Local Description Research Articles

Recent advances in local feature detector and descriptor: a literature survey

The computer vision system is the technology that deals with identifying and detecting the objects of a particular class in digital images and videos. Local feature detection and description play an essential role in many computer vision applications like object detection, object classification, etc. The accuracy of these applications depends on the performance of local feature detectors and descriptors used in the methods. Over the past decades, new algorithms and techniques have been introduced with the development of machine learning and deep learning techniques. The machine learning techniques can lead the work to the next level when sufficient data is provided. Deep learning algorithms can handle a large amount of data efficiently. However, this may raise questions in a researcher’s mind about selecting the best algorithm and best method for a particular application to increase the performance. The selection of the algorithms highly depends on the type of application and amount of data to be handled. This encouraged us to write a comprehensive survey of local image feature detectors and descriptors from state-of-the-art to the recent ones. This paper presents feature detection and description methods in the visible band with their advantages and disadvantages. We also gave an overview of current performance evaluations and benchmark datasets. Besides, the methods and algorithms are described to find the features beyond the visible band. Finally, we concluded the survey with future directions. This survey may help researchers and serve as a reference in the field of the computer vision system.

Enhanced Sparsity Prior Model for Low-Rank Tensor Completion.

Conventional tensor completion (TC) methods generally assume that the sparsity of tensor-valued data lies in the global subspace. The so-called global sparsity prior is measured by the tensor nuclear norm. Such assumption is not reliable in recovering low-rank (LR) tensor data, especially when considerable elements of data are missing. To mitigate this weakness, this article presents an enhanced sparsity prior model for LRTC using both local and global sparsity information in a latent LR tensor. In specific, we adopt a doubly weighted strategy for nuclear norm along each mode to characterize global sparsity prior of tensor. Different from traditional tensor-based local sparsity description, the proposed factor gradient sparsity prior in the Tucker decomposition model describes the underlying subspace local smoothness in real-world tensor objects, which simultaneously characterizes local piecewise structure over all dimensions. Moreover, there is no need to minimize the rank of a tensor for the proposed local sparsity prior. Extensive experiments on synthetic data, real-world hyperspectral images, and face modeling data demonstrate that the proposed model outperforms state-of-the-art techniques in terms of prediction capability and efficiency.

Parametrization of a Conjugacy Class of the Special Linear Group

A local description of the foliation of the group SL(n) into conjugacy classes, and also the foliation of 𝔰𝔩∗(n) into coadjoint orbits, requires introducing parameters on a conjugacy class (a coadjoint orbit). Under the assumption that the parameters are rational functions of natural coordinates (the matrix elements) on SL(n), the problem is reduced to solving a system of linear equations. This system arises from the requirement that the parameters be invariant with respect to translations along vector fields normal to the conjugacy class. Similarly, the problem of parametrization of coadjoint orbits in 𝔰𝔩∗(n) can be solved using the Cartan–Weyl basis for 𝔰𝔩(n). The adjoint action is the differential of the conjugation action. Consequently, the parameters on the conjugacy classes and the coadjoint orbits are related by a transformation determined by the mapping from the algebra (n) to the group SL(n). The groups SL(3), SL(4) are considered as examples. Bibliography: 13 titles.

Combining multiple spatial statistics enhances the description of immune cell localisation within tumours

Digital pathology enables computational analysis algorithms to be applied at scale to histological images. An example is the identification of immune cells within solid tumours. Image analysis algorithms can extract precise cell locations from immunohistochemistry slides, but the resulting spatial coordinates, or point patterns, can be difficult to interpret. Since localisation of immune cells within tumours may reflect their functional status and correlates with patient prognosis, novel descriptors of their spatial distributions are of biological and clinical interest. A range of spatial statistics have been used to analyse such point patterns but, individually, these approaches only partially describe complex immune cell distributions. In this study, we apply three spatial statistics to locations of CD68+ macrophages within human head and neck tumours, and show that images grouped semi-quantitatively by a pathologist share similar statistics. We generate a synthetic dataset which emulates human samples and use it to demonstrate that combining multiple spatial statistics with a maximum likelihood approach better predicts human classifications than any single statistic. We can also estimate the error associated with our classifications. Importantly, this methodology is adaptable and can be extended to other histological investigations or applied to point patterns outside of histology.

