Adjacent Objects Research Articles

Variable range random walk

Exploiting the coherent medium approximation, I investigate a random walk on objects distributed randomly in a continuous space when the jump rate depends on the distance between two adjacent objects. In one dimension, it is shown that when the jump rate decays exponentially in the long distance limit, a non-diffusive to diffusive transition occurs as the density of sites is increased. In three dimensions, the transition exists when the jump rate has a super Gaussian decay.

Contrast Sensitivity Changes after Phototherapeutic Keratectomy in Heterozygote Granular Corneal Dystrophy Type 2

Purpose: Contrast sensitivity, the ability to distinguish the relative difference in luminance of an object from its surrounding or adjacent objects, is a useful measure of visual function. In granular corneal dystrophy type 2 (GCD2), opacity of the corneal stroma causes deterioration in visual function. We compared the contrast sensitivity of GCD2 patients before and after phototherapeutic keratectomy (PTK) to evaluate the perioperative visual function in these patients.Methods: This study included 22 eyes of heterozygote GCD2 patients. The visual acuity and contrast sensitivity were measured before and after PTK. The contrast sensitivity was measured in mesopic and photopic background luminances, with glare (G) levels of 0-2 (G0, G1, and G2, respectively) and spatial frequencies at 1.5, 3, 6, 12, and 18 cycles per degree (cpd).Results: The contrast sensitivity increased after PTK at 1.5 and 3 cpd in mesopic and photopic conditions with G0-2 glare (p < 0.05). At 6 cpd, the contrast sensitivity increased in the mesopic condition with G1 glare, and in the photopic condition with G0-2 glare (p < 0.05). However, there was no change in contrast sensitivity at any glare level at 12 and 18 cpd after PTK.Conclusions: In GCD2 patients, the contrast sensitivity increased significantly after PTK. The vision of GCD2 patients, which is decreased due to corneal opacity, is improved after PTK.

Extraction of indoor objects based on the exponential function density clustering model

Indoor point cloud includes wall, ceiling, floor and many other indoor objects. Extraction of wall, ceiling, floor and many objects in the room is the key for many applications including object identification, facility management and reconstruction of building. In view of this, this paper uses exponential function to construct density clustering model according to the local density within cutoff distance. First, the distances between boundary and indoor point cloud are used to construct the constraint condition of wall density clustering. Second, ceiling and floor are extracted according to the density clustering of z direction and local density model. Third, the distance δi is determined according to the size of local density. Simultaneously, we find that the cluster centers are recognized as points for which the product value of the local density and distance δi is anomalously large. Finally, the cluster belonging of each point to the cluster center is determined according to the distance between each point and cluster center. Finally, indoor objects extraction is achieved by judging the distance between neighboring clusters. We conduct the extraction of indoor objects of different type scenes and compare with clustering and deep learning methods. Comparison results show that the proposed method is superior to the clustering and deep learning methods when adjacent objects are not close to each other, but inferior to the deep learning methods when they are next to each other. Additionally, the extraction accuracy, recall and F1-score are calculated according to the matching rate, TP, FP and FN. It illustrates that the performance of the proposed method is affected by the degree of closeness between objects.

POSITION-SENSITIVE ATTENTION BASED ON FULLY CONVOLUTIONAL NEURAL NETWORKS FOR LAND COVER CLASSIFICATION

Abstract. Pixel-wise land cover classification is a fundamental task in remote sensing image interpretation, aiming to identify planimetric features (e.g., trees, waters, buildings etc.) from earth's surface. Recently, deep learning methods based on fully convolutional neural networks (FCN) become the mainstream approach for land cover classification, thanks to their superior performance in the image context perception and features learning. However, for high-resolution remote sensing images with huge quantity of object details, some deep learning based methods often ignore many important details by nature, specially, in the procedure of pooling operation and stacking convolutions in conventional FCN, it can leads to ambiguous classification of adjacent objects. To refine lost details caused by the stacking convolutions, we propose a position-sensitive attention (PSA) based on skip connections for land cover classification with high-resolution remote sensing images, which designs to deliver a weight that is sensitive to the spatial details in remote sensing images, the PSA module is able to improve pixel-level details scattered across spatial positions. Experimental results demonstrate that our method can be feasible to existing FCN-based models, 1% improvement in F1-score is obtained on 2021 "Shengteng Cup" competition dataset after using PSA, when comparing to several state-of-the-art methods, similar or even better performance is achieved on the ISPRS Vaihingen 2D dataset, but with less parameters.

Enhancing critical resolution of a ghost imaging system by using a vortex beam.

In an imaging system, resolution and signal-to-noise ratio (SNR) are two important indexes to characterize imaging quality. Ghost imaging is a novel imaging method whose imaging resolution and SNR are affected by the speckle size. In this paper, the relation between speckle size and resolution as well as that between speckle size and SNR in the GI system is analyzed in detail. It is shown that the critical resolution, resolvable minimum-separation between two adjacent objects, is approximately equal to the speckle size (speckle diameter). There exists an optimum SNR when the speckle size is larger than the object size. Based on our conclusion, we propose a scheme to enhance the critical resolution of the GI system by using a vortex beam, and the enhancement ability under different topological charges is clearly presented, which can be quantized by a simple formula.

Scale-aware attention network for weakly supervised semantic segmentation

Weakly supervised semantic segmentation (WSSS) using image-level labels greatly alleviates the burden of obtaining large amounts of pixel-wise annotations. To create pseudo segmentation labels, most WSSS algorithms rely on the local response regions in the class activation maps (CAMs). However, such activation maps only focus on the local discriminative parts of the object, because the classification network does not require the entire object to optimize the objective function. To enhance the network’s ability to focus more on non-discriminative parts of the object and generate high-quality pseudo-masks, the Scale-aware Attention Network (SAN) is proposed. Specifically, a pyramidal attention module is introduced to propagate discriminative information to adjacent object regions by adaptively selecting contextual features from the convolutional pyramid with varied filter scales. A multi-scale prediction fusion structure with a joint loss is proposed to make better use of the complementary information of localization maps in different scales. The dense and integral localization maps are obtained in the inference stage by weighted fusion of the multi-scale predictions, which are then used to train segmentation models. This novel SAN has demonstrated its effectiveness in a series of experiments. It has achieved a state-of-the-art result of 71.9% mIoU on the PASCAL VOC 2012 segmentation test set compared with other approaches under the same level of supervision.

Positioning the Adjacent Buried Objects Using UWB Technology Combine with Levenberg Marquardt Algorithm

The determination of the buried objects and cracks in building structures is an important issue in real-life. In this paper, we propose a new method called Correlation Function Separation Technique (CFST) combine with the Lervenberg-Marquardt Algorithm (LMA) using the Impulse Radio Ultra-Wide Band (IR-UWB) penetrating system to improve the accuracy in detecting and positioning of the adjacent buried objects in building structures. Based on the UWB signal processing, the proposed method can be used to determine both the relative permittivity of the environment and the position of the buried objects, especially the adjacent objects. The analytical method is validated by mathematical proofs and Matlab simulations, and the position errors are used to assess the performance of proposed method. The numerical results shown that the proposed method can be used for positioning the adjacent buried objects in the homogeneous environment which has an average positioning error of 3.52cm, which is smaller than that of the conventional method based on B-canned radar images processing.

Repulsion Distance Intersection over Union Loss for Crowd Scene Detection

The detection of many people or crowds increases the false-positive prediction rate owing to the creation of more bounding boxes. In this study, a loss function that reduces the false positive predictions of an object detection model is proposed. The proposed loss function induces not only the bounding box to be closer to the ground truth (GT) box but also the center point distance between the GT box and the overlapped adjacent object or other bounding boxes to be longer. In addition, a balanced feature pyramid was introduced to enhance the precision of object prediction. Applying the proposed method, 𝑡ℎ𝑒 log-average miss rate on false positive per image in [10<SUP>−2</SUP>, 100] was 1.89% lower while average precision and Jaccard index were 0.5% and 0.58% higher, respectively, than those of the conventional method, which indicates that it effectively reduces false positive predictions.

Probabilistic context integration‐based aircraft behaviour intention classification at airport ramps

Self-driving baggage tractors driving on airport ramps/aprons present new trends that promote better airport operation procedures and proliferate the aviation market. Airport ramps have unique mobility requirements when it comes to layout, population, demand, and patterns. Estimating aircraft movement is highly crucial and must be required because of safety reasons and airport operations rules. The movement of aircraft at the airport ramp is not dynamic but relatively static and slow. However, it is more complicated in estimating attributes. Even though the aircraft is parked and stationary, all operational vehicles should be cautious on aircraft pushing in or out from the parking stand. However, this is never dealt with in any research since aircraft detection observed at the airport ramp is not suggested in previous research. In this paper, a context-aware fusion-based method for aircraft intention detection at airport ramps is proposed. A parallel extraction of behavioural features from the aircraft and situational context detection from other adjacent objects are suggested. Using the proposed method, aircraft can be detected and aircraft movement context can be estimated. The feasibility of this algorithm is demonstrated based on the fluent dataset obtained at Cincinnati Airport.

Object-Based Classification Framework of Remote Sensing Images With Graph Convolutional Networks

Object-based image classification (OBIC) on very-high-resolution (VHR) remote sensing (RS) images is utilized in a wide range of applications. Nowadays, many existing OBIC methods only focus on features of each object itself, neglecting the contextual information among adjacent objects and resulting in low classification accuracy. Inspired by a spectral graph theory, we construct a graph structure from objects generated from VHR RS images and propose an OBIC framework based on truncated sparse singular value decomposition and graph convolutional network (GCN) model, aiming to make full use of relativities among objects and produce an accurate classification. Through conducting experiments on two annotated RS image data sets, our framework obtained 97.2% and 66.9% overall accuracy, respectively, in automatic and manual object segmentation circumstances, within a processing time of about 1/100 of convolutional neural network (CNN)-based methods’ training time.

Star Mass Inertia Dictates the Speed of Light

This paper indicated that Newton’s law of gravitation is a statistical relation of two adjacent objects with a distance and Einstein’s general relativity shows the internal connection of material distribution through space-time warping in our universe so general relativity is referred as a gravity theory. Here the paper gives an extension of general relativity in the sense of the metric theory of gravity which is consistent with Einstein’s equivalence principle and generates a weak field approximation which extends Newtonian dynamics. Thus, the extended theory of gravity can infer that the light has a speed limit of photon escaping the mass inertia of a star.

Four Kinds of Potential Mass-Energy Space-Time Distribution Unified by a Dynamical Approach

The unification of gravity and electricity since the early 20th century and the unification of four fundamental forces since the 1970s have become a mainstream of physical study by using different theories and methods. This paper introduces a dynamical approach to respectively reveal the physical natures of gravity and electricity as well as strong and weak forces. The results showed that the nature of gravity is a dynamical product of mass-energy linear distributions between two adjacent objects and the nature of electricity is a dynamical product of mass-energy linear distributions between two adjacent unlike charges. Potential nuclear energies are mathematically shown by the products of mass-energy linear distributions between proton-neutron particles as well as hadron-lepton particles, which are also initially generated through external perpendicular forces with the maximum efficiency. Thus, four kinds of potential mass-energy space-time distribution rather than four long- and short-range forces showed a common mathematical expression or similarity among them by using this approach.

Zoning of rice growing territory of Krasnodar region by the set of features of Pyricularia oryzae Cav. isolation

The geographical location and relief of the Krasnodar Territory determines a wide variety of agro-climatic and agro-technological conditions for the production of crops. On the territory of the Krasnodar Territory there is the largest part of agricultural land intended for the production of rice crops in Russia. Annual infection of rice plants with pathogenic Piricularia Orizae Cav. due to a wide range of strain diversity and indicates a significant dependence of the population genotype on the territorial distribution area. Therefore, the identification of zones that determine the conditions for the spread of blast within the population plays an important role in studying the dynamics of pathogen development in the territory of rice cultivation. The results of the research work showed that the Krasnodar rice growing area has a uniform landscape without sharp changes in the relief profile. The location of adjacent geographical objects affects the agro-climatic conditions of the southern part of the territory, which indicates the likelihood of isolation of the population of rice blast strains in the southern part of the rice growing area. An analysis of the agricultural zoning of the Krasnodar Territory showed the isolation of the southern part of the rice-growing territory according to the characteristics of the soil cover and the isolation of the conditions of agrotechnological measures. Features of the water regime of the irrigation systems of the Krasnodar Territory indicate the differentiation of the factors of infection and the spread of the pathogen on the basis of the isolation of irrigation flows. The territory has a clear division along the channels of the Kuban and Protoka rivers.

Contextual-Aware Land Cover Classification With U-Shaped Object Graph Neural Network

The timely and accurate land cover mapping with remote sensing images played a huge role in ecosystem monitoring. However, due to the spectral variability and spatial complexity of high-resolution remote sensing images, it is often difficult to find an efficient method to achieve accurate land cover classification. To explore useful contextual and hierarchical features in remote sensing images, this paper proposes a U-shaped object graph neural network (U-OGNN), which is mainly composed of self-adaptive graph construction (SAGC), hierarchical graph encoder, and decoder. For self-adaptive graph construction, the similarity measurement is applied to generate contextual-aware graph structure, by feeding deep features extracted from convolution and multi-layer attention operations. Graph encoder and decoder fuse multi-level information over different scales by capturing hierarchical features of adjacent objects. In this way, the proposed method is able to predict land-cover types by considering multi-level contextual information accurately. Experiments on GID land-cover classification datasets, the overall accuracies of the U-OGNN reach 87.81 %.

Reconstructing images of two adjacent objects passing through scattering medium via deep learning

In this paper, to the best of our knowledge, we first present a deep learning based method for reconstructing the images of two adjacent objects passing through scattering media. We construct an imaging system for imaging of two adjacent objects located at different depths behind the scattering medium. In general, as the light field of two adjacent objects passes through the scattering medium, a speckle pattern is obtained. We employ the designed adversarial network, which is called as YGAN, for reconstructing the two images simultaneously from the speckle. It is shown that based on the trained YGAN, we can reconstruct images of the two adjacent objects with high quality. In addition, the influence of object image types, and the location depths of the two adjacent objects on the imaging fidelity will be studied. Results demonstrate the strong generalization ability and effectiveness of the YGAN. Even in the case where another scattering medium is inserted between the two objects, the YGAN can reconstruct the object images with high fidelity. The technique presented in this paper can be used for applications in areas of medical image analysis, such as medical image classification, segmentation, and studies of multi-object scattering imaging, three-dimensional imaging etc.

Critical resolution in ghost imaging system with pseudo-thermal light

The resolution limit, which characterizes the resolution power of an optical imaging system, is defined as the resolvable minimum-separation between two points. In classical optical imaging, the resolution limit is given by the Rayleigh diffraction limit. In this paper, we propose a concept of critical resolution, which is used to describe the minimum separation of two adjacent objects that an imaging system can resolve. The critical resolution of pseudo-thermal ghost imaging (PGI) system is analyzed. To quantify the critical resolution, we develop a mathematical formula from which one can evaluate whether a prescribed object can be resolved in a PGI system. The evolution process from critical resolution to resolution limit is presented when the object size towards to zero. The results show that the resolution limit of PGI system is 1.98 times the speckle size. Our results can not only promote the development of GI technology, but also provide theoretical guidance for the design of GI system.

A coarse-to-fine boundary refinement network for building footprint extraction from remote sensing imagery

Extracting building footprints from remotely sensed imagery has long been a challenging task and is not yet fully solved. Obstructions from nearby shadows or trees, varying shapes of rooftops, omission of small buildings, and varying scale of buildings hinder existing automated models for extracting sharp building boundaries. Different reasons account for these challenges. In convolutional neural network-based methods, the down-sampling operation loses spatial details of the input images; and small buildings are omitted from the high-level features. The sheltering trees and adjacent objects shadowing may cause errors since semantic information cannot be effectively preserved. Moreover, the insufficient use of multi-scale building features causes blurry edges in the predictions for buildings with complex shapes. To address these challenges, we propose a novel coarse-to-fine boundary refinement network (CBR-Net) that accurately extracts building footprints from remote sensing imagery. Unlike the existing semantic segmentation methods that directly generate building predictions at the highest level, we designed a module that progressively refines the building prediction in a coarse-to-fine manner. In this way, the advantages of both the high-level and low-level features can be retained. We also present a novel boundary refinement (BR) module that enhances the ability of the CBR-Net model to perceive and refine building edges. The BR module refines building prediction by perceiving the direction of each pixel in a remotely sensed optical image to the center of the nearest object to which it might belong. The refined results are used as pseudo labels in a self-supervision process that increases model robustness to noisy labels or obstructions. Experimental results on three public building datasets, including the WHU building dataset, the Massachusetts building dataset, and the Inria aerial image dataset, demonstrate the effectiveness of the proposed method. In evaluation tests, CBR-Net outperformed other state-of-the-art algorithms on the three datasets by maintaining both the continuous entities and accurate boundaries of buildings. The source code of the proposed CBR-Net is available at https://github.com/HaonanGuo/CBRNet.

Metonymy in scientific linguistic discourse

The purpose of the article was to show the features of the functioning of different types of metonymy in scientific linguistic discourse, which is understood as a verbalized epistemic situation common to the scientific sphere of communication, taken in the entire totality of linguistic and extralinguistic factors and enshrