一种针对高光谱图象识别与分类的二进编码方法

针对高光谱遥感图像中地物识别与分类问题,在对传统二进编码方法的优缺点进行分析的基础上,给出了一种新的二进编码策略,并给出了相应的快速匹配算法。该算法原理简单,易于实现,能够提高不同地物的可分性,试验结果表明,算法可以得到较好的分类效果。     

多光谱手指静脉成像与采集系统设计

多光谱手指静脉图像信息采集对提高手指静脉识别性能具有重要意义,从活体组织的不同介质对不同波长的入射光有不同的散射和吸收性能这一角度出发,建立了多光谱手指静脉成像采集系统。重点介绍基于数码管控制芯片的多波段光源控制电路而搭建的硬件系统及设计的相应软件系统。该系统不仅解决了多模态手指静脉图像成像与采集的一些基本问题,而且为多光谱手指静脉识别奠定了数据基础和实验环境。这对促进手指静脉识别技术的发展具有重要意义。     

一种基于红外图像的目标自动识别算法

红外成像制导具有在各种复杂战术环境下自主搜索、捕获、识别和跟踪目标的能力,代表了当代红外制导技术的发展趋势。提出了一种红外图像预处理、跟踪、分类的自动目标识别算法,利用小波变换、形态学方法对红外图像进行预处理,提取不同频带的惯性不变矩作为特征量,利用神经网络进行分类识别,结果表明该算法具有很高的识别率,对于精确制导武器的目标识别研究具有一定的参考价值。     

基于HSV色彩空间和光谱特征的遥感图像阴影检测方法

阴影是遥感图像的重要特征之一，遥感图像阴影检测方法存在偏蓝和暗色调非阴影误检的问题。本文提出一种基于HSV色彩空间和光谱特征的遥感图像阴影检测方法，该方法依据大气散射理论，利用海岸线波段、红边波段和近红外波段构建伪彩色图像，并转换到HSV色彩空间，构造权重修正归一化饱和度—强度指数；利用海岸线波段和近红外波段构造归一化光谱比值图像；利用权重修正归一化饱和度—强度指数和归一化光谱比值图像组建联合阴影指数；最后通过Otsu对联合阴影指数阈值分割得到最终阴影检测结果。该方法对多幅WorldView-3图像进行测试，阴影检测精度达到91%以上，相比现有同类方法提高约1%。实验结果表明本文方法能够有效缓解偏蓝和暗色调非阴影误检的问题，有助于提高阴影检测精度。     

考虑执行机构饱和的迎角跟踪控制

设计提出了 1种针对高光谱图像分类任务的 3D-MSCNN模型。在 PCA降维的基础上,利用 3D空谱特征提取网络和 2D多尺度特征提取网络实现高光谱图像特征提取,充分发挥高光谱图像空谱信息价值,增强对不同尺度地表覆盖的表达能力。最后,利用 Softmax分类损失函数实现高光谱图像分类任务。实验结果表明,本文算法在 In. dian Pines和 Pavia University数据集上都取得了较好的分类效果。与 CD-CNN、3D-CNN、SS-Net和 HybirdSN等方法相比,本文算法能够有效提升总体精度、平均精度和 Kappa系数等客观评价指标。     

基于多尺度 3D-2D卷积神经网络的高光谱图像分类

设计提出了 1种针对高光谱图像分类任务的 3D-MSCNN模型。在 PCA降维的基础上,利用 3D空谱特征提 取网络和 2D多尺度特征提取网络实现高光谱图像特征提取,充分发挥高光谱图像空谱信息价值,增强对不同尺度 地表覆盖的表达能力。最后,利用 Softmax分类损失函数实现高光谱图像分类任务。实验结果表明,本文算法在 In. dian Pines和 Pavia University数据集上都取得了较好的分类效果。与 CD-CNN、3D-CNN、SS-Net和 HybirdSN等方法相 比,本文算法能够有效提升总体精度、平均精度和 Kappa系数等客观评价指标。     

基于SS-ROI分割和CNN的交互手势识别

针对月面机器人在执行空间任务期间与航天员之间交互时的手势识别问题,提出一种基于SS-ROI分割和CNN的手势语义解读算法。利用SS算法的多策略融合优势,将待测图像生成目标候选区域,提取的各区域PHOG特征送入SVM分类器中进行分类,定位手势区域。采用7层网络结构对16种手势样本进行训练,得到迭代模型。对手势区域输入的迭代模型进行类别判断,得到识别结果。实验结果表明,算法验证识别率达到97.63%。     

