U-net is an image segmentation technique developed primarily for image segmentation tasks. These traits provide U-net with a high utility within the medical imaging community and have resulted in extensive adoption of U-net as the primary tool for segmentation tasks in medical imaging. The success of U-net is evident in its widespread use in nearly all major image modalities, from CT scans and MRI to X-rays and microscopy. Furthermore, while U-net is largely a segmentation tool, there have been instances of the use of U-net in other applications. Given that U-net’s potential is still increasing, this narrative literature review examines the numerous developments and breakthroughs in the U-net architecture and provides observations on recent trends. We also discuss the many innovations that have advanced in deep learning and discuss how these tools facilitate U-net. In addition, we review the different image modalities and application areas that have been enhanced by U-net.

U-Net and Its Variants for Medical Image Segmentation: A Review of Theory and Applications

We propose an end-to-end place recognition model based on a novel deep neural network. First, we propose to exploit the spatial pyramid structure of the images to enhance the vector of locally aggregated descriptors (VLAD) such that the enhanced VLAD features can reflect the structural information of the images. To encode this feature extraction into the deep learning method, we build a spatial pyramid-enhanced VLAD (SPE-VLAD) layer. Next, we impose weight constraints on the terms of the traditional triplet loss (T-loss) function such that the weighted T-loss (WT-loss) function avoids the suboptimal convergence of the learning process. The loss function can work well under weakly supervised scenarios in that it determines the semantically positive and negative samples of each query through not only the GPS tags but also the Euclidean distance between the image representations. The SPE-VLAD layer and the WT-loss layer are integrated with the VGG-16 network or ResNet-18 network to form a novel end-to-end deep neural network that can be easily trained via the standard backpropagation method. We conduct experiments on three benchmark data sets, and the results demonstrate that the proposed model defeats the state-of-the-art deep learning approaches applied to place recognition.

Spatial Pyramid-Enhanced NetVLAD With Weighted Triplet Loss for Place Recognition