No-reference stereoscopic image quality assessment based on global and local content characteristics

No-reference stereoscopic images quality assessment (NR-SIQA) via deep learning has gained increasing attention. In this paper, we propose a no-reference stereoscopic image quality assessment method based on global and local content characteristics. The proposed method simulates the perception route of human visual system, and derives features from the fused view and single view through the global feature fusion sub-network and local feature enhancement sub-network. As for the fused view, a cross-fusion strategy is applied to model the process in the V1 visual cortex, and the multi-scales pooling (MSP) is utilized to integrate context information under different sub-regions for effective global feature extraction. As for the single view, the asymmetric convolution block (ACB) is introduced to strengthen the local information description. By jointly considering the fused view and single view, the proposed network can efficiently extract the features for quality assessment. Finally, a weighted average strategy is applied to estimate the visual quality of stereoscopic image. Experimental results on 3D quality databases demonstrate that the proposed network is superior to the state-of-the-art metrics, and achieves an excellent performance.

Multi-input 2-dimensional deep belief network: diabetic retinopathy grading as case study

The most important action in treating diabetic retinopathy is early diagnosis and its progression degree. This paper presents a two-dimensional Deep Belief Network based on Mixed-restricted Boltzmann Machine capable of receiving multiple two-dimensional inputs. Using multiple inputs provides more appropriate prior information for learning. In this proposed method, the image is transferred to the HSV color space and then the 3D color image is converted to a 2D matrix using a weighted mean. This weighted mean is calculated based on the entropy criterion. The resulting two-dimensional matrix is not in pixel and is merely a raw description of the image. The local, regional and global descriptions are extracted from this matrix and provided for the network. The proposed deep network automatically extracts the appropriate features to determine the progression degree of diabetic retinopathy by the network. Window by window image processing can overcome one of the basic problems of image classification, i.e. the small number of labeled data. Experiments showed that the proposed method is superior when compared to other methods.

Outcrop and thin section scale descriptions of Silurian carbonates within the Hiawatha National Forest, Mackinac County, Upper Peninsula, Michigan, USA

Silurian carbonate outcrops within the southern part of the St. Ignace District of the Hiawatha National Forest (Mackinac County, Michigan) were examined to develop a local description of the exposed geology. The bedrock geology consists of the Manistique Group (dominantly the Cordell Formation) and the Engadine Group (Rockview Formation, Rapson Creek Formation, and Bush Bay Formation). Building on observations and differentiation from previous investigations, outcrops were selected to describe the macroscopic characteristics of these formations in attempt to develop a local description that would be more useful compared to confusing regional descriptions for researchers and forest officials. In Mackinac County, the Cordell Formation is a dark-to-light gray, chert-rich dolostone with abundant corals, stromataporoids, brachiopods, and echinoids. The Rockview Formation is a dark gray-to-white, chert-poor dolostone with microbiolite laminations, and brachiopod molds. The Rapson Creek Formation is a light gray-to-white, chert-free dolostone with abundant stromatoporoids. The Bush Bay Formation is light reddish brown to white in color, chert-poor, crystalline dolostone with abundant echinoids, brachiopods, and corals.

