Improved Three-Dimensional Image Reconstruction Technique for Multi-Component Ground Penetrating Radar Data

For electromagnetic imaging the vectorial character of the emitted field and the radiation characteristics of both source and receiver play an important role. Recently, a new imaging algorithm was presented dedicated to the electromagnetic case. The radiation characteristics of GPR source and receiver antennas and the vectorial nature of the electromagnetic waves are taken into account for a monostatic fixed offset GPR survey. This resulted in a representative image of a point scatterer. Comparison with scalar imaging algorithms shows that for a homogeneous space the SAR image has an opposite sign compared to the multi-component image, whereas the Gazdag image has a phase shift of about 90° with respect to the multi-component image. In this paper, modified scalar imaging algorithms are introduced that minimize these differences. However, between the images obtained in a homogeneous half-space phase differences of 10–20° are still present. These differences indicate the possible error in nature of the physical property contrast when it is determined with the modified scalar imaging algorithms. The multi-component imaging algorithm returns a representative image because it uses the appropriate Greens functions to eliminate the propagation effects. For practical imaging strategies, only far-field radiation characteristics can be used to compensate for the propagation effects due to the large computing time needed to evaluate the total-field expressions. Synthetic analysis of the imaging of a point scatterer calculated using total-field expressions shows that using the far-field expressions in the multi-component imaging algorithm the images better approximate the actual contrast than using the modified scalar imaging algorithms. Experimental results are presented from imaging several buried objects with different medium properties and different orientations. The phase differences in the experimental data are similar to those obtained synthetically. This likeness indicates that using the multi-component imaging algorithm, a more reliable image is obtained than with the modified scalar imaging algorithms.     

融合五自由度位姿信息的单目内窥镜目标尺寸测量

针对低分辨率、弱纹理、缺少参照物的小场景下内窥镜图像目标测量困难问题,提出了融合5自由度电磁传感器的单目内窥镜下目标尺寸测量方法。首先,分析了基于5自由度传感器的内窥镜定位原理,得到内窥镜沿主光轴方向的位移。接着,以结石目标为例分析了图像目标关键信息的导航采集方法,即通过语义分割网络获得图像目标的轮廓信息,进而与内窥镜主光轴进行重合度判定,记录符合重合条件关键帧的图像目标长度信息与对应的位姿信息。最后,基于针孔相机成像模型,将目标成像比例关系与内窥镜沿主光轴的位移结合,建立了目标尺寸测量方法。实验结果表明：所提方法的测量误差控制在10%以内;对长度1～9 mm目标的平均测量误差为0.33 mm。能够满足单目内窥镜检查中对目标尺寸测量的稳定可靠、精度高、省时省力等需求。     

Data Processing and Imaging in GPR System Dedicated for Landmine Detection

Improvements of GPR technology can be attained by making adjustments specific for the application of landmine detection on three levels: system design, data acquisition and data processing. In this paper we describe data processing algorithms specially developed for a novel video impulse ultra-wide band front end. With this front end, three-dimensional measurements (C-scans) have been carried out over a controlled test site, using a non-metallic scanner. The test site contained surface-laid and shallow buried landmines, both antitank and antipersonnel, made of plastic, wood, and metal. Because of practical limitations, the data have been acquired on an irregular grid. We have designed data preprocessing and imaging algorithms such that they take into account the specific antenna geometry and its elevation above the ground as well as the irregularity of the data acquisition grid. We show that by tuning the data pre-processing and imaging to the newly designed radar front end and to the particular data acquisition strategy, we obtain clear subsurface images. The resulting images show the ability of the GPR system to detect and visualize small surface laid and shallow buried targets.     