Segmentation and Detection.- Graph Cut Optimization for the Piecewise Constant Level Set Method Applied to Multiphase Image Segmentation.- Tubular Anisotropy Segmentation.- An Unconstrained Multiphase Thresholding Approach for Image Segmentation.- Extraction of the Intercellular Skeleton from 2D Images of Embryogenesis Using Eikonal Equation and Advective Subjective Surface Method.- On Level-Set Type Methods for Recovering Piecewise Constant Solutions of Ill-Posed Problems.- The Nonlinear Tensor Diffusion in Segmentation of Meaningful Biological Structures from Image Sequences of Zebrafish Embryogenesis.- Composed Segmentation of Tubular Structures by an Anisotropic PDE Model.- Extrapolation of Vector Fields Using the Infinity Laplacian and with Applications to Image Segmentation.- A Schrodinger Equation for the Fast Computation of Approximate Euclidean Distance Functions.- Semi-supervised Segmentation Based on Non-local Continuous Min-Cut.- Momentum Based Optimization Methods for Level Set Segmentation.- Optimization of Divergences within the Exponential Family for Image Segmentation.- Convex Multi-class Image Labeling by Simplex-Constrained Total Variation.- Geodesically Linked Active Contours: Evolution Strategy Based on Minimal Paths.- Validation of Watershed Regions by Scale-Space Statistics.- Adaptation of Eikonal Equation over Weighted Graph.- A Variational Model for Interactive Shape Prior Segmentation and Real-Time Tracking.- Image Enhancement and Reconstruction.- A Nonlinear Probabilistic Curvature Motion Filter for Positron Emission Tomography Images.- Finsler Geometry on Higher Order Tensor Fields and Applications to High Angular Resolution Diffusion Imaging.- Bregman-EM-TV Methods with Application to Optical Nanoscopy.- PDE-Driven Adaptive Morphology for Matrix Fields.- On Semi-implicit Splitting Schemes for the Beltrami Color Flow.- Multi-scale Total Variation with Automated Regularization Parameter Selection for Color Image Restoration.- Multiplicative Noise Cleaning via a Variational Method Involving Curvelet Coefficients.- Projected Gradient Based Color Image Decomposition.- A Dual Formulation of the TV-Stokes Algorithm for Image Denoising.- Anisotropic Regularization for Inverse Problems with Application to the Wiener Filter with Gaussian and Impulse Noise.- Locally Adaptive Total Variation Regularization.- Basic Image Features (BIFs) Arising from Approximate Symmetry Type.- An Anisotropic Fourth-Order Partial Differential Equation for Noise Removal.- Enhancement of Blurred and Noisy Images Based on an Original Variant of the Total Variation.- Coarse-to-Fine Image Reconstruction Based on Weighted Differential Features and Background Gauge Fields.- Edge-Enhanced Image Reconstruction Using (TV) Total Variation and Bregman Refinement.- Nonlocal Variational Image Deblurring Models in the Presence of Gaussian or Impulse Noise.- A Geometric PDE for Interpolation of M-Channel Data.- An Edge-Preserving Multilevel Method for Deblurring, Denoising, and Segmentation.- Fast Dejittering for Digital Video Frames.- Sparsity Regularization for Radon Measures.- Split Bregman Algorithm, Douglas-Rachford Splitting and Frame Shrinkage.- Anisotropic Smoothing Using Double Orientations.- Image Denoising Using TV-Stokes Equation with an Orientation-Matching Minimization.- Augmented Lagrangian Method, Dual Methods and Split Bregman Iteration for ROF Model.- The Convergence of a Central-Difference Discretization of Rudin-Osher-Fatemi Model for Image Denoising.- Theoretical Foundations for Discrete Forward-and-Backward Diffusion Filtering.- L 0-Norm and Total Variation for Wavelet Inpainting.- Total-Variation Based Piecewise Affine Regularization.- Image Denoising by Harmonic Mean Curvature Flow.- Motion Analysis, Optical Flow, Registration and Tracking.- Tracking Closed Curves with Non-linear Stochastic Filters.- A Multi-scale Feature Based Optic Flow Method for 3D Cardiac Motion Estimation.- A Combined Segmentation and Registration Framework with a Nonlinear Elasticity Smoother.- A Scale-Space Approach to Landmark Constrained Image Registration.- A Variational Approach for Volume-to-Slice Registration.- Hyperbolic Numerics for Variational Approaches to Correspondence Problems.- Surfaces and Shapes.- From a Single Point to a Surface Patch by Growing Minimal Paths.- Optimization of Convex Shapes: An Approach to Crystal Shape Identification.- An Implicit Method for Interpolating Two Digital Closed Curves on Parallel Planes.- Pose Invariant Shape Prior Segmentation Using Continuous Cuts and Gradient Descent on Lie Groups.- A Non-local Approach to Shape from Ambient Shading.- An Elasticity Approach to Principal Modes of Shape Variation.- Pre-image as Karcher Mean Using Diffusion Maps: Application to Shape and Image Denoising.- Fast Shape from Shading for Phong-Type Surfaces.- Generic Scene Recovery Using Multiple Images.- Scale Space and Feature Extraction.- Highly Accurate PDE-Based Morphology for General Structuring Elements.- Computational Geometry-Based Scale-Space and Modal Image Decomposition.- Highlight on a Feature Extracted at Fine Scales: The Pointwise Lipschitz Regularity.- Line Enhancement and Completion via Linear Left Invariant Scale Spaces on SE(2).- Spatio-Featural Scale-Space.- Scale Spaces on the 3D Euclidean Motion Group for Enhancement of HARDI Data.- On the Rate of Structural Change in Scale Spaces.- Transitions of a Multi-scale Image Hierarchy Tree.- Local Scale Measure for Remote Sensing Images.

Scale space and variational methods in computer vision : Second International Conference, SSVM 2009, Voss, Norway, June 1-5, 2009 : proceedings

Two decades after the introduction of matrix assisted laser desorption/ionization (MALDI) and electrospray ionization (ESI), soft ionization mass spectrometry represents a powerful toolset for the structural investigation of synthetic polymers. The present review highlights the current state-of-the-art, covering the latest developments of novel techniques, enabling instrumentation as well as the important applications of soft ionization MS from the beginning of 2008. Special attention is paid to the role that soft ionization MS has played in the mechanistic investigation of radical polymerization processes since 2005. © 2010 The Royal Society of Chemistry.

Mass spectrometry in polymer chemistry: A state-of-the-art up-date

This study had two main aims: (1) to provide a comprehensive review of terrestrial laser scanner (TLS) point cloud registration methods and a better understanding of their strengths and weaknesses; and (2) to provide a large-scale benchmark data set (Wuhan University TLS: Whu-TLS) to support the development of cutting-edge TLS point cloud registration methods, especially deep learning-based methods. In particular, we first conducted a thorough review of TLS point cloud registration methods in terms of pairwise coarse registration, pairwise fine registration, and multiview registration, as well as analyzing their strengths, weaknesses, and future research trends. We then reviewed the existing benchmark data sets (e.g., ETH Dataset and Robotic 3D Scanning Repository) for TLS point cloud registration and summarized their limitations. Finally, a new benchmark data set was assembled from 11 different environments (i.e., subway station, high-speed railway platform, mountain, forest, park, campus, residence, riverbank, heritage building, underground excavation, and tunnel environments) with variations in the point density, clutter, and occlusion. In addition, we summarized future research trends in this area, including auxiliary data-guided registration, deep learning-based registration, and multi-temporal point cloud registration.

Registration of large-scale terrestrial laser scanner point clouds: A review and benchmark

In this paper, we address the semisupervised distance metric learning problem and its applications in classification and image retrieval First, we formulate a semisupervised distance metric learning model by considering the metric information of inner classes and interclasses In this model, an adaptive parameter is designed to balance the inner metrics and intermetrics by using data structure Second, we convert the model to a minimization problem whose variable is symmetric positive-definite matrix Third, in implementation, we deduce an intrinsic steepest descent method, which assures that the metric matrix is strictly symmetric positive-definite at each iteration, with the manifold structure of the symmetric positive-definite matrix manifold Finally, we test the proposed algorithm on conventional data sets, and compare it with other four representative methods The numerical results validate that the proposed method significantly improves the classification with the same computational efficiency

Manifold Preserving: An Intrinsic Approach for Semisupervised Distance Metric Learning

LieTrICP: An improvement of trimmed iterative closest point algorithm

Many deep learning models are vulnerable to the adversarial attack, i.e., imperceptible but intentionally-designed perturbations to the input can cause incorrect output of the networks. In this paper, using information geometry, we provide a reasonable explanation for the vulnerability of deep learning models. By considering the data space as a non-linear space with the Fisher information metric induced from a neural network, we first propose an adversarial attack algorithm termed one-step spectral attack (OSSA). The method is described by a constrained quadratic form of the Fisher information matrix, where the optimal adversarial perturbation is given by the first eigenvector, and the vulnerability is reflected by the eigenvalues. The larger an eigenvalue is, the more vulnerable the model is to be attacked by the corresponding eigenvector. Taking advantage of the property, we also propose an adversarial detection method with the eigenvalues serving as characteristics. Both our attack and detection algorithms are numerically optimized to work efficiently on large datasets. Our evaluations show superior performance compared with other methods, implying that the Fisher information is a promising approach to investigate the adversarial attacks and defenses.

/pdf/the-adversarial-attack-and-detection-under-the-fisher-42wjp7g2lk.pdf

The Adversarial Attack and Detection under the Fisher Information Metric

A new structure-preserving quaternion QR decomposition method for color image blind watermarking

Virus image classification using multi-scale completed local binary pattern features extracted from filtered images by multi-scale principal component analysis

Yaxin Peng

Papers

Manifold Preserving: An Intrinsic Approach for Semisupervised Distance Metric Learning

LieTrICP: An improvement of trimmed iterative closest point algorithm

The Adversarial Attack and Detection under the Fisher Information Metric

A new structure-preserving quaternion QR decomposition method for color image blind watermarking

Virus image classification using multi-scale completed local binary pattern features extracted from filtered images by multi-scale principal component analysis