海面目标的稀疏检测方法研究
发布时间：2018-12-16 11:50
【摘要】：海面目标检测因受到海杂波的影响,传统目标检测方法易产生高虚警问题,如何有效抑制非平稳的、相关性强的海杂波,提高海面目标的检测能力,一直是雷达检测领域的难点问题。在高频近似情况下,海面目标(例如舰船等)的后向散射常呈现出多散射中心的特点,散射中心数量一般远小于观测区域的可分辨单元数量,基本符合压缩感知(CS,ComPressive Sensing)处理方法对目标散射稀疏先验的要求。因此,本文利用压缩感知技术重点开展了以下研究工作:1.在高斯噪声背景下,研究了迭代软阈值(IST,Iterative Soft Thresholding)算法的点目标重构能力和输出噪声特性。重点讨论了基于IST的两种固定门限检测器,推导了相应的检测概率和虚警概率的解析表达式,仿真分析表明,欠采样下的IST固定门限检测器性能优于匹配滤波最优检测器。另外,给出了基于IST的恒虚警(CFAR)检测器架构及其性能分析。2.在海杂波背景和点目标检测条件下,建立了海杂波复合K分布模型,研究了 OMP(Orthogonal Matching Pursuit)和 FOCUSS(Focal Undetermined System Solver)算法对中低海情海杂波的抑制性能,相比经典的白化滤波方法,它们具有更好的海杂波滤除能力。针对海杂波的复杂动力学行为,初步探索了海杂波抑制的深度学习(DeeP Learning)方法,利用卷积自编码器(CAE,Convolutional Auto-Encode)对回波谱图中的海杂波和目标进行了有效分离,初步验证了方法的可行性。3.在海杂波背景下,研究了扩展目标的稀疏检测方法。扩展目标的多散射中心通常呈现区域连续分布的特点,这里利用目标连续区域边界的稀疏性和非零的目标散射点与周围散射点之间的连续依赖关系,提出了结合总体变分(Total Variation,TV)正则化约束的SF-LASSO算法,仿真结果表明SF-LASSO算法能较准确反演目标位置和基本轮廓。
[Abstract]:Because the sea surface target detection is affected by sea clutter, the traditional target detection method is easy to produce high false alarm problem. How to effectively suppress the non-stationary and strongly correlated sea clutter and improve the detection ability of the sea surface target, It has always been a difficult problem in the field of radar detection. In the case of high frequency approximation, the backscattering of sea surface targets (such as ships, etc.) often presents the characteristics of multiple scattering centers. The number of scattering centers is generally much smaller than the number of discernible units in the observed region, which basically accords with compression sensing (CS,). A priori requirement for sparse target scattering by ComPressive Sensing) processing method. Therefore, this paper focuses on the following research work using compressed sensing technology: 1. In the background of Gao Si noise, the point target reconstruction ability and output noise characteristics of iterative soft threshold (IST,Iterative Soft Thresholding) algorithm are studied. In this paper, two kinds of fixed threshold detectors based on IST are discussed, and the analytical expressions of detection probability and false alarm probability are derived. The simulation results show that the performance of IST fixed threshold detector under under-sampling is better than that of matched filter optimal detector. In addition, the architecture and performance analysis of CFAR (CFAR) detector based on IST are given. 2. Under the condition of sea clutter background and point target detection, the composite K distribution model of sea clutter is established, and the suppression performance of OMP (Orthogonal Matching Pursuit) and FOCUSS (Focal Undetermined System Solver) algorithms to sea clutter in middle and low sea conditions is studied, compared with the classical whitening filtering method. They have better filtering capability of sea clutter. Aiming at the complex dynamic behavior of sea clutter, the depth learning (DeeP Learning) method for sea clutter suppression is preliminarily explored. The sea clutter and target in echo spectrum are effectively separated by convolution self-encoder (CAE,Convolutional Auto-Encode). The feasibility of the method is preliminarily verified. 3. In the background of sea clutter, the sparse detection method of extended targets is studied. The multi-scattering centers of extended targets usually show the characteristics of continuous regional distribution. In this paper, the sparsity of the boundary of the continuous region of the target and the continuous dependence between the non-zero scattering points and the surrounding scattering points are used. A new SF-LASSO algorithm with global variation (Total Variation,TV) regularization constraints is proposed. The simulation results show that the SF-LASSO algorithm can accurately retrieve the target position and the basic contour.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.51
本文编号：2382284
[Abstract]:Because the sea surface target detection is affected by sea clutter, the traditional target detection method is easy to produce high false alarm problem. How to effectively suppress the non-stationary and strongly correlated sea clutter and improve the detection ability of the sea surface target, It has always been a difficult problem in the field of radar detection. In the case of high frequency approximation, the backscattering of sea surface targets (such as ships, etc.) often presents the characteristics of multiple scattering centers. The number of scattering centers is generally much smaller than the number of discernible units in the observed region, which basically accords with compression sensing (CS,). A priori requirement for sparse target scattering by ComPressive Sensing) processing method. Therefore, this paper focuses on the following research work using compressed sensing technology: 1. In the background of Gao Si noise, the point target reconstruction ability and output noise characteristics of iterative soft threshold (IST,Iterative Soft Thresholding) algorithm are studied. In this paper, two kinds of fixed threshold detectors based on IST are discussed, and the analytical expressions of detection probability and false alarm probability are derived. The simulation results show that the performance of IST fixed threshold detector under under-sampling is better than that of matched filter optimal detector. In addition, the architecture and performance analysis of CFAR (CFAR) detector based on IST are given. 2. Under the condition of sea clutter background and point target detection, the composite K distribution model of sea clutter is established, and the suppression performance of OMP (Orthogonal Matching Pursuit) and FOCUSS (Focal Undetermined System Solver) algorithms to sea clutter in middle and low sea conditions is studied, compared with the classical whitening filtering method. They have better filtering capability of sea clutter. Aiming at the complex dynamic behavior of sea clutter, the depth learning (DeeP Learning) method for sea clutter suppression is preliminarily explored. The sea clutter and target in echo spectrum are effectively separated by convolution self-encoder (CAE,Convolutional Auto-Encode). The feasibility of the method is preliminarily verified. 3. In the background of sea clutter, the sparse detection method of extended targets is studied. The multi-scattering centers of extended targets usually show the characteristics of continuous regional distribution. In this paper, the sparsity of the boundary of the continuous region of the target and the continuous dependence between the non-zero scattering points and the surrounding scattering points are used. A new SF-LASSO algorithm with global variation (Total Variation,TV) regularization constraints is proposed. The simulation results show that the SF-LASSO algorithm can accurately retrieve the target position and the basic contour.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.51
【相似文献】
相关期刊论文 前6条
1 宁兰忠;;海面目标测控系统的一种设计方法[J];无线电工程;1987年02期
2 陈开元;宋元;;海面目标威胁评估指标体系分析[J];舰船电子工程;2009年06期
3 杨士英;;对海面远距离目标的高精度定位方法研究[J];电子信息对抗技术;2010年06期
4 杨宝成,沈国土,洪镇青,高兰香,毛宏霞,朱文勇,高景;海面目标热像理论模拟中的几何构型问题[J];华东师范大学学报(自然科学版);2001年01期
5 陈小龙;关键;何友;;微多普勒理论在海面目标检测中的应用及展望[J];雷达学报;2013年01期
6 孙春韦，，袁潭林，詹明阳，金亦华;2D搜索雷达对海面目标的群跟踪[J];雷达与对抗;1995年04期
相关重要报纸文章 前1条
1 本报特约撰稿 王群;让F—35们无处遁形的特殊雷达[N];中国国防报;2014年
相关硕士学位论文 前3条
1 赵洪宇;机载雷达对海目标数据处理方法研究[D];电子科技大学;2014年
2 张琦;海面目标的稀疏检测方法研究[D];中国科学技术大学;2017年
3 冯艳华;海面目标红外辐射特性研究[D];西安电子科技大学;2009年
本文编号：2382284
本文链接：https://www.wllwen.com/shoufeilunwen/xixikjs/2382284.html
