宽带雷达一维高分辨距离像识别研究

对一维高分辨率距离像（HRRP）进行预处理,解决高分辨距离像姿态、平移和幅度敏感性问题。对HRRP进行了目标子空间提取,基于子空间使用最大相关系数法对目标进行识别。实验结果表明,基于子空间法的目标识别具有较好的识别结果和较快的处理速度。     

基于主分量分析的一维距离像雷达目标识别

一维距离像是自动目标识别的一种重要特征,它对目标姿态变化很敏感,只有通过进一步处理提取稳定特征才能够有效用于识别。针对距离像的这种姿态敏感性,首先分析了主分量分析(PCA)的降噪原理与核主分量分析(KPCA)的特征提取能力,然后提出先用PCA滤波对一维距离像降噪再用KPCA提取较大姿态角范围内稳定特征的雷达目标一维距离像识别框架,并用四类目标的实测数据进行分类实验,表明该算法确实能够提高识别性能。     

基于改进小波降噪的雷达一维距离像识别

针对低信噪比下雷达目标一维距离像质量不高、影响目标识别率的问题,将小波阈值降噪的方法应用到雷达目标一维距离像识别研究中,设计了一种新的小波阈值函数,提出了基于小波阈值降噪的雷达一维距离像识别的方法。利用仿真数据进行实验验证,以LVQ（Learning Vector Quantization）神经网络作为分类识别器,进行目标的分类识别研究。结果表明,将小波阈值降噪用于雷达目标一维距离像识别,在低信噪比时能够有效地降低噪声,提高距离像的质量,从而提高目标一维距离像识别率,同时实验也验证了所提出的新阈值函数相较于原阈值函数能更加有效地降低噪声,提高识别率。     

基于LVQ网络优化的雷达目标识别算法

目标一维距离像在雷达目标识别领域中具有很高的研究价值,神经网络有很强的自适应能力,被广泛应用于目标识别领域中。通过研究分析,将学习向量量化（Learning Vector Quantization, LVQ）神经网络应用于雷达目标一维距离像识别。针对LVQ神经网络对初始连接权值敏感的问题和如何增强网络的分类识别性能,提出利用粒子群优化（Particle Swarm Optimization, PSO）算法对其进行优化。在此基础上提出了基于PSO-LVQ神经网络的雷达目标一维距离像识别新方法。通过3类飞机实测数据实验,验证了PSO算法优化LVQ神经网络初始连接权值的可行性和PSO-LVQ识别算法的有效性。     

基于卷积神经网络的雷达目标HRRP分类识别方法

卷积神经网络通过卷积和池化操作提取图像在各个层次上的特征进而对目标进行有效识别,是深度学习网络中应用最广泛的一种。文中围绕一维距离像雷达导引头自动目标识别,开展基于卷积神经网络的目标高分辨距离像分类识别方法研究。首先,基于空中目标一维距离像姿态敏感性仿真生成近似平行交会条件下不同类型目标的高分辨距离像数据集;其次,构建一种一维卷积神经网络结构对目标高分辨距离像进行分类识别;作为比较,针对同类高分辨距离像数据集,分析了主成分分析-支持向量机方法的目标分类识别效果。结果表明:基于卷积神经网络的目标分类识别算法有更好的识别能力,对高分辨距离像的姿态敏感性具有较强的适应性。     

基于一维距离像的雷达目标识别研究进展

现代高分辨雷达为雷达目标识别提供了新的途径,相应的高分辨一维距离像具有易获取和处理的特点,同时还能够揭示出目标径向结构的分布特性,因而在目标识别领域具有良好的应用前景.文中分析了基于一维距离像的雷达目标识别研究现状,包括一维距离像的模型、距离像特性分析、特征提取方法以及分类器的设计方法等,并且指出了高分辨一维距离像目标识别今后的研究方向.     

