一种基于MOD字典学习的图像超分辨率重建新算法

将低分辨率图像重建成高分辨率图像是图像处理领域中的一个重要课题。Yang提出一种基于联合字典学习的图像超分辨率重建算法,其算法样本选取与字典训练方法较为复杂。提出一种基于MOD字典学习的图像超分辨率重建新算法,首先采用少量的训练样本代替Yang的大量训练样本,然后使用MOD字典学习算法代替Yang的FFS字典学习算法,最后利用字典对图像进行稀疏表示与重建。实验结果表明,所提出的算法速度较快,并且重建图像的质量较高。     

High-photon-throughput snapshot colour imaging using a monochromatic digital camera and a pupil-domain diffuser

Colour RGB imaging with high throughput was achieved by a monochromatic digital camera with a dispersive diffuser at its pupil. An acquired snapshot monochromatic image was converted to colour coordinates through spectra, by resorting to digital processing with a compressed sensing-based algorithm of spectral imaging. Results of optical evaluation and calibration of an optical system and colour imaging experiments are reported.     

Visibility of ghost imaging in a two-arm microscope imaging system

Ghost imaging with a classical thermal source is investigated in a two-arm microscope imaging system. The dependence of the imaging visibility on the aperture of the reference lens is discussed. It is shown that by using large apertures, good visibility as well as enhancing resolution can be obtained. The effects from the distance the object is moved away from the original plane are also studied, and one can obtain good visibility with a well-resolved image by changing the distance.     

基于聚类和协同表示的超分辨率重建

图像超分辨率重建是利用单幅或多幅降质的低分辨率图像重建得到高分辨率图像,以提高图像的视觉效果并获得更多可用的信息。本文提出结合图像特征聚类和协同表示的超分辨率重建方法。在训练阶段根据图像的特征信息对图像样本进行聚类并利用图像特征的差异性训练不同的字典,克服了传统训练单个字典方法对图像特征表示不足的缺点。而且利用协同表示方法求得不同聚类的高、低分辨率图像样本之间的映射矩阵,提高了图像重建速度。实验表明,本文方法与其他方法相比,不仅提高了重建图像的PSNR和SSIM指标,而且改善了视觉效果。     

Widefield scanning imaging with optical super-resolution

An economical, pollution-free microsphere-based widefield scanning imaging method is presented. This system is able to visualize the surface pattern of the sample through a transparent dielectric microsphere stuck onto a glass probe. The microsphere endows the system with super-resolution capability, while the field of view can easily be expanded by scanning and image stitching. The feasibilities and advantages of this method have been verified experimentally.     

水下热扰动的光学成像失真问题研究

为了研究水下热扰动环境对光学成像的畸变、模糊等失真问题的影响,利用水下图像的灰度分布、结构相似性图像度量(SSIM)和归一化最大灰度梯度清晰度评价函数来评价目标图像在径向和轴向上的畸变和模糊等失真程度,得到水下热扰动对光学成像变化的规律。实验数据表明,随着成像系统与目标的轴向距离增加,图像的畸变和模糊程度越来越大。轴向距离L1=500 mm时,对应图像的SSIM值优于0.7,归一化清晰度值优于0.8;轴向距离L3=1500 mm时,对应图像的SSIM值低于0.2,归一化清晰度值不足0.6;此外,轴向距离L1时,成像在径向上,距离发热源越近,边缘漂移越大,即成像图像畸变越严重;最后,相同轴向和径向条件下,目标在不同时刻的图像SSIM和归一化清晰度值有优劣,该结论可为后续的水下图像复原提供参考。     

彩色图像多尺度引导的深度图像超分辨率重建

为获得更优的深度图像超分辨率重建结果,本文构建了彩色图像多尺度引导深度图像超分辨率重建卷积神经网络。该网络使用多尺度融合方法实现高分辨率(HR)彩色图像特征对低分辨率(LR)深度图像特征的引导,有益于恢复图像细节信息。在对LR深度图像提取特征的过程中,构建了多感受野残差块(MRFRB)提取并融合不同感受野下的特征,然后将每一个MRFRB输出的特征连接、融合,得到全局融合特征。最后,通过亚像素卷积层和全局融合特征,得到HR深度图像。实验结果表明,该算法得到的超分辨率图像缓解了边缘失真和伪影问题,有较好的视觉效果。     

光纤集成化高分辨率无标记差分显微成像系统

得益于荧光标记技术,远场超分辨显微成像技术的不断发展为人们打开了一道微观世界的大门。远场超分辨光学显微镜也成为现代医学、生命科学等领域研究发展的重要工具和设备。但是相比之下,远场无标记显微成像技术发展却相对缓慢。为此,本文提出了一种采用光纤器件的集成化差分显微成像方法。该方法通过特制的光纤模式选择耦合器(MSC)实现了差分成像系统中空心光斑的生成,并解决了空心光斑和实心光斑严格对准困难的问题。搭建了一套结构简单紧凑,高度集成化的无标记显微成像系统。实验中,采用直径为150 nm的金颗粒和最小间距约50 nm的无标记聚合物线对结构对系统进行成像测试,分辨率相较传统扫描共聚焦显微镜大幅提升。

     

Ptychographical imaging with partially saturated diffraction patterns

In ptychographical imaging with X-ray or electron beam, the recorded speckle intensity pattern of far field diffraction was often saturated by the presence of strong zeroth-order which was undesirable. A beam-stop was widely used to block the area of strong central diffraction in the pattern. The image was then reconstructed by disregarding the diffraction constraints within that area where inaccuracies exists due to saturation or blocking of the beam. Though reconstructions were obtained in this way by omitting saturation regions, the rationale behind this method was not discussed in detail and its effect on the final image quality was not studied. The physics of this method was analyzed theoretically and experimentally in this study. The convolution effect between illumination and object functions was able to retrieve and bring back the information lost in the blocked or saturated area from its neighborhood. This was possible while the saturated or blocked area was not wider than the spectral width of illumination, and once it exceeds this minimum, the quality of reconstruction would degrade seriously. The theoretical model was verified by numerical simulations and experiments.     

Multi-plane imaging for arbitrary reconstruction positions in digital holography

A novel method for multi-plane imaging in digital holography is proposed: holograms of objects located at various places are recorded and then reconstructed simultaneously through one-step Fresnel diffraction using the quadratic distorted phase factor (QDPF). The theory of one-step Fresnel diffraction with the QDPF is deduced and experimental results support the theoretical investigation.     

Holographic digital Fourier microscopy for selective imaging of biological tissue

We present an application of digital Fourier holography for selective imaging of scatterers with different sizes in turbid media such as biological tissues. A combination of Fourier holography and high‐resolution digital recording, digital Fourier microscopy (DFM) permits crucial flexibility in applying filtering to highlight scatterers of interest in the tissue. The high‐resolution digital hologram is a result of the collation of Fourier holographic frames to form a large‐size composite hologram. It is expected that DFM has an improved signal‐to‐noise ratio as compared to conventional direct digital imaging, e.g., phase microscopy, as applied to imaging of small‐size objects. The demonstration of the Fourier filtering capacity of DFM using a biological phantom represents the main focus of this article. © 2005 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 14, 253–258, 2004; Published online inWiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.20031     

变分贝叶斯框架中的多帧图像盲超分辨

多帧图像超分辨理论是通过融合多帧低分辨率图像的信息来重构高空间分辨率图像。准确估计低分辨率图像的模糊核是进行有效信息融合的先决条件。传统超分辨方法通常假设模糊核已知且采用固定的高斯滤波模糊核,并且模型参数需要费时的手动调整。本文在变分贝叶斯框架下获得相应的超分辨算法,该算法对高分辨率图像、模糊核和模型参数同时进行最优估计。对比实验表明,模糊核自适应估计的盲超分辨方法总体性能优于现有的变分贝叶斯框架下的图像超分辨方法,特别是在高信噪比场景,推荐方法优势更加明显。     

Temperature recovery and contrast computations in NDE thermographic imaging systems

Thermographic imaging systems find application in many nondestructive evaluation problems. However, because of the inherent nature of the image-formation process, several degradations arise which preclude a more generalized use of thermographic NDE. In this paper, we analyze radiometric and noise degradations and propose a calibration process for the imaging systems which allows recovery of object surface temperature evolution. Moreover, a constrast computation technique which combines both the time and spatial reference techniques is shown to increase the defect visibility and make possible a more quantitative (sizing) defect evaluation.     

调和映射约束下的超分辨率图像重建

针对超分辨率图像重建过程中的正则化约束问题,本文提出采用p(x)调和映射进行正则化重建,根据超分辨率图像观察模型及正则约束,给出相应的能量泛函,并采用动态偏微分方程演化来求解能量泛函.该算法在重建的过程中能够根据图像空间特性自适应地采用不同的p(x)范数进行正则化,在图像的平滑区域采用近似2次范数进行正则化,而在图像的边缘区域采用近似1次范数进行正则化.实验结果均表明该算法不仅能有效地重建图像边缘,而且能有效地改善一次范教约束重建的分片常数效应.     

Research on the Application of Super Resolution Reconstruction Algorithm for Underwater Image

Underwater imaging is widely used in ocean, river and lake exploration, but it is affected by properties of water and the optics. In order to solve the lower-resolution underwater image formed by the influence of water and light, the image super-resolution reconstruction technique is applied to the underwater image processing. This paper addresses the problem of generating super-resolution underwater images by convolutional neural network framework technology. We research the degradation model of underwater images, and analyze the lower-resolution factors of underwater images in different situations, and compare different traditional super-resolution image reconstruction algorithms. We further show that the algorithm of super-resolution using deep convolution networks (SRCNN) which applied to super-resolution underwater images achieves good results.     

大景深光学成像系统关键技术的研究

在普通光学成像系统中引入相位模板和数字图像复原技术,可以增大系统的景深。本文设计了一种新的由多个等面积环形区域组成的纯相位模板,相邻的环形分区交替补偿高斯物面两侧离焦。同时,基于频域滤波理论和MTF离焦不变性质,给出了相应的不依赖于离焦程度的图像复原滤波器的表达式。系统的MTF曲线、MTF密度图和Spoke图结果表明系统具有大景深特性。通过适当调整模板参量,系统景深可以达到普通光学系统的4倍以上。系统可以应用到生物和医学的显微成像、机器视觉等多个领域。     

Exploring the formation of focal adhesions on patterned surfaces using super-resolution imaging

The formation of focal adhesions on various sizes of fibronectin patterns, ranging from 200 μm to 250 nm, was systematically investigated by total internal reflection fluorescence microscopy and super-resolution imaging. It was found that cells adhered to and spread on these micro/nanopatterns, forming focal adhesions. On a micrometer scale the shape of the focal adhesions was elongated. However, on the nanometer scale, the shape of focal adhesions became dotlike. To further explore the distribution of focal adhesion proteins formed on surfaces, a localization-based super-resolution imaging technique was employed in order to determine the position and density of vinculin proteins. A characteristic distance of 50 nm was found between vinculin molecules in the focal adhesions, which did not depend on the size of the fibronectin nanopatterns. This distance was found to be crucial for the formation of focal adhesions. In addition, the density of vinculin at the focal adhesions formed on the nanopatterns increased as the pattern size decreased. The density of the protein was found to be 425 ± 247, 584 ± 302, and 703 ± 305 proteins μm(-2) on the 600, 400, and 250 nm fibronectin patterns respectively. Whereas 226 ± 77 proteins μm(-2) was measured for the matured focal adhesions on homogeneous fibronectin coated substrates. The increase in vinculin density implies that an increase in mechanical load was applied to the focal adhesions formed on the smaller nanopatterns.     

对光电成像系统干扰效果的评估方法

提出了对两类光电成像系统干扰效果的评估准则。对观瞄用光电成像系统,可以根据对成像功能的损伤破坏程度将激光致盲干扰效果划分为四个等级,根据对目标识别的影响程度将烟幕干扰效果划分为三个等级。为定量评估烟幕对观瞄用光电成像系统的干扰效果,可以采用成像系统在实施干扰前后图像之间的相关函数作为评估指标,按照相关函数值的大小将干扰效果划分为三个等级。对跟踪用光电成像系统,可依据有干扰时成像系统能否输出有效跟踪误差以及跟踪误差是否超出正常跟踪精度的三倍判定干扰是否有效,并依据干扰成功率的大小划分干扰等级。     

Preliminary study of coded-source-based neutron imaging at the CPHS

A cold neutron radiography/tomography instrument is under construction at the Compact Pulsed Hadron Source (CPHS) at Tsinghua University, China. The neutron flux is so low that an acceptable neutron radiographic image requires a long exposure time in the single-hole imaging mode. The coded-source-based imaging technique is helpful to increase the utilization of neutron flux to reduce the exposure time without loss in spatial resolution and provides high signal-to-noise ratio (SNR) images. Here we report a preliminary study on the feasibility of coded-source-based technique applied to the cold neutron imaging with a low-brilliance neutron source at the CPHS. A proper coded aperture is designed to be used in the beamline instead of the single-hole aperture. Two image retrieval algorithms, the Wiener filter algorithm and the Richardson-Lucy algorithm, are evaluated by using analytical and Monte Carlo simulations. The simulation results reveal that the coded source imaging technique is suitable for the CPHS to partially solve the problem of low neutron flux.     

Broadband holography applied to eddy current imaging by using signals with multiplied phases

In this paper an eddy current imaging method for nondestructive testing purposes is presented which utilizes the concept of broadband holography. An eddy current coil which is used simultaneously as an antenna for eddy current generation and as a probe for detection of response of interaction between eddy currents and flaws, respectively, is moved along a synthetic aperture during the imaging procedure generating synthetic eddy current pulses by scanning a certain frequency range. In terms of wave propagation phenomena the penetration depth (range) of eddy currents in conducting media is small compared to the equivalent wavelength of this type of fields. Therefore, adequate resolution can only be obtained in the reconstructed cross-sectional images by phase multiplication of received multifrequency signals, which is equivalent to a fictitious reduction of wavelengths. Experimental results verify the imaging capability of this method with improved resolution compared to conventional eddy current testing methods.