Classification of targets by polarimetric information

研究项目的目的是利用电磁信息的全矢量特性对雷达目标进行分类。基于辛克莱散射矩阵的雷达偏振测量原理使我们能够根据以下项目对目标进行分类：特征偏振状态偏振增强将散射矩阵分解为球面、双平面、螺旋三种基本类型。极化各向异性系数极化熵利用上述项，我们试图检索目标上的信息。首先，我们设计并开发了一个全偏振FM-CW雷达系统，该系统在x波段实时工作，每秒可获取44个散射矩阵。本文还提出了一种波纹板波导作为偏振校准器。该校准器具有直线偏振特性，工作在超宽带范围内。经过对雷达系统的极化定标，得到了精确的散射矩阵。然后，将该雷达系统应用于实验室测量中的目标分类。将散射矩阵分解为3个基本分量，得到了较为满意的结果，验证了极化信息在雷达遥感中的重要性。该系统还应用于地下物体的探测。利用极化各向异性系数和特征极化状态，雷达系统对埋地目标进行了较好的分类。对于使用NASA-JPL AIRSAR、航天飞机成像雷达c /X-SAR和CRL/NASDA AIRSAR获取的合成孔径雷达（SAR）图像分析，我们引入了额外的偏振信息，即偏振熵。基于极大似然方法的熵与散射矩阵元、小波变换相结合提供了较高的分类性能。

The purpose of the research project is to classify radar targets using full vector nature of electromagnetic information. The principle of radar polarimetry based on Sinclair scattering matrix enables us to classify targets by the following items :Characteristic polarization statePolarimetric enhancementDecomposition of scattering matrix into three basic types, sphere, diplane, helix.Polarimetric anisotropy coefficientPolarimetric entropyUsing the above items, we tried to retrieve information on target.At first, we have designed and developed a full polarimetric FM-CW radar system operative in real-time mode in the X-band, which acquires 44 scattering matrices per second. A corrugated plate waveguides is also proposed as a polarimetric calibrator. The calibrator has a straight line polarization characteristic and operates in ultra wideband width. After the polarimetric calibration of the radar system, it yielded accurate scattering matrices.Then, the radar system was applied to classification of targets in a laboratory measurement. The decomposition of scattering matrix into 3 basic components provided quite satisfactory results, and it verified the importance of polarimetric information in radar remote sensing. The system has been also applied to the detection of objects in the underground. Using polarization anisotropy coefficient together with characteristic polarization states, the radar system yielded a fine classification results of buried target.For Synthetic Aperture Radar (SAR) image analysis, acquired with NASA-JPL AIRSAR, Shuttle Imaging Radar-C/X-SAR, and CRL/NASDA AIRSAR, we introduced an additional polarimetric information, i.e., polarimetric entropy. The combination of entropy with scattering matrix element, wavelet transform, based on the maximum likelihood method provided high-level classification performance.