Two-dimensional gridless super-resolution method for ISAR imaging

In the past decade, the compressive sensing (CS) based ISAR imaging methods have gained much interest due to the capability of obtaining high-quality images with under-sampled data. However, both the performance and the application of CS ISAR imaging methods are limited by sparse representations and iterative reconstruction algorithms where the former may be hard to find and the latter may lead to low imaging efficiency. The deep-network (DN) based image reconstruction methods appeared in recent years have been shown to be able to dramatically reduce the computational complexity and ensure the image reconstruction at the same time. The Deep-ADMM-Net (DAN) is a network constructed by iterative steps of the traditional optimization algorithm, Alternating Direction Method of Multipliers (ADMM). The mainly characteristic of DAN is that it is interpretable and efficient. We propose a DAN based ISAR imaging method. The well trained DAN can reconstruct high-quality ISAR image using much fewer measurements than CS based imaging methods. Experimental results show that our proposed imaging method is superior to the existing CS method in both image reconstruction quality and computational efficiency

针对低信噪比下基于压缩感知(CS)的ISAR成像方法性能下降甚至失效的问题，该文提出了一种改进CS的成像方法，即用能量门限分离含目标的距离单元和噪声单元，同时在方位向上用相干投影来提高观测数据的信噪比，利用迭代加权的1-范数优化以增强真实散射点能量并且压制噪声。 改进CS算法适用于强噪声的环境下的ISAR成像，并可利用很少脉冲数获得高分辨的目标像。实测数据处理验证了该方法可以有效克服强噪声与杂波。同时，仅用16个回波脉冲就可得到高分辨ISAR图像，验证了该方法在非常有限脉冲条件下的稳健性。

A new approach has been proposed to reduce the computations involved in the ISAR imaging, which uses harmonic wavelet- (HW) based time-frequency representation (TFR). Since the HW-based TFR falls into a category of nonparametric time-frequency (T-F) analysis tool, it is computationally efficient compared to parametric T-F analysis tools such as adaptive joint time-frequency transform(AJTFT), adaptive wavelet transform(AWT), and evolutionary AWT (EAWT). Further, the performance of the proposed method of ISAR imaging is compared with the ISAR imaging by other nonparametric T-F analysis tools such as short-time Fourier transform(STFT) and Choi-Williams distribution (CWD). In the ISAR imaging, the use of HW-based TFR provides similar/better results with significant (92%) computational advantage compared to that obtained by CWD. The ISAR images thus obtained are identified using a neural network-based classification scheme with feature set invariant to translation, rotation, and scaling.

The compressive sensing(CS) and sparse representation(SR) technique provide new way to improve the ISAR image resolution. At present, in most traditional CS and SR based ISAR imaging methods, the target is regarded as a set of isolated scattering centers. As a matter of fact, adjacent scattering centers compose a lot of small sets which can reflect the structure of the target. It is no doubt that these small sets can be exploited to improve the ISAR image quality, but the traditional CS and SR ISAR imaging methods ignore this. So in this paper, a block sparse signal recovery algorithm-Block Sparse Bayesian Learning(BSBL) method is introduced to modify the traditional method, and the simulate experiment show that this method can get better ISAR image than the other methods.

B ENOIT is a fractal analysis software product for Windows 95, Windows 98, or Windows NT used to find order and patterns in seemingly chaotic data, particularly where traditional statistical approaches to data analysis fail. It is widely used in disciplines as diverse as biology, chemistry, physics, economics, medicine, and geology. The U.S. Geological Survey, for example, employs fractal analysis to accurately predict the volume of undiscovered deposits of oil and natural gas, on the basis of data from known deposits. BENOIT measures user-supplied data by standard fractal methods. For a fractal, measures change in value as the scale decreases in size because ever-smaller pieces become included in the analysis. Measures are plotted as a function of ruler size on a log-log plot, and a fractal dimension is calculated from the slope of the resulting line. Users select one of 10 analytical measures with the software. Five of the available measures in the program act upon bitmap images in Windows BMP format. These are described as the “self-similar” or two-dimensional (2D) methods, while the remaining group of routines act upon time-series or 1D data. The latter group requires data to be in a simple but specific data format, such as is available in Excel. The program also features a data generator that produces files with a given fractal dimension. Users may find this useful for testing and control purposes. The self-similar or image methods available in BENOIT measure different characteristics of bitmap objects in ways that should be scale-invariant. A real dataset normally has some fractal limit, and outside the limit, the fractal dimension will return a trivial value (1 for time-series or 2 for image data). Upper and lower fractal limits are controlled by the size of the dataset. Self-similar methods available in BENOIT are well known in fractal analysis: box dimension, perimeter-area dimension, information dimension, and ruler dimension. All methods are explained in standard Help files that contain several pages of information for each topic. The 1D analysis routines use “self-affine” methods of analysis. Self-affine fractals differ from self-similiar fractals in that their parts need to be rescaled by different factors in different coordinates to resemble the original. In the roughness-length method, the root-mean-square variation or roughness of the data is calculated for a variety of horizontal scales. The operation provides an estimate of the Hurst exponent, H , in a log-log plot, which is related to the fractal dimension. Standard self-affine methods available include R/S (Rescaled Range) analysis, power spectrum, roughness-length, variogram, and wavelets. Printing of log-log figures is provided, but without many features that would be found in a spreadsheet. Documentation for the program is available online. BENOIT has a highly visual interface, complete with an animated grid or ruler for self-similar fractal methods, and it gives users control of all calculations that the program performs, unlike other fractal software. Benoit is not without flaws. Some operations, such as name registration with the Windows NT 4.0 taskbar and the Open File requester, do not conform to standard Windows conventions. It would be of help to have an outline or flowchart of the operation of the different routines available in BENOIT for newcomers to fractal analysis. In summary, the variety of fractal analysis methods available in BENOIT, together with generally detailed help files and significant user control of operations, make BENOIT a good resource for learning about and using fractal analysis methods.

Three-dimensional (3D) imaging is increasingly shaping cancer research. This piece highlights key technical considerations when adopting 3D over traditional two-dimensional (2D) imaging methods, emphasizing its unique ability to capture complex and heterogeneous tumor architecture. We also discuss the importance of multiomics integration and how virtual staining workflows can address scalability issues. Finally, we propose hybrid strategies that balance 2D and 3D imaging for a well-rounded understanding of inter- and intratumoral heterogeneity, underscoring the complementary value of both modalities in advancing cancer research. This article is part of a special series: Driving Cancer Discoveries with Computational Research, Data Science, and Machine Learning/AI .

Factors affecting blood vessel lumen definition for two-dimensional and three-dimensional inflow magnetic resonance (MR) imaging methods are considered. Vessel definition is affected (a) by the amount of dephasing of the blood in the vessels, both for uncompensated and velocity-compensated gradients; (b) by the image reconstruction technique (normal Fourier reconstruction when asymmetric echoes are collected or a maximum-intensity projection technique in post-processing); (c) by loss of signal due to T2* dephasing; (d) by misregistration; (e) by vessel wall motion; and (f) by partial-volume effects. The first two factors were found to dominate for resolution on the order of 1 mm3. To overcome these dephasing problems, the authors developed asymmetric echo, velocity-compensated sequences with TEs as short as 4.8 msec. The data were then reconstructed with an iterative partial Fourier algorithm, enabling improved lumen definition to be obtained in phantoms and in vivo.

Two-dimensional imaging with a single-sided NMR probe

Convolutional neural network and level-set spectral element method for ultrasonic imaging of delamination cavities in an anisotropic composite structure

Highly reflective material target captured by the camera will form a mixture of the original target and its reflection image, and the above method is unable to accurately segment the image. To solve this problem, we proposed a method based on two-dimensional gray histogram contour projection segmentation algorithm. This algorithm is based on the grayscale image grayscale-field to establish the two-dimensional gray histogram, and use the two-dimensional histogram oblique segmentation method to filter out noise, then the Otsu method is used to projection segmentation and look for the best threshold segmentation in a straight line, so as to get the target image to be detected. Finally, use the convex polygon method to process the edge of the target. Experimental results show that the proposed method can accurately segment the target from the blurred image.

Molecular structural retrieval based on electron diffraction has been proposed to determine the atomic positions of molecules with sub-angstrom spatial and femtosecond temporal resolutions. Given its success on small molecular systems, in this work, we pointout that the accuracy of structure retrieval is constrained by the availability of a wide range of experimental data in the momentum space in all molecular systems. To mitigate the limitations, for laser-induced electron diffraction, here we retrieve molecular structures using two-dimensional (energy and angle) electron momentum spectra in the laboratory frame for a number of small molecular systems, which have previously been studied with 1D methods. Compared to the conventional single-energy or single-angle analysis, our 2D methods effectively expand the momentum range of the measured data. Besides utilization of the 2D data, two complementary methods are developed for consistency check on the retrieved results. The 2D nature of our methods also offers a way of estimating the error from retrieval, which has never been explored before. Comparing with results from prior experiments, our findings show evidence that our 2D methods outperform the conventional 1D methods. Paving the way to the retrieval of large molecular systems, in which their tunneling ionization rates are challenging to obtain, we estimate the error of using the isotropic model in place of including the orientation-dependent ionization rate.

The Scanning Laser Ophthalmoscope (SLO) is an essential medical tool for diagnosis of retinal disease. It uses a small amount of laser to scan the retinal at high speed and transmits the fundus images to the video monitor for medical auxiliary diagnosis. However, like all optical imaging technologies, due to the interference of hardware equipment and external conditions, it is often not ideal imaging. In most clinical cases of laser ophthalmoscope, only low-resolution retinal images can be used to assist medical diagnosis. For this reason, we propose a new depth super-resolution method of retinal image based on laser scanning ophthalmoscope. The retinal image enhanced by local Laplacian operator is introduced into an efficient full convolution neural network. The convolution network uses Adam algorithm to replace the traditional SGD(Stochastic gradient descent) method, which runs faster and faster, and the reconstructed image effect is better. In this work, we subjectively evaluate our algorithm, apply it to real retinal images and compare it with several traditional super-resolution reconstruction methods. The experimental results show that this method has achieved good results in improving the overall quality of laser scanning ophthalmoscope image.

Multidimensional Signal, Image, and Video Processing and Coding

In order to solve the problem that conventional inverse synthetic aperture radar three-dimensional (InISAR 3D) imaging methods need too much computation and so many restrictions in image registration step, this study proposes an InISAR 3D imaging algorithm based on joint translational compensation. The algorithm employs joint translational compensation of range and cross-range direction to achieve the aim of imaging and registration. While obtaining the two-dimensional ISAR images, the accurate registration of different ISAR images is completed. The algorithm needs less calculations and fewer constraints. Simulation results verify the effectiveness of it.

A new resolution enhancement method for sandstone thin-section images using perceptual GAN