一种获取雷达目标一维距离像的方法

利用高分辨一维距离像进行目标识别在现代雷达中已成为一个重要的研究课题。本文基于雷达信号设计和互相关接收处理技术,研究了一种利用目标回波,逐次获取一维距离像的方法。与冲激雷达相比,这种方法能显著提高信噪比,并且这个方法的收敛速度快,一般迭代10次以内即达到稳定。     

基于一维距离像的雷达目标识别方法研究

通过对光学区雷达目标一维距离像的介绍和分析,指出利用一维距离像进行雷达目标分类和识别的可行性,并针对一维距离像对姿态角度化敏感这一难点问题,提出两种比较实用的解决方案。     

一种改进的雷达高分辨距离像统计识别方法

基于雷达目标一维高分辨距离像的统计目标识别中,需解决两大问题:其一是如何处理距离像对姿态敏感和平移敏感,其二是如何准确地描述距离像的统计特征.直接将一维距离像用于目标识别通常很难取得好的识别效果.本文将高斯混合模型(GMM)应用到空中目标高分辨一维距离像统计建模中,提出了一种改进的高斯混合模型模糊聚类分析方法并用于目标识别.与传统的k-means聚类算法的实验结果比较表明,该方法是有效、稳健的,在低信噪比条件下具有较好的识别效果.     

基于神经网络的雷达目标识别

本文讨论了基于径向基函数网络（ＲＢＦＮ）的雷达目标识别问题，在分析了一维距离象特点的基础上，提出了采用非相关幅度平均一维距离象以获取稳定模式这一有效方法，在指出传统经验公式局限性后，给出了一种基于训练样本空间分布来估计高斯函数形状参数的方法，用微波暗室试验数据进行转台成象并对一维距离象三种模式进行识别分类的结果表明，本文所提出的方法用于研究雷达目标识别是有效的。     

Computing advective velocities from satellite images of sea surfacetemperature

A method of employing sequential satellite images of sea surface temperature (SST) patterns to estimate surface advective current velocities is described. SST images are obtained by processing data received twice daily from the Advanced Very High Resolution Radiometer. The current velocities are estimated by applying a maximum cross correlation (MCC) technique on two time-lapsed images. The MCC technique involves computing matrices of cross correlation coefficients and identifying correlation peaks. A two-stage multilevel statistical test over two-dimensional correlations is developed for determining the relative significance levels of velocity estimates. The test also identifies areas where the MCC technique cannot be effective. Aspects of implementation and limitations of the MCC technique for computing current velocity are also discussed. Advective velocity fields computed through MCC for the Chatham Rise area in New Zealand extending from (40.5°S, 173.5°E) to (49°S, 178°W) show good agreement with the known geostrophic flow patterns in this area     

基于显著性及主成分分析的红外小目标检测

将红外小目标检测作为目标与背景的二分类问题.先根据点扩散函数原理,仿真生成红外小目标训练样本,再用主成分分析方法提取目标样本的主特征,建立目标的主成分空间.对测试样本,只要判断其在主成分空间的重构残差,便可识别其是否为目标.为了提高算法的实时性,提出了一种基于显著性和主成分分析的红外小目标检测算法,先通过频域残差方法检测目标可能存在的显著性区域,再在此区域内做识别.实验结果证明该方法快速、有效.     

A Kernel PCA Shape Prior and Edge Based MRF Image Segmentation

We introduce both shape prior and edge information to Markov random field (MRF) to segment target of interest in images.Kernel Principal component analysis (PCA) is performed on a set of training shapes to obtain statistical shape representation.Edges are extracted directly from images.Both of them are added to the MRF energy function and the integrated energy function is minimized by graph cuts.An alignment procedure is presented to deal with variations between the target object and shape templates.Edge information makes the influence of inaccurate shape alignment not too severe,and brings result smoother.The experiments indicate that shape and edge play important roles for complete and robust foreground segmentation.     

子空间模型下的仿射不变目标跟踪

针对目标跟踪过程中目标可能出现的快速变化和严重遮挡等问题,提出了一种基于新的子空间表示的目标跟踪算法。采用距离不变量对尺度不变特征变换(SIFT)特征点匹配对进行提纯。用提纯后的特征点匹配对,通过线性拟合得到仿射变化参数。在粒子滤波的理论框架下,采用快速的迭代算法,建立目标的主分量(PCA)子空间表示,结合计算得到的仿射变化参数,构造有效的目标观测模型完成跟踪。同时,采用在线学习的方法对SIFT特征点和PCA子空间进行定时更新。大量实验表明,提出的算法能快速有效地完成对姿态和形状剧烈变化的目标的精确跟踪。     

Robust proportionate adaptive filter based on maximum correntropy criterion for sparse system identification in impulsive noise environments

Proportionate-type adaptive filtering (PtAF) algorithms have been successfully applied to sparse system identification. The major drawback of the traditional PtAF algorithms based on the mean square error (MSE) criterion show poor robustness in the presence of impulsive noises or abrupt changes because MSE is only valid and rational under Gaussian assumption. However, this assumption is not satisfied in most real-world applications. To improve its robustness under non-Gaussian environments, we incorporate the maximum correntropy criterion (MCC) into the update equation of the PtAF to develop proportionate MCC (PMCC) algorithm. The mean and mean square convergence performance analysis are also performed. Simulation results in sparse system identification and echo cancellation applications are presented, which demonstrate that the proposed PMCC exhibits outstanding performance under the impulsive noise environments.     

