海杂波分形特性分析及目标检测方法

发布时间：2018-04-21 14:23

本文选题：目标检测 + 联合特征　；参考：《南京信息工程大学》2014年硕士论文

【摘要】：海杂波是雷达接收白海表面的散射回波,它在很大程度上影响了雷达对冰山、油轮和其他海上目标的定位和检测。因此,研究海杂波特性对减轻散射的影响、提高海杂波中目标检测是非常关键的。分形是非线性科学的重要分支,在处理非线性、非平稳信号方面发挥了很好的优势,尤其是在雷达杂波的建模和目标检测方面得到了很好的发展与应用。本文重在研究分形理论在海杂波和微弱目标检测领域的应用,理论研究和实测数据验证相结合,分析海杂波形成的基本机理和分形理论对于海上微弱目标检测的优势,结合IPIX实测雷达数据来检验所提检测方法的有效性。具体的研究内容如下：介绍了分形及其相关理论,尤其是分形维数和多重分形谱的定义和求解方法,分析了分形及其相关理论在雷达目标检测领域的应用,应用单一和多重分形方法分别对实测海杂波数据的分形特征进行仿真分析。在雷达回波中提取海杂波的分形参数和功率谱熵特征组成二维向量,运用凸包训练算法得到纯海杂波分类区域,同样对未知待测海杂波提取这两个特征参量,以此特征参量所对应的点是否在此分类区域内判别是否存在目标。利用IPIX雷达数据,证明了所提算法优于用单个特征差异作为统计量的方法,在相同虚警概率下检测效果明显提高,为雷达目标检测提供了新的检测方案。在实测数据的基础上,分析海杂波和目标回波的多重分形谱及广义Hurst指数,多组数据实验发现纯海杂波和目标回波的广义Hurst指数曲线分离度较大,尤其是在低尺度条件下,而不同距离单元的纯海杂波间广义Hurst指数曲线靠的很近。基于这些发现,我们选取了三个多重分形特征参数来量化两者之间的差异,利用SVM分类器对目标回波和纯海杂波进行分类实验,实验结果表明：无论是对于训练样本还是测试样本,基于多重分形特征的目标检测对目标单元和纯海杂波单元的正确检测率都在90%以上,同时,指出不同极化方式下分类准确率存在一定差异,说明了采用多重分形特征参数进行检测能基本实现了目标单元和纯海杂波单元的分离,具有一定的实用价值。
[Abstract]:Sea clutter is a radar that receives scattering echoes from the white sea surface. To a great extent, it affects radar location and detection of icebergs, tankers and other offshore targets. Therefore, it is very important to study the characteristics of sea clutter to reduce the scattering and improve the target detection in sea clutter. Fractal is an important branch of nonlinear science, which has played a good role in dealing with nonlinear and non-stationary signals, especially in radar clutter modeling and target detection. This paper focuses on the application of fractal theory in the field of sea clutter and weak target detection. The basic mechanism of sea clutter formation and the advantage of fractal theory for weak target detection at sea are analyzed by combining theoretical research with the verification of measured data. The effectiveness of the proposed detection method is verified by using IPIX radar data. The specific contents of the study are as follows: This paper introduces the fractal theory, especially the definition and solution of fractal dimension and multifractal spectrum, and analyzes the application of fractal theory in radar target detection. The fractal characteristics of measured sea clutter data are simulated and analyzed by single and multifractal methods. The fractal parameters and power spectral entropy features of sea clutter are extracted from radar echo to form a two-dimensional vector. The classification region of pure sea clutter is obtained by using convex hull training algorithm. The two characteristic parameters are also extracted for unknown sea clutter. Whether the points corresponding to this characteristic parameter exist in this classification area. Using IPIX radar data, it is proved that the proposed algorithm is superior to the method using single feature difference as statistic, and the detection effect is improved obviously under the same false alarm probability, which provides a new detection scheme for radar target detection. Based on the measured data, the multifractal spectrum and generalized Hurst exponent of sea clutter and target echo are analyzed. It is found that the generalized Hurst exponent curve of pure sea clutter and target echo is more separated, especially under the condition of low scale. The generalized Hurst exponent curves of pure sea clutter with different distances are very close. Based on these findings, we select three multifractal feature parameters to quantify the difference between them, and use SVM classifier to classify target echo and pure sea clutter. The experimental results show that the correct detection rate of target unit and pure sea clutter unit based on multifractal feature is more than 90% for both training samples and test samples. It is pointed out that there are some differences in classification accuracy under different polarization modes, which indicates that the detection of multifractal feature parameters can basically realize the separation of target units from pure sea clutter units, which is of certain practical value.
【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P715.9

【参考文献】