基于子空间投影的未知背景航拍高光谱图像恒虚警目标检测

针对航拍高光谱图像中未知背景地物特征条件下小目标的检测问题,给出一种检测算法。利用目标的低概率特性,通过模糊聚类获取高光谱图像中背景的光谱特性;然后将高光谱数据向背景光谱信号的正交子空间及目标信号子空间投影以抑制背景和噪声信号;最后在特征层利用广义似然比检验构造出具有恒虚警特性的检测器,完成融合检测过程。理论分析和实验结果表明了算法的有效性。     

残差超复数偶对分解的多光谱和全色图像融合方法

提出一种残差超复数偶对分解的多光谱和全色图像融合方法。用超复数分别对多光谱图像和全色图像的残差图像建模,并对多光谱图像的超复数残差模型沿彩色空间的灰度轴方向分别进行超复数偶对分解,得到包含亮度信息的单部分和包含色度信息的复部分。分析表明用高分辨率全色图像的超复数残差图像来替换低分辨率多光谱图像分解后得到的复部分,可以恢复出高分辨率的多光谱图像的残差,从而实现多光谱图像和全色图像的融合。仿真结果验证了所提方法的有效性,同时验证了该方法不存在人眼可见的光谱畸变。各种现有图像融合评估方法的评估结果表明:该方法优于亮度、色度、饱和度(IHS)、主元分析（PCA）和小波变换的融合方法。     

基于改进ORB-GMS-SPHP算法的快速图像拼接方法

针对传统图像拼接算法存在拼接速度慢、图像拼接有色差等问题，提出了一种基于ORB-GMS-SPHP算法的大视场多图像拼接方法。该方法首先利用高斯函数构建尺度空间，将高斯尺度空间划分为多个网格，在每个网格内借助FAST算法提取尺度空间特征点，使用BRIEF算法提取描述符并匹配，得到更加均匀分布的特征点；然后使用运动网格统计算法筛选匹配点；最后采用SPHP算法融合图像重叠区域，从而得到完整的拼接图像。将改进的ORB-GMS-SPHP算法与现有的传统特征点匹配算法在特征点匹配精度和特征点匹配速度进行对比与评价，验证了该方法特征点匹配速度快、精度高，并且可以保留更多的正确匹配点的特点。将该拼接方法与传统拼接方法在拼接速度、图像配准均方误差RMSE以及视觉主观判断拼接色差进行对比与评价，验证了该拼接方法具有较快的拼接速度、更高的拼接精度和无明显色差。该方法在2 736像素×3 648像素图像中,特征点匹配时间降低至6.463 s，图像配准精度RMSE降低至3.87。实验证明该方法特征点匹配速度快、精度高,且拼接精度高、无明显色差。     

贴壁方柱湍流场TR—PIV实验研究

采用时变粒子图像速度场测试技术（TR—PIV），对低速循环水槽中贴壁二维方柱绕流湍流场进行了细致的测量。通过对实验得到的30000个连续瞬态速度场进行分析，得到了时均速度场和流线图谱，以及流向、法向速度分量的脉动强度场和涡量场。此外，对脉动速度时序信号进行谱分析，得到了流场中低频大尺度相干结构脱落频率。通过对瞬态速度场的分析，揭示了贴壁方柱绕流的旋涡脱落及其发展过程。     

Measurements of L-band inland-water surface-clutter Doppler spectra

Although radar surface-clutter reflectivities from terrain are generally much greater than those from water, strong Bragg resonances at low but non-zero Doppler frequencies in backscatter from small inland bodies of water might potentially cause false alarms for moving target indicator (MTI) or other Doppler signal-processing techniques designed for target detection in ground clutter. To provide data for investigating this concern, measurements of L-band radar backscatter were recorded from the surface of a small inland freshwater reservoir in central Massachusetts. These measurements were of unusually high system stability and spectral purity so as to provide up to 80 dB of available spectral dynamic range. Strong Bragg spikes occurred in the clutter Doppler spectra from the reservoir at low (3 to 4 Hz) but non-zero Doppler frequencies. This strong Bragg resonance was persistent in time and space throughout the measurements. Spectral results are presented for all four combinations of linear polarization. Comparison with tree clutter spectral results indicates that, when an occasional water body comes under surveillance at vertical polarization in otherwise generally forested terrain, water clutter spectral density is expected to exceed surrounding-terrain tree clutter spectral densities in the Bragg-offset Doppler vicinity by large amounts     

重庆大宁河特大桥成桥状态抖振响应分析

目前桥梁的抗风研究主要针对平原开阔地区,复杂山地风作用下桥梁的抗风问题研究很少。以重庆大宁河钢桁架特大拱桥为对象,通过现场实测,得到了该桥桥址处的风环境特性,包括平均风速沿高度的变化规律、湍流强度、脉动风功率谱密度函数以及湍流积分尺度等参数。根据风特性参数,计算出该桥在成桥状态下的抖振力和响应值,结果表明：即使在45m/s风速下,抖振位移响应也在公路桥梁抗风设计规范（JTG/T D60-01-2004）规定范围内。山区地貌中的平均风速沿高度的变化规律不符合公路桥梁抗风设计规范给出的指数变化率分布,湍流强度明显大于规范值。     

Far ultraviolet imaging from the IMAGE spacecraft. 2. Wideband FUV imaging

The Far Ultraviolet Wideband Imaging Camera (WIC) complements the magnetospheric images taken by the IMAGE satellite instruments with simultaneous global maps of the terrestrial aurora. Thus, a primary requirement of WIC is to image the total intensity of the aurora in wavelength regions most representative of the auroral source and least contaminated by dayglow, have sufficient field of view to cover the entire polar region from spacecraft apogee and have resolution that is sufficient to resolve auroras on a scale of 1 to 2 latitude degrees. The instrument is sensitive in the spectral region from 140–190 nm. The WIC is mounted on the rotating IMAGE spacecraft viewing radially outward and has a field of view of 17° in the direction parallel to the spacecraft spin axis. Its field of view is 30° in the direction perpendicular to the spin axis, although only a 17°×17° image of the Earth is recorded. The optics was an all-reflective, inverted Cassegrain Burch camera using concentric optics with a small convex primary and a large concave secondary mirror. The mirrors were coated by a special multi-layer coating, which has low reflectivity in the visible and near UV region. The detector consists of a MCP-intensified CCD. The MCP is curved to accommodate the focal surface of the concentric optics. The phosphor of the image intensifier is deposited on a concave fiberoptic window, which is then coupled to the CCD with a fiberoptic taper. The camera head operates in a fast frame transfer mode with the CCD being read approximately 30 full frames (512×256 pixel) per second with an exposure time of 0.033 s. The image motion due to the satellite spin is minimal during such a short exposure. Each image is electronically distortion corrected using the look up table scheme. An offset is added to each memory address that is proportional to the image shift due to satellite rotation, and the charge signal is digitally summed in memory. On orbit, approximately 300 frames will be added to produce one WIC image in memory. The advantage of the electronic motion compensation and distortion correction is that it is extremely flexible, permitting several kinds of corrections including motions parallel and perpendicular to the predicted axis of rotation. The instrument was calibrated by applying ultraviolet light through a vacuum monochromator and measuring the absolute responsivity of the instrument. To obtain the data for the distortion look up table, the camera was turned through various angles and the input angles corresponding to a pixel matrix were recorded. It was found that the spectral response peaked at 150 nm and fell off in either direction. The equivalent aperture of the camera, including mirror reflectivities and effective photocathode quantum efficiency, is about 0.04 cm². Thus, a 100 Rayleigh aurora is expected to produce 23 equivalent counts per pixel per 10 s exposure at the peak of instrument response.     

The importance of spectroscopy in the 80–800 å region for plasma diagnostics in the solar atmosphere

We discuss the importance of the spectral range from about 80 to 800 Å for determining physical conditions in different regions of the solar atmosphere. We give examples of line ratios that may be used to determine electron densities in quiet Sun regions, active regions, and flares. We discuss the possibility of determining electron temperatures from line ratios in the EUV.We show that profiles as well as intensities of spectral lines must be obtained for a proper interpretation of the spectra. We give approximate parameters for a solar grazing incidence spectrograph suitable for the study of the 80–800 Å wavelength region.Given as an invited review paper at the EGAS Meeting in Munich, Germany, 11–14 July 1978.     

Locating forward opportune landing sites using spectral images fromsatellites

The ability of future combat airlifters to deliver cargo into short, austere opportune landing sites which are very close to operating field units is expected to greatly improve the effectiveness of the US warfighter. But how do we find and select those sites in a quickly-changing battle scenario? The use of spectral image data from satellites such as EarthWatch as the medium for identifying opportune landing sites is the concept of interest and is described. A study was performed to explore the feasibility and data requirements for this concept. The study's objectives were two-fold: (1) evaluate and demonstrate airlifter value; and (2) evaluate and demonstrate opportune landing site selection concept feasibility. Spectral data was collected in several image resolutions for selected areas within the US. Potential opportune landing sites were selected across large areas and the selection was refined based on higher resolution spectral images. Site surveys were used to grade the accuracy of selection. The number of opportune landing sites (as short as 600 feet) relative to traditional runway lengths was determined and compared     

Thermal cyclotron radiation from a hot coronal loop with helical magnetic field

The spectral and polarization properties of thermal cyclotron radio emission from a hot coronal loop with a current along the axis are computed. The magnetic field is supposed to have a component along the loop axis as well as a poloidal part due to the current, both components being of comparable magnitude. In this specific configuration a helical magnetic field is present with a remarkable minimum of its absolute value along the loop axis and a maximum at its periphery. The presence of one or two maxima of magnetic field value along the line of sight results in increasing optical thickness of the gyroresonance layers at appropriate frequencies in the microwave band and, therefore, in enhanced radio emission at those harmonics which are optically thin (for example,s=4). These cannot be observed in models with the commonly employed magnetic field configuration (longitudinal along the loop axis).We show that the frequency spectrum of thermal cyclotron radiation from a hot coronal loop with a helical magnetic field differs from that of the standards-component source (with smooth frequency characteristics and polarization corresponding toe-mode) in that plenty of fine structures (line-like features and cut-offs) are present and theo-mode is prevalent in some frequency intervals. The enhanced radio emission at high harmonics and the complicated form of frequency spectrum in the model considered imply that some microwave sources, which are poorly explained in traditional models of solar active regions, may be associated with helical magnetic fields in hot coronal loops. Computations allow one to indicate spectral and polarizational peculiarities of local sources testifying to the presence of a helical magnetic field.     

一种基于预测和变换混合设计的超光谱图像压缩方法

提出了一种新的超光谱图像压缩方法。该方法针对三维小波压缩编码方法对专有数据的特殊属性利用效率较低的缺陷,设计了一种应用于超光谱图像的预测编码和变换编码相混合的实现方案。首先,由超光谱图像的高谱间相关性,推导出图像各波段间的一般表达式。然后,根据这一表达式建立了超光谱图像的波段预测方式。该预测方式以波段为单位,参考波段采用小波变换编码技术进行压缩。为了提高预测的精度,参考波段选择器和波段预测器分别基于超光谱图像的数学统计规律和各波段直方图形状相似的特性进行设计。最后,将预测偏差值通过三维小波编码技术进行压缩。实验结果表明,本文设计的方法与目前先进的超光谱压缩技术相比能够提供具有竞争力的性能提升,且具有良好的兼容性、灵活性和渐进传输能力。     

Feature enhancement and ATR performance using nonquadratic optimization-based SAR imaging

We present an evaluation of the impact of a recently proposed synthetic aperture radar (SAR) imaging technique on feature enhancement and automatic target recognition (ATR) performance. This image formation technique is based on nonquadratic optimization, and the images it produces appear to exhibit enhanced features. We quantify such feature enhancement through a number of criteria. The findings of our analysis indicate that the new feature-enhanced SAR image formation method provides images with higher resolution of scatterers, and better separability of different regions as compared with conventional SAR images. We also provide an ATR-based evaluation. We run recognition experiments using conventional and feature-enhanced SAR images of military targets, with three different classifiers. The first classifier is template based. The second classifier makes a decision through a likelihood test, based on Gaussian models for reflectivities. The third classifier is based on extracted locations of the dominant target scatterers. The experimental results demonstrate that the new feature-enhanced SAR imaging method can improve the recognition performance, especially in scenarios involving reduced data quality or quantity.     

基于图像处理思想的自适应网格技术

为了提高激波和旋涡等复杂流动现象的捕捉精度和计算效率,基于图像处理的思想,研究了非结构网格的自适应加密和稀疏算法.将流场计算网格与图像像素类比,借鉴图像处理的方法,提出了一种流场自适应区域的分割算法.该方法能自动对自适应区进行分层,中间区域网格可以达到原来多次自适应的效果,而且提高了生成网格的质量.将算法嵌入到流场求解器中,根据流场求解结果对网格进行自适应,并使用算例进行了验证.