Reducing the dimensionality of plant spectral databases

Ground-based measurements of plant reflectance and transmittance are essential for remote sensing projects oriented toward agriculture, forestry, and ecology. This paper examines the application of principal components analysis (PCA) in the storage and reconstruction of such plant spectral data. A novel piecewise PCA approach (PPCA), which takes into account the biological factors that affect the interaction of solar radiation with plants, is also proposed. These techniques are compared through experiments involving the reconstruction of reflectance and transmittance curves for herbaceous and woody specimens. The spectral data used in these experiments were obtained from the Leaf Optical Properties Experiment (LOPEX) database. The reconstructions were performed aiming at a root-mean-square error lower than 1%. The results of these experiments indicate that PCA can effectively reduce the dimensionality of plant spectral databases from the visible to the infrared regions of the light spectrum, and that the PPCA approach can further maximize the accuracy/cost ratio of the storage and reconstruction of plant spectral reflectance and transmittance data.     

有监督的高光谱图像伪装目标检测方法

针对伪装目标检测问题,提出了一种有监督的高光谱伪装目标检测方法。以植被型伪装目标为研究对象,在分析伪装材料与绿色植被光谱之间特性的基础上,先通过光谱重排、光谱微分以及光谱差异性增强处理,对植被型伪装材料与真实植被（背景）之间的光谱差异进行放大,然后利用主成分分析（PCA）变换进行降维,从而实现了一种适用于大面积植被型伪装目标的高光谱检测方法。实验结果表明,该检测方法在检测时间和检测效果上要优于基于加权的约束能量最小化法（WCM-CEM）和基于非监督目标生成处理的正交子空间投影法（UTGP-OSP）。     

相似性约束的深度置信网络在SAR图像目标识别的应用

特征提取是合成孔径雷达(SAR)图像目标识别的关键环节。SAR图像中存在的相干斑点和非光滑特性使得传统针对光学图像的特征提取方法变得很难应用。虽然可以采用深度置信网络(DBN)自动地进行特征学习,但是该方法属于无监督学习方法,这使得学习到的特征与具体的任务是无关的。该文提出一种叫做相似性约束的受限玻尔兹曼机模型。该模型在学习过程中通过约束特征向量之间的相似性达到引入监督信息的目的。另外,可以将多个相似性约束的受限玻尔兹曼机堆叠成一种新的深度模型,称其为相似性约束的深度置信网络模型。实验结果表明在SAR图像目标识别应用中,该方法相比主成分分析(PCA)以及原始DBN具有更好的识别性能。     

Computing Coastal Ocean Surface Currents From Infrared and Ocean Color Satellite Imagery

Many previous studies have demonstrated the viability of estimating advective ocean surface currents from sequential infrared satellite imagery using the maximum cross-correlation (MCC) technique when applied to 1.1-km-resolution Advanced Very High Resolution Radiometer (AVHRR) thermal infrared imagery. Applied only to infrared imagery, cloud cover and undesirable viewing conditions (gaps in satellite data and edge-of-scan distortions) limit the spatial and temporal coverage of the resulting velocity fields. In addition, MCC currents are limited to those represented by the displacements of thermal surface patterns, and hence, isothermal flow is not detected by the MCC method. The possibility of supplementing MCC currents derived from thermal AVHRR imagery was examined, with currents calculated from 1.1-km-resolution Moderate Resolution Imaging Spectroradiometer (MODIS) and Sea-viewing Wide Field-of-view Sensor (SeaWiFS) ocean color imagery, which often have spatial patterns complementary to the thermal infrared patterns. Statistical comparisons are carried out between yearlong collections of thermal and ocean color derived MCC velocities for the central California Current. It is found that the image surface patterns and resulting MCC velocities complement one another to reduce the effects of poor viewing conditions and isothermal flow. The two velocity products are found to agree quite well with a mean correlation of 0.74, a mean rms difference of 7.4 cm/s, and a mean bias less than 2 cm/s which is considerably smaller than the established absolute error of the MCC method. Merging the thermal and ocean color MCC velocity fields increases the spatial coverage by approximately 25% for this specific case study