Signal Processing, 2017, 36(4): 1639-1653.
[14] 潘立登, 吴宁川. 径向基函数神经网络正交最小二乘改进算法的实现[J]. 北京化工大学学报 (自然科学版), 2002, 29(4): 82-84.
[15] 丁昊, 董云龙, 刘宁波, 等. 海杂波特性认知研究进展与展望[J]. 雷达学报, 2016, 5(5): 499-516.