Characterization of Local 3-D Rough Dielectric Surfaces Using Standoff GPR Measurements with Widebeam Antennas

In real life most ground surfaces are not flat but rough. The observation of surface roughness depends on the wavelength and angle of the incident wave. In order to be able to detect shallow subsurface objects, on one hand we need to use higher frequencies to achieve better range resolution. One the other hand we have to deal with rough surfaces relative to shorter wavelengths. In this paper a wideband ground-penetrating radar (GPR) phase measurement and processing technique for characterizing three-dimensional (3-D) rough dielectric surfaces is presented. The method is based on the measurement of phase data by a standoff GPR with wide-beam antennas at short range over 3-D rough ground surfaces. The principle of this method was verified experimentally in the measurement of a composite surface. The height of the composite surface varies from 0 to 8 cm. The antennas are open-ended waveguide antennas whose frequency range is 2.3 GHz to 4.3 GHz. They are broadband, have low gain and wide beamwidth. The experimental tests demonstrate that the 3-D rough surfaces can be characterized locally by using a monochromatic and multifrequency broadband phase processing and imaging method. The results show good agreement between the imagery of the surface height distribution obtained by this method and the actual geometry of the 3-D rough surfaces.     

Estimation of Hydraulic Property of an Unconfined Aquifer by GPR

Controlled water productions were performed at a water source area of Ulaanbaatar city, Mongolia to evaluate the effectiveness of ground penetrating radar (GPR) for detecting and monitoring dynamic groundwater movements in the subsurface and for estimating the hydraulic properties of the aquifer. Field experiments in Ulaanbaatar were carried out in 2001 and 2002. GPR data were acquired using 100 MHz antennas. This paper reports the results of GPR methods to monitor the groundwater migration caused by the pumping operation and GPR’s potential ability to estimate hydraulic properties of the aquifer. The GPR data sets were acquired very carefully by locating the antenna position accurately. The residual trace shows a feature that is a combination of the water level reflections acquired at two different times in the pumping test. It helped to determine travel time and its effective reflection point from the top of the water saturated zone. The residual wavelet varies versus offset from the pumping well for a given residual image. Common midpoint (CMP) data and velocity analysis indicated the depth of the water table and the water content in the unsaturated and saturated zone. Combining hydrogeologic data with quantitative information yielded by GPR data, hydraulic properties of the aquifer could be estimated by assuming a hydraulic model. It was concluded that GPR can be successfully employed to monitor groundwater migration and to estimate hydraulic properties of the aquifer.     

Analysis of multiscale measurements of porous microstructures based on 3D optical microscopes

In this article, a multiscale measurement strategy is introduced to analyze porous microstructures and the main influences on the measurement accuracy of 3D optical microscopes are investigated. The purpose is to explore the fundamental relationship between resolution, magnification and imaging in optical systems derived from using different optical lenses and their impacts on the characterization of porous microstructures. A confocal laser scanning microscope with different lenses is used for the data acquisition. Afterwards, a post-processing for data combined with image processing is carried out to analyze the geometry differences of identical pores. The results show that the numerical aperture is the primary factor causing measurement differences of the same micro object rather than the magnification of a lens and the calibrated image pixel resolution. Moreover, the assessed geometry differences strongly depend on the size or the scale of the microstructures. This phenomenon can be treated as a good verification example for the classic Abbe-theory.     

Characterizing Subsurface Archaeological Structures with Full Resolution 3D GPR at the Early Dynastic Foundations of Saqqara Necropolis,Egypt

Currently, Ground Penetrating Radar (GPR) used in archaeological prospection is based on 2-D parallel line methodologies characterized by line spacing from 0.25 to 1 m (common line separation is 0.5 m) with different GPR antennas and extensive interpolation used to fill data gaps. High resolution 3-D GPR images of the subsurface can be obtained by recording data with a quarter wavelength grid spacing in all directions. Recently, we used a new GPR system which is a combination of commercial GPR with a rotary laser positioning system developed at Tohoku University for full-resolution subsurface imaging. In this paper we will show how the high density 3-D GPR data acquired over an area of about 14 m?×?28 m can improve the image quality and reveal the subsurface archaeological structure of early dynastic foundations in the Saqqara area. The GPR vertical cross-sections and the horizontal depth slices extracted from the full-resolution 3-D GPR reveal great information about ancient human activities, most likely burial mounds. GPR data at depth greater than 1.3 m were overwhelmed by “ringing features” (repeated horizontal harmonic-like features) most probably caused by the presence of underlying shallow layers of low resistivity shale and claystone. A 2-D electric resistivity tomography (ERT) profile was acquired using a multi-electrode system with 1 m electrode spacing. The ERT section shows high resistivity for the near surface desert sand and gravel deposits. The second geoelectric layer detected by ERT shows a low resistivity value consistent with the presence of a highly conductive layer at a depth of about 1.3 m. Integration of such different geophysical tools (e.g. GPR with ERT) helps to interpret the repeated horizontal features in the 3-D GPR data.     