Rapid Creation of a Data Product for the World's Specimens of Horseshoe Bats and Relatives, a Known Reservoir for Coronaviruses

Genomic evidence suggests that the causative virus of COVID-19 (SARS-CoV-2) was introduced to humans from horseshoe bats (family Rhinolophidae) (Andersen et al. 2020) and that species in this family as well as in the closely related Hipposideridae and Rhinonycteridae families are reservoirs of several SARS-like coronaviruses (Gouilh et al. 2011). Specimens collected over the past 400 years and curated by natural history collections around the world provide an essential reference as we work to understand the distributions, life histories, and evolutionary relationships of these bats and their viruses. While the importance of biodiversity specimens to emerging infectious disease research is clear, empowering disease researchers with specimen data is a relatively new goal for the collections community (DiEuliis et al. 2016). Recognizing this, a team from Florida State University is collaborating with partners at GEOLocate, Bionomia, University of Florida, the American Museum of Natural History, and Arizona State University to produce a deduplicated, georeferenced, vetted, and versioned data product of the world's specimens of horseshoe bats and relatives for researchers studying COVID-19. The project will serve as a model for future rapid data product deployments about biodiversity specimens. The project underscores the value of biodiversity data aggregators iDigBio and the Global Biodiversity Information Facility (GBIF), which are sources for 58,617 and 79,862 records, respectively, as of July 2020, of horseshoe bat and relative specimens held by over one hundred natural history collections. Although much of the specimen-based biodiversity data served by iDigBio and GBIF is high quality, it can be considered raw data and therefore often requires additional wrangling, standardizing, and enhancement to be fit for specific applications. The project will create efficiencies for the coronavirus research community by producing an enhanced, research-ready data product, which will be versioned and published through Zenodo, an open-access repository (see doi.org/10.5281/zenodo.3974999). In this talk, we highlight lessons learned from the initial phases of the project, including deduplicating specimen records, standardizing country information, and enhancing taxonomic information. We also report on our progress to date, related to enhancing information about agents (e.g., collectors or determiners) associated with these specimens, and to georeferencing specimen localities. We seek also to explore how much we can use the added agent information (i.e., ORCID iDs and Wikidata Q identifiers) to inform our georeferencing efforts and to support crediting those collecting and doing identifications. The project will georeference approximately one third of our specimen records, based on those lacking geospatial coordinates but containing textual locality descriptions. We furthermore provide an overview of our holistic approach to enhancing specimen records, which we hope will maximize the value of the bat specimens at the center of what has been recently termed the "extended specimen network" (Lendemer et al. 2020). The centrality of the physical specimen in the network reinforces the importance of archived materials for reproducible research. Recognizing this, we view the collections providing data to iDigBio and GBIF as essential partners, as we expect that they will be responsible for the long-term management of enhanced data associated with the physical specimens they curate. We hope that this project can provide a model for better facilitating the reintegration of enhanced data back into local specimen data management systems.

A New Multiscale Computational Model for Low Salinity Waterflooding in Clay Bearing Sandstones

We develop a new multiscale model to compute effective properties such as relative permeability, contact angle and partition coefficients in low salinity enhanced oil recovery processes for two-phase flow in sandstones containing reactive surfaces of kaolinite clay. In this setting, we construct a three-scale approach which entails the local nanoscale description ruled by the electro-chemistry of a confined electrolyte solution containing Na+, $$Ca^{2+}$$ , $$H^+$$ , $$Cl^-$$ and $$OH^-$$ ions residing between bounded crude-oil droplets at residual saturation and clay substrate. Our analysis focuses on the case of surface complexation geochemical reactions between the ionic species of the invading water and the electrically charged kaolinite and oil–water interfaces. In this scenario, we construct a local electric double layer problem for the electric potential based on a non-symmetric Poisson–Boltzmann equation supplemented by nonlinear boundary conditions with the magnitude of the surface charge strongly dictated by the geochemical reactions. By invoking the local mechanical equilibrium of the electrolyte solution and solving numerically the nonlinear problem using the finite element method, we compute the local ionic profiles and reconstruct numerically the disjoining pressure and adsorption isotherms for each ionic species for a wide range of brine compositions and pH of the water phase. Furthermore, combining the disjoining pressure results with the Frumkin/Derjaguin wetting theory allows to compute the dependence of the contact angle on wettability, pH and salinity. Subsequently, the formal homogenization procedure is adopted to upscale the pore-scale flow and ion transport to the macroscale giving rise to a new Darcy scale coupled flow/transport model. The hyperbolic part of the nonlinear homogenized model is solved analytically in an 1D example of enhanced oil recovery.

