Research on High Resolution Satellite Image Classification Algorithm based on Convolution Neural Network

Artifi cial neural network is a kind of artificial intelligence method to simulate the function of human brain, and deep learning technology can establish a depth network model with hierarchical structure on the basis of artificial neural network. Deep learning brings new development direction to artificial neural network. Convolution neural network is a new artificial neural network method, which combines artificial neural network and deep learning technology, and this new neural network is widely used in many fields of computer vision. Modern image recognition algorithm requires classifi cation system to adapt to different types of tasks, and deep network and convolution neural network is a hot research topic in neural networks. According to the characteristics of satellite digital image, we use the convolution neural network to classify the image, which combines texture features with spectral features. The experimental results show that the convolution neural network algorithm can effectively classify the image.

Information retrieval methods for high resolution γ-ray spectra

A program based on MATLAB 7.0 platform was developed to locate characteristic peak position and calculate net area of characteristic peak.The formula for the calculation of relative standard deviation of net peak area by Sterlinski’s method was found excellent in searching single peaks and resolving overlapping peaks in high resolution gamma-ray spectrum.Gaussian function fitting method using Levenberg-Marquardt algorithm was applied to calculate net area of peaks.A standard test spectrum supplied by the IAEA in 1995 was analyzed by the program and another two widely used commercial software.The analysis results show the program was superior to the latter two in searching single peaks and resolving overlapping peaks.The optimized fitting indexes are found between 0.962 and 0.996,which shows that the program adopted is feasible and accurate for extracting the net peak area in high resolution gamma-ray spectra.

Automatic Extraction of Urban Road Centerlines from High-Resolution Satellite Imagery Using Automatic Thresholding and Morphological Operation Method

The commercial high-resolution imaging satellite with 1 m spatial resolution IKONOS is an important data source of information for urban planning and geographical information system (GIS) applications. In this paper, a morphological method is proposed. The proposed method combines the automatic thresholding and morphological operation techniques to extract the road centerline of the urban environment. This method intends to solve urban road centerline problems, vehicle, vegetation, building etc. Based on this morphological method, an object extractor is designed to extract road networks from highly remote sensing images. Some filters are applied in this experiment such as line reconstruction and region filling techniques to connect the disconnected road segments and remove the small redundant. Finally, the thinning algorithm is used to extract the road centerline. Experiments have been conducted on a high-resolution IKONOS and QuickBird images showing the efficiency of the proposed method.

High resolution 3D nonlinear integrated inversion

The high resolution 3D nonlinear integrated inversion method is based on nonlinear theory. Under layer control, the log data from several wells （or all wells） in the study area and seismic trace data adjacent to the wells are input to a network with multiple inputs and outputs and are integratedly trained to obtain an adaptive weight function of the entire study area. Integrated nonlinear mapping relationships are built and updated by the lateral and vertical geologic variations of the reservoirs. Therefore, the inversion process and its inversion results can be constrained and controlled and a stable seismic inversion section with high resolution with velocity inversion, impedance inversion, and density inversion sections, can be gained. Good geologic effects have been obtained in model computation tests and real data processing, which verified that this method has high precision, good practicality, and can be used for quantitative reservoir analysis.

基于函数−特征子空间的水下高分辨方位估计算法

针对有限阵列孔径下传统方位估计方法对低频目标方位分辨能力弱的问题,提出一种基于函数−特征子空间理论的高分辨方位估计算法。该方法在特征子空间算法的基础上,改进其投影扫描向量的子空间分解形式,并在计算流程中引入函数波束形成算法,获得基于指数控制的函数−特征子空间方位估计算法。理论分析与仿真结果表明,该算法涵盖了常规的特征子空间算法,且当指数绝对值小于1时,该算法在保证抗噪性能的前提下,空间方位谱具有更低的旁瓣和更窄的主瓣。试验数据结果表明,该算法在分辨能力方面优于特征子空间等波束形成方法,具有较好的实际应用前景。

基于波束补偿的成像增强算法研究

近场高分辨率成像的需求日益增加,传统的后向投影成像算法忽略了周围复杂电磁环境给成像带来的不良影响,在对回波进行相干积累的过程中不可避免地将周围杂波映射到目标图像网格上,导致图像质量较差。针对这一问题,提出一种基于天线方向图波束补偿的后向投影成像增强技术。该技术通过提前生成自适应能量衰减系数矩阵,对目标回波幅度进行衰减补偿,能有效抑制复杂电磁环境下强散射点给成像过程带来的不良影响。实测及实验结果表明,所提方法不仅在抑制杂波的前提下增强了图像质量,也能降低成像过程中相干积累的次数,提高了成像效率。