A simple and stable autofocusing protocol for long multidimensional live cell microscopy

Focus maintenance is a challenging problem in multidimensional wide‐field microscopy. Most automated microscopes use software algorithms, which are applied to z‐sections of the object, to select for the plane with the best signal to noise ratio. When applied automatically in multidimensional imaging applications, autofocus routines significantly increase light exposure and can become cytotoxic if applied too frequently. In addition, automated focusing procedures can readily focus on unwanted high contrast objects. By labelling a defined position with a fluorescent marker, we were able to separate the focusing procedure from the actual image acquisition positions and therefore overcome some of the major drawbacks of routine autofocus procedures. To implement this method in a multidimensional acquisition experiment, we created a visual basic‐based program, which is run prior to each image acquisition. This technique allows tight control of focus whilst keeping light toxicity in live cell imaging experiments to a minimum.     

Polarimetric Borehole Radar System for Fracture Measurement

Radar polarimetry is a technology that overcomes the limitation between the radar resolution and the penetration depth of borehole radar. We have developed a stepped-frequency polarimetric borehole radar system. This is a polarimetric borehole radar system which measures the full-radar polarimetry in a borehole by changing the antenna arrangements. By using a network analyzer and an optical analog signal link, this system has a frequency bandwidth of 2–500MHz, which is suitable for two different antennas for radar polarimetry. We also propose a technique for polarimetric antenna calibration. In order to understand the potential of polarimetric borehole radar, field measurement were carried out at the Mirror Lake fractured-rock research site (NH, USA). The radar penetration depth from the borehole in the reflection measurement was over 10m, at the frequency range of 2–402MHz. We observed many clear reflections from fractures in each polarization status. Even in the raw data, we found the differences in the radar profile for different polarization status. We believe the polarimetric feature is closely related to the roughness of each fracture, and it is also related to the physical properties of the fracture such as water permeability.     

高质量X射线检测数字化成像及图像采集

X射线数字成像检测是工业无损检测的新技术.非胶片实时数字成像检测以其高效率、低成本,特别是数字图像的可交换性和存储方便等特点成为射线检测的发展趋势.通常使用的由射线像增强器、视频摄像机、图像采集卡和计算机组成的系统成像质量差且只能在低能X射线下使用,作者经过反复研究和实验,研制了基于单晶闪烁屏和科学级CCD相机的可用于高、低能X射线高分辨率、高对比灵敏度数字成像检测系统.介绍了整个检测系统的组成,设计了科学级制冷CCD相机、预处理电路、A/D转换电路、微机EPP方式快速数据采集电路和软件等.给出了某铝铸件(厚150mm)在加速器高能X射线下的成像结果.实验证明该系统比普通像增强器和视频相机组成的系统具有适应射线能量范围宽和成像质量好的优点(成像时间长一些).系统的空间分辨率大于3lp/mm, 检测物体50mm(等效钢厚)以上时,对比度灵敏度小于1%.     

基于激光三角法和嵌入式的微位移实时检测系统

基于激光三角法和嵌入式的微位移实时检测技术,采用激光三角法测距和灰度重心法提取激光条纹中心的原理,使线结构光、透镜、CCD位置信息满足Scheimpflug共轭清晰成像条件。通过图像处理获取像面坐标系位移信息,得到待测物体物面坐标系的位移,利用灰度重心法实时提取结构光条纹中心的特点,结合嵌入式系统尺寸小、稳定、便携的特性,实现对待测物体进行实时、非接触和微位移测量。     

动态电容层析成像图像重建算法

提出了融合ECT测量信息和被测对象动态演化信息的新型图像重建模型;基于Tikhonov正则化方法,建立一个同时考虑了ECT测量信息、被测对象动态演化信息、时间与空间约束的新型图像重建目标泛涵,将图像重建问题转化为最优化问题;提出了集成分裂Bregman迭代法优势的新型算法求解该目标泛涵。数值仿真结果表明,所提出的图像重建算法其图像重建质量均优于OIOR算法、STR算法及PLI算法;同时由于所提出的图像重建算法同时考虑了测量数据和重建模型的不精确性,其抵抗测量噪声的能力得以提高。     

正交调制复合光栅投影的三维测量机理及仿真

介绍了采用复合光栅来进行快速测量获取物体三维信息的方法。该方法通过投影由载频光栅携带与其正交的相移光栅而形成的复合光栅来进行三维面形测量。建立了基于复合光栅方法的三维测量理论模型,运用复合光栅完整模拟物体恢复的过程,模拟结果证明了该方法的可行性。该方法具有祛噪功能,由于只需投影一幅复合光栅图像和采集一幅变形光栅图像就实现了物体的重建,因此适用于物体三维面形的快速实时测量。     

快速旋转式偏振成像探测装置的设计

为了获取更快的偏振成像探测速率,本文对现有的机械旋转式偏振成像装置进行了改进,设计了连续旋转检偏器的成像方式,并改进图像处理过程,进一步提高了偏振成像速度。该装置克服了传统机械旋转式偏振成像装置体积大、成像速度慢、效率低的不足,利用电机带动检偏器快速匀速旋转,并与相机的曝光同步,能够快速便捷地实现偏振图像的获取。同时为了实现更高的偏振图像采集速率,对偏振图像采用了流水线式的处理方式,利用每相邻的3幅强度图解算得到偏振图像,使偏振图像与强度图像具有相同的成像速率。经过测试,该装置能够很好地完成偏振图像的采集,获取被测目标的偏振度和偏振角,在稳定工作状态下平均获取一幅偏振信息耗时0.033 s,且具有较好的工作稳定性。本文所做工作提升了机械旋转式偏振成像仪的工作效率,也为进一步提升机械式偏振成像仪的成像速度、实现对运动目标探测打下了基础。     

基于PXI总线和LVDS技术的高速图像采集系统的设计与实现

针对现代化生产和科学研究图像传感器采集数据量大、要求传输速度快等特点,提出了一种基于PXI总线和LVDS高速图像采集系统.介绍了系统的总体结构,重点阐述了各部分的功能实现,对图像数据传输的乒乓操作进行了系统仿真,并对系统整体功能进行了测试.仿真实验证明,系统具有稳定的图像数据采集和高速传输能力,能够满足工程需要.     

Dynamic characterization of a high speed gamma-ray tomograph

A HSGT (High Speed Gamma-ray Tomograph) has been designed and built at the University of Bergen with the objective to monitor rapid changes in multiphase hydrocarbon flow regimes. In order to perform real-time image reconstruction with photon integration times as low as 10 ms, a novel DACS (Data Acquisition and Control System) has been developed. The DACS is based on FPGA (Field-Programmable Gate Array) programming of the CompactRIO module from National Instruments to minimize its data acquisition and control time. The CompactRIO module includes a reconfigurable FPGA, which provides hardware-level data acquisition and control determinism with a time resolution of 25 ns. The data acquisition and control time for the HSGT obtained with the novel DACS interface design is 0.18 ms, which corresponds to a data transmission bandwidth of 1.35 Mbytes/s given that the HSGT data frame consists of 85 channels each comprising a 24 bit resolution. The DACS also facilitates FPGA sensor data pre-processing, i.e. normalization, of the acquired tomograph data to speed up the image reconstruction. Dynamic characterization of the HSGT for rotational and translational movements is presented in this paper, which is based on calculation of the RMSE (Root Mean Square Error) of the acquired tomogram compared to that of the test phantom. The test phantom consists of two spherical holes with different radius in a polypropylene sample. The results of the dynamic characterization show that the HSGT can sustain imaging of a rotational object with angular velocities ～30 rad/s. For translational movement (free fall) the HSGT is able to detect internal cross-sectional structures with velocities up to ～4 m s^?1.     

Photon event distribution sampling: an image formation technique for scanning microscopes that permits tracking of sub-diffraction particles with high spatial and temporal resolutions

In photon event distribution sampling, an image formation technique for scanning microscopes, the maximum likelihood position of origin of each detected photon is acquired as a data set rather than binning photons in pixels. Subsequently, an intensity-related probability density function describing the uncertainty associated with the photon position measurement is applied to each position and individual photon intensity distributions are summed to form an image. Compared to pixel-based images, photon event distribution sampling images exhibit increased signal-to-noise and comparable spatial resolution. Photon event distribution sampling is superior to pixel-based image formation in recognizing the presence of structured (non-random) photon distributions at low photon counts and permits use of non-raster scanning patterns. A photon event distribution sampling based method for localizing single particles derived from a multi-variate normal distribution is more precise than statistical (Gaussian) fitting to pixel-based images. Using the multi-variate normal distribution method, non-raster scanning and a typical confocal microscope, localizations with 8 nm precision were achieved at 10 ms sampling rates with acquisition of ~200 photons per frame. Single nanometre precision was obtained with a greater number of photons per frame. In summary, photon event distribution sampling provides an efficient way to form images when low numbers of photons are involved and permits particle tracking with confocal point-scanning microscopes with nanometre precision deep within specimens.     

Refractive and relativistic effects on ITER low field side reflectometer design

The ITER low field side reflectometer faces some unique design challenges, among which are included the effect of relativistic electron temperatures and refraction of probing waves. This paper utilizes GENRAY, a 3D ray tracing code, to investigate these effects. Using a simulated ITER operating scenario, characteristics of the reflected millimeter waves after return to the launch plane are quantified as a function of a range of design parameters, including antenna height, antenna diameter, and antenna radial position. Results for edge/SOL measurement with both O- and X-mode polarizations using proposed antennas are reported.     

Nondestructive Permittivity Profile Retrieval of Non-Planar Objects by Free Space Microwave Techniques

In this paper, a nondestructive method for the non-contact characterization of non-planar dielectric objects is presented. Traditionally, this kind of measurement is achieved by using spot-focused antennas associated to a vector network analyzer. In this study, we propose a free space measurement system, fitted out with a classical horn antenna, which demonstrates its usefulness in many practical cases. The permittivity has been computed from the measurement of the reflection coefficient of metal-backed samples of different shapes at 2.45 GHz. Signal processing tools, based on deconvolution techniques, are developed to enhance the accuracy of the permittivity measurement.     

Pushing the resolution limits in cryo electron tomography of biological structures

Cryo electron tomography is a three-dimensional imaging technique that is suitable for imaging snapshots of the structural arrangements of biomolecular complexes and macromolecules, both in vitro and in the context of the cell. In terms of attainable resolution, cryo electron tomographic reconstructions now show resolvable details in the 5-10 nm range, connecting optical microscopy with molecular imaging techniques. In view of the current developments in super-resolution light microscopy and correlative light and electron microscopy, cryo electron tomography will be increasingly important in structural biology as a tool to bridge light microscopy with molecular imaging techniques like NMR, X-ray diffraction and single particle electron microscopy. In cell biology, one goal, often referred to as visual proteomics, is the molecular mapping of whole cells. To achieve this goal and link cryo electron tomography to these high-resolution techniques, increasing the attainable resolution to 2-5 nm is vital. Here, we provide an overview of technical factors that limit the resolution in cryo electron tomography and discuss how during data acquisition and image processing these can be optimized to attain the highest possible resolution. Also, existing resolution measurement approaches and current technological developments that potentially increase the resolution in cryo electron tomography are discussed.