Improved Georeferencing: Three essential guiding documents

To understand biological and geological events and the history of collected samples, it is essential to determine and communicate location information accurately. The accuracy of a georeference depends upon the circumstances of the event. Historical collections depend on having clear verbatim locality descriptions, the correct interpretation of data written on labels, and on the availability of gazetteers and maps of suitable scale and time. Observation and tracking data localities depend on GPS (Global Positiioning System) accuracy, and on presence or absence of nearby obstructions such as buildings, forest cover, cliffs, etc. Marine data depend on the accurate determination of the surface location and the techniques to position a dive event from that location and to determine its depth and extent. Many people are using smartphones or maps such as Google Earth and Google Maps to determine their georeferences – but are they suitable and accurate enough to determine locations and elevations? New editions of the Georeferencing Best Practices (Chapman and Wieczorek 2020), the Georeferencing Quick Reference Guide (Zermoglio et al. 2020), and the Georeferencing Calculator Manual (Bloom et al. 2020), were published earlier this year and address all the issues listed above and many more. These documents were based on earlier versions but have been updated and improved considerably – adding information for marine biomes, caves, lithographic stratifications, transects, and the use of elevation, as well as including many more illustrations and examples. The expansion of an extensive georeferencing glossary adds to consistency in the use of terms The trio of documents now provides consistent guidance about how to georeference diverse locality types and detailed instructions on how to calculate uncertainty using many different coordinate reference systems and datums (horizontal and vertical) along with the importance of recording this information. Finally, they provide guidance on how to set up a georeferencing project and how to relate the results to the Darwin Core Standard (Darwin Core Task Group 2009). For the last decade, Darwin Core (Wieczorek et al. 2012) has been one of the preferred standards for sharing biodiversity data, including associated location information. Darwin Core currently has 44 terms in its Location class, allowing sharing from administrative divisions, to elevations and depths, coordinates in different formats, and georeference metadata, among others. Although Darwin Core provides definitions for each of its terms, their correct use is sometimes poorly understood, resulting in information being captured incorrectly, or not captured, documented or shared at all. We will re-introduce these documents, discuss their content, importance, and differences from previously published versions. The newly revised documents provide guidance on capturing and documenting georeferences, clarifying the georeferencing process and showing how to capture information using Darwin Core appropriately. They will improve the location data associated with biological events and our understanding of these events.

Creating Incremental Models of Indoor Environments through Omnidirectional Imaging

In this work, an incremental clustering approach to obtain compact hierarchical models of an environment is developed and evaluated. This process is performed using an omnidirectional vision sensor as the only source of information. The method is structured in two loop closure levels. First, the Node Level Loop Closure process selects the candidate nodes with which the new image can close the loop. Second, the Image Level Loop Closure process detects the most similar image and the node with which the current image closed the loop. The algorithm is based on an incremental clustering framework and leads to a topological model where the images of each zone tend to be clustered in different nodes. In addition, the method evaluates when two nodes are similar and they can be merged in a unique node or when a group of connected images are different enough to the others and they should constitute a new node. To perform the process, omnidirectional images are described with global appearance techniques in order to obtain robust descriptors. The use of such technique in mapping and localization algorithms is less extended than local features description, so this work also evaluates the efficiency in clustering and mapping techniques. The proposed framework is tested with three different public datasets, captured by an omnidirectional vision system mounted on a robot while it traversed three different buildings. This framework is able to build the model incrementally, while the robot explores an unknown environment. Some relevant parameters of the algorithm adapt their value as the robot captures new visual information to fully exploit the features’ space, and the model is updated and/or modified as a consequence. The experimental section shows the robustness and efficiency of the method, comparing it with a batch spectral clustering algorithm.

LRF-Net: Learning Local Reference Frames for 3D Local Shape Description and Matching.

The local reference frame (LRF) acts as a critical role in 3D local shape description and matching. However, most existing LRFs are hand-crafted and suffer from limited repeatability and robustness. This paper presents the first attempt to learn an LRF via a Siamese network that needs weak supervision only. In particular, we argue that each neighboring point in the local surface gives a unique contribution to LRF construction and measure such contributions via learned weights. Extensive analysis and comparative experiments on three public datasets addressing different application scenarios have demonstrated that LRF-Net is more repeatable and robust than several state-of-the-art LRF methods (LRF-Net is only trained on one dataset). We show that LRFNet achieves 0.686 MeanCos performance on the UWA 3D modeling (UWA3M) dataset, outperforming the closest method by 0.18. In addition, LRF-Net can significantly boost the local shape description and 6-DoF pose estimation performance when matching 3D point clouds.

Products of Finite Connected Subgroups

For a non-empty class of groups L, a finite group G=AB is said to be an L-connected product of the subgroups A and B if ⟨a,b⟩∈L for all a∈A and b∈B. In a previous paper, we prove that, for such a product, when L=S is the class of finite soluble groups, then [A,B] is soluble. This generalizes the theorem of Thompson that states the solubility of finite groups whose two-generated subgroups are soluble. In the present paper, our result is applied to extend to finite groups previous research about finite groups in the soluble universe. In particular, we characterize connected products for relevant classes of groups, among others, the class of metanilpotent groups and the class of groups with nilpotent derived subgroup. Additionally, we give local descriptions of relevant subgroups of finite groups.

Local grammars and diachronic speech act analysis

AbstractThis paper proposes a method that is designed to facilitate diachronic speech act analysis. The proposed method draws on the corpus linguistic concept of local grammar – an approach which seeks to account for, not the whole of a language, but one meaning or function only. Local grammar descriptions capture both formal and semantic regularities of speech act realisations, and local grammars offer a more reliable way to quantify speech act realisations across time. It is particularly in this respect that it is argued that a local grammar approach can be useful for diachronic speech act studies, which is demonstrated subsequently by tracing one particular speech act, namely “apology”, in a sample of the Corpus of Historical American English (coha).

Application of the IDEMIX Concept for Internal Gravity Waves in the Atmosphere

AbstractThe model Internal Wave Dissipation, Energy and Mixing (IDEMIX) presents a novel way of parameterizing internal gravity waves in the atmosphere. IDEMIX is based on the spectral energy balance of the wave field and has previously been successfully developed as a model for diapycnal diffusivity, induced by internal gravity wave breaking in oceans. Applied here for the first time to atmospheric gravity waves, integration of the energy balance equation for a continuous wave field of a given spectrum, results in prognostic equations for the energy density of eastward and westward gravity waves. It includes their interaction with the mean flow, allowing for an evolving and local description of momentum flux and gravity wave drag. A saturation mechanism maintains the wave field within convective stability limits, and a closure for critical-layer effects controls how much wave flux propagates from the troposphere into the middle atmosphere. Offline comparisons to a traditional parameterization reveal increases in the wave momentum flux in the middle atmosphere due to the mean-flow interaction, resulting in a greater gravity wave drag at lower altitudes. Preliminary validation against observational data show good agreement with momentum fluxes.

Fluctuational electrodynamics in atomic and macroscopic systems: van der Waals interactions and radiative heat transfer

We present an approach to describing fluctuational electrodynamic (FED) interactions, particularly van der Waals (vdW) interactions as well as radiative heat transfer (RHT), between material bodies of vastly different length scales, allowing for going between atomistic and continuum treatments of the response of each of these bodies as desired. Any local continuum description of electromagnetic (EM) response is compatible with our approach, while atomistic descriptions in our approach are based on effective electronic and nuclear oscillator degrees of freedom, encapsulating dissipation, short-range electronic correlations, and collective nuclear vibrations (phonons). While our previous works using this approach have focused on presenting novel results, this work focuses on the derivations underlying these methods. First, we show how the distinction between "atomic" and "macroscopic" bodies is ultimately somewhat arbitrary, as formulas for vdW free energies and RHT look very similar regardless of how the distinction is drawn. Next, we demonstrate that the atomistic description of material response in our approach yields EM interaction matrix elements which are expressed in terms of analytical formulas for compact bodies or semianalytical formulas based on Ewald summation for periodic media; we use this to compute vdW interaction free energies as well as RHT powers among small biological molecules in the presence of a metallic plate as well as between parallel graphene sheets in vacuum, showing strong deviations from conventional macroscopic theories due to the confluence of geometry, phonons, and EM retardation effects. Finally, we propose formulas for efficient computation of FED interactions among material bodies in which those that are treated atomistically as well as those treated through continuum methods may have arbitrary shapes, extending previous surface-integral techniques.

Local projections for high-dimensional outlier detection

A novel approach for outlier detection is proposed, called local projections, which is based on concepts of the Local Outlier Factor (LOF) (Breunig et al. in Lof: identifying density-based local outliers. In: ACM sigmod record, ACM, volume 29, pp. 93–104, 2000) and ROBPCA (Hubert et al. in Technometrics 47(1):64–79, 2005). By using aspects of both methods, this algorithm is robust towards noise variables and is capable of performing outlier detection in multi-group situations. The idea is to focus on local descriptions of the observations and their neighbors using linear projections. The outlyingness of an observation is determined by a weighted distance of the observation to all identified projection spaces, with weights depending on the appropriateness of the local description. Experiments with simulated and real data demonstrate the usefulness of this method when compared to existing outlier detection algorithms.

Dielectric response with short-ranged electrostatics

The dielectric nature of polar liquids underpins much of their ability to act as useful solvents, but its description is complicated by the long-ranged nature of dipolar interactions. This is particularly pronounced under the periodic boundary conditions commonly used in molecular simulations. In this article, the dielectric properties of a water model whose intermolecular electrostatic interactions are entirely short-ranged are investigated. This is done within the framework of local molecular-field theory (LMFT), which provides a well-controlled mean-field treatment of long-ranged electrostatics. This short-ranged model gives a remarkably good performance on a number of counts, and its apparent shortcomings are readily accounted for. These results not only lend support to LMFT as an approach for understanding solvation behavior, but also are relevant to those developing interaction potentials based on local descriptions of liquid structure.

Rupture of alumina coatings on C35 steel: A numerical simulation

The knowledge of the rupture mechanism of multilayer systems is necessary for improving the bonding of materials, particularly for the ceramic-to-metal junctions, and especially in the case of the ceramic coatings on metallic substrates. For this purpose, coupling experimental measurements of macroscopic interfacial adhesions with numerical tools of fracture mechanics allow understanding how the stress distribution and the cracking mechanisms are influenced by the local environment. This brings an explanation to the cracks initiation and propagation and to the role of local characteristics of the interfacial zone. The present work illustrates this, in the case of alumina coatings strongly bonded to C35 steel substrates previously pre-oxidized in CO2. The experimental tool is based on the “silver print test”. It consists in covering the central part of the samples with a thin layer of silver paint before coating in order to create a defect zone. Concerning the numerical part, the finite element method is used as a powerful tool to describe the mechanical phenomena at the local scale influencing the crack propagation. The obtained numerical results show a good agreement with the experimental observations and they allow a local description of the phenomena influencing the cracking mechanism. Moreover, they lead to a concrete proposition for improving the coatings adherence.

Evolution Equations for Quantum Semi-Markov Dynamics.

Using a newly introduced connection between the local and non-local description of open quantum system dynamics, we investigate the relationship between these two characterisations in the case of quantum semi-Markov processes. This class of quantum evolutions, which is a direct generalisation of the corresponding classical concept, guarantees mathematically well-defined master equations, while accounting for a wide range of phenomena, possibly in the non-Markovian regime. In particular, we analyse the emergence of a dephasing term when moving from one type of master equation to the other, by means of several examples. We also investigate the corresponding Redfield-like approximated dynamics, which are obtained after a coarse graining in time. Relying on general properties of the associated classical random process, we conclude that such an approximation always leads to a Markovian evolution for the considered class of dynamics.