基于高分辨率再分析海流数据的三种涡旋识别方法在南海的评估

本文基于南海11 a逐日高分辨率再分析海流数据,利用三种涡旋识别方法——缠绕角(Winding Angle,WA)法、速度矢量几何(VectorGeometry,VG)法和OW(OkuboWeiss)参数法对南海中尺度涡进行识别和追踪,对比了三种方法对中尺度涡的探测能力,并分析了在高分辨率再分析产品中的适用性。研究表明,产品空间分辨率的提高能够放大OW方法的W参数噪声,使该方法存在涡旋过量检测和涡旋分割现象,导致识别的涡旋数量偏多、半径偏小,涡旋平均成功探测率最低(76.2%);与OW方法相比,WA方法在一定程度上降低了涡旋过量探测率和漏判率,涡旋平均成功探测率提高至85.1%,但涡旋识别时间较长;与前两种方法相比,VG方法的涡旋平均成功探测率可达93.2%,综合评估参数(过量探测率×漏判率÷成功探测率)优势显著(4.5%),且计算高效。因此,在基于高分辨率再分析产品进行中尺度涡识别时,VG方法具有更加合理的涡旋探测结果和更强的适用性。

基于空间协模拟的叠前地震随机反演方法

无论是迭代类随机反演还是线性贝叶斯随机反演,一般采用序贯类随机模拟表征油藏的非均质性。序贯类方法大多依赖变差函数或训练图像描述模型参数的空间相关性,需要逐点计算模拟结果,因此并行实现困难、计算效率低。为此,将一种条件化快速傅里叶变换滑动平均(FFT⁃MA)模拟引入线性反演框架,提出基于空间协模拟的叠前地震随机反演方法。首先,在贝叶斯框架下整合地震数据和测井低频信息,获得弹性参数的后验概率分布。然后,根据FFT⁃MA算法生成概率场,以测井数据为条件数据对贝叶斯后验应用概率场协同模拟,从而获得井震数据约束的叠前弹性参数高分辨率随机反演结果。该方法无需迭代更新模型参数,能极大地提高随机反演的计算效率。最后,由模型试算和实际资料应用测试方法的效果。模型试算表明,所提方法在高分辨率储层预测及计算效率方面均优于常规方法,可稳定、准确地描述小尺度储层特征。实际资料应用表明,所提方法得到的高分辨反演结果与测井数据契合良好,提高了随机反演在薄储层定量表征方面的实用性。

应用遗传算法估计非稳态地震数据的混合相位子波及Q值

地震高分辨率处理或反演的目的是获得精确的反射系数或弹性参数模型。然而地层滤波效应模糊了地震记录中的地层反射信息,因此有必要消除这种滤波效应。现有的大部分提高分辨率方法并不能完全摒弃关于地震子波和Q模型的一些假设。为了获得更为切合实际的地震子波,同时自适应获取Q模型,文中将地震子波相位估计与Q模型估计相结合,提出了基于遗传算法的非稳态地震数据混合相位子波及Q值估计方法。首先通过井旁道记录拟合得到初始地震子波的振幅信息,然后依据子波Z变换的根关于单位圆移动与否构建用于遗传算法的编码链条。另一方面,十进制Q模型对应的二进制表示形式同样能够利用编码链条表征,因此利用全局优化算法能够同时估计地震混合相位子波以及Q模型。结合根变换和遗传算法不断改变子波相位的同时自适应生成Q模型,利用子波和Q模型构造的时变子波矩阵与测井反射系数得到合成地震记录,并与井旁道记录进行匹配,最终得到合理的混合相位子波和地层Q模型,进而构造时变子波矩阵进行时变反褶积。通过井旁道记录与测井数据拟合得到的混合相位子波,与实际地震子波相位更为接近,理论数据和实际数据处理结果证实了该方法的有效性。

Anti-swarm UAV radar system based on detection data fusion

There is a growing body of research on the swarm unmanned aerial vehicle(UAV)in recent years,which has the characteristics of small,low speed,and low height as radar target.To confront the swarm UAV,the design of anti-UAV radar system based on multiple input multiple output(MIMO)is put forward,which can elevate the performance of resolution,angle accuracy,high data rate,and tracking flexibility for swarm UAV detection.Target resolution and detection are the core problem in detecting the swarm UAV.The distinct advantage of MIMO system in angular accuracy measurement is demonstrated by comparing MIMO radar with phased array radar.Since MIMO radar has better performance in resolution,swarm UAV detection still has difficulty in target detection.This paper proposes a multi-mode data fusion algorithm based on deep neural networks to improve the detection effect.Subsequently,signal processing and data processing based on the detection fusion algorithm above are designed,forming a high resolution detection loop.Several simulations are designed to illustrate the feasibility of the designed system and the proposed algorithm.

一种奇异值分解与子空间加权联合的改进MUSIC算法

在低信噪比、小快拍数等非理想条件下,经典DOA估计算法对邻近目标的分辨率严重下降,甚至失去分辨能力。针对这一问题,提出了一种将重构的接收信号协方差矩阵进行奇异值分解并与改进的加权子空间方法相结合的改进算法。该算法充分利用互相关信息构建新的接收信号协方差矩阵,并对噪声子空间信息采用新的校正方法,对噪声特征值进行改造,之后对噪声子空间进行加权,最后与信号子空间加权技术相联合。仿真实验证明,改进算法在低信噪比和小快拍数条件下可以分辨间隔4°的相邻目标,统计分析表明该算法的分辨率明显优于经典MUSIC算法。

基于关键点检测的坐姿识别方法

针对直接应用深度学习分类算法所得模型泛用性较差的问题,提出了一种分两个步骤完成人体坐姿识别的方法。所提方法首先提取图片中人体上身关键点的坐标信息,在关键点检测环节采用高分辨率主干网络,并进一步改进了模型结构,在下采样环节中引入SE(Squeeze-and-Excitation)注意力机制,加强了空间位置特征的表达,取得了更高的检测平均精准度;然后采用随机森林算法对关键点进行坐姿分类。实验结果表明:所提方法识别准确率可以达到94%以上,并且在陌生场景下有更好的泛用性,能适应实际应用中复杂的人物环境。

CT高分辨算法与标准算法图像重建密度测量对亚厘米肺结节鉴别诊断的临床价值

目的探讨高分辨算法与标准算法重建图像对亚厘米肺结节CT值测量的差异性以及对良恶性鉴别诊断的价值。方法选取我院有手术病理结果的高分辨算法薄层重建图像测量CT值≥0 HU(根据CT值定义公式,水的CT值为0 HU)的亚厘米肺结节患者98例,且同时具有标准算法薄层重建图像。为进一步区分其内完全实性与少量含气结节,以水的CT值为界限,按照标准算法重建图像测量CT值结果≥0 HU(n=52)、<0 HU(n=46)分成两组,比较两组不同重建算法测得病灶CT值的差异,以及病理良恶性概率变化,分析诊断效能。结果全部病灶高分辨算法CT值均高于标准算法,其差值在两组间差异有统计学意义(P<0.01)。高分辨算法测得CT值≥0 HU的全部入组病例中,恶性占37%(36/98),良性占63%(62/98)。采用标准算法测得的病灶CT值仍然≥0 HU的病灶共52例,良性占比升至85%(44/52);病灶CT值变为<0 HU共46例,恶性占比升至61%(28/46)。ROC曲线显示标准算法图像CT值对病灶良恶性的鉴别诊断效能高于高分辨算法。结论采用标准算法重建图像观察和测量病灶的密度较高分辨算法更为真实准确。

面向高分辨率激光扫描点云的图像色彩界面交互配色设计研究

配色图像样本杂色的存在会使得配色方案与源图像标准颜色存在较大的误差,导致配色协调舒适度较差。针对上述问题,进行面向高分辨率激光扫描点云的图像色彩界面交互配色设计研究。该研究中利用高分辨率激光扫描技术采集标准点云数据并作为参考样本,针对点云数据实施去噪处理,以减少杂色的干扰。将点云的RGB数据转换为HSV数据并参考表伯克霍夫美度值建立配色模型。将果蝇算法和粒子群算法结合在一起,避免陷入局部最优,利用果蝇-粒子群算法求取配色模型最优解,得出图像色彩界面交互配色最佳设计方案。结果表明:配色方案2的伯克霍夫美度值为0.976,要大于其他5个方案,由此说明配色方案2是配色模型的最优解,配色效果达到最佳,配色更加协调舒适。

基于高分遥感图像处理技术的公路勘查方法研究

针对公路高分遥感勘测技术中,遥感图像易受天气干扰、数据分析精确度低等问题,研究了基于高分遥感图像处理技术的公路勘查方法。设计了双行信息采样电路以及采样信息保持电路,实现了公路勘查过程中的高清图像信息采集,由此避免因自身电源频率、外部磁场等因素的干扰而产生图像噪点,保证采集图像信息的准确性。此外,采用一种暗通道先验模型去云雾技术,能够有效排除云雾天气对遥感图像采集的干扰。通过基于Storm的大数据算法,提高系统的整体数据分析及传输能力。试验结果表明,该方法不仅增强了去云雾能力,还可以还原正常图像85%左右的图像特征点。该方法具有很强的数据处理、抗干扰以及数据传输能力。

基于信号幅度的^(3)He管中子探测器刻度算法优化研究

本研究针对中国散裂中子源下的高压中子衍射谱仪中核心组件——位置灵敏型^(3)He管中子探测器的刻度问题进行了深入探究。通过实验验证和理论分析,揭示了传统刻度方法中未考虑到的中子信号幅度对刻度参数的影响,以及由此产生的位置分布畸变和空间分辨率降低的问题。为解决这些问题,本研究提出了一种改进的刻度算法,该算法能够有效地考虑中子信号幅度的变化,从而显著改善了探测器系统的空间分辨能力。经过实验结果分析显示,新的刻度算法不仅消除了位置分布的拖尾现象,而且提高了不同位置处的空间分辨率,与理论预期相符。本研究的成果为高压中子衍射谱仪在极端条件下的应用提供了重要的技术支持,并为进一步优化中子探测器的刻度方法和提高中子衍射技术的应用精度奠定了基础。

高分辨率地震勘探数据的噪声抑制算法研究

近年来,随着地震勘探技术的发展,高分辨率地震勘探数据成为地震勘探的重要数据源,这些数据提供了更细致、更清晰的地下结构信息,对于提高资源勘探的精度和效率具有重要意义。然而,在地震数据处理过程中,不可避免地存在噪声,严重影响了地震数据的分辨率和准确性。因此,为获得高分辨率地震数据,噪声抑制的重要性显而易见。基于此,详细讨论各类噪声的特性,探讨当前主要噪声抑制算法,比较这些算法的性能,对提高高分辨率地震勘探数据的质量,进一步揭示地下复杂结构信息,提高地震勘探和资源勘查准确性具有重要意义。

Feature Extraction of Localized Scattering Centers Using the Modified TLS-Prony Algorithm and Its Applications

This paper presents an all-parametric model of radar target in optic region, in which the localized scattering center's frequency and aspect angle dependent scattering level, distance and azimuth locations are modeled as the feature vectors. And the traditional TLS-Prony algorithm is modified to extract these feature vectors. The analysis of Cramer-Rao bound shows that the modified algorithm not only improves the restriction of high signal-to-noise ratio(SNR)threshold of traditional TLS-Prony algorithm, but also is suitable to the extraction of big damped coefficients and high-resolution estimation of near separation poles. Finally, an illustrative example is presented to verify its practicability in the applications. The experimental results show that the method developed can not only recognize two airplane-like targets with similar shape at low SNR, but also compress the original radar data with high fidelity.

IMPROVEMENT OF PHASE GRADIENT AUTOFOCUS ALGORITHM

The phase error estimated by phase gradient autofocus(PGA) is not based on a finite order polynomial mode, so PGA has a good autofocus property for arbitrary order phase error and is fit for high resolution airborne SAR. But PGA has two shortcomings: first, it has a worse estimation property for fast changing phase error; second, there exists a section of linear phase in the phase error estimated by this algorithm. This paper introduces the idea of rank one phase estimate (ROPE) autofocus technique, and improves PGA. The improved PGA(IPGA) can successfully overcome both these shortcomings of PGA.

基于强度调制的偏振光谱测量技术

针对现阶段强度调制偏振光谱测量技术在光程差域内不同通道间存在频谱信息串扰的不足,采用加窗插值傅里叶变换法,研究了不同切趾函数对强度调制偏振光谱测量技术的影响。首先,进行强度调制偏振光谱测量技术理论分析,完成强度调制模块的设计;其次,根据强度调制模块的设计指标,模拟入射光的光强信息及采用不同切趾函数的信息解调复原过程;最终,对强度调制偏振光谱测量系统进行仿真,根据仿真模型结合高分辨率光谱仪、平行光管等器件完成偏振光谱测量装置的搭建,进行偏振光谱测量实验。研究结果表明,偏振光谱测量系统设计指标选择合适的切趾函数可以减少光强信息截断过程中造成的频谱能量泄露,使光程差域内不同通道之间的串扰信息减少、偏振光谱复原精度提高,不同切趾函数切趾处理后线偏振光的偏振度最小误差由0.0768减小到0.0014,偏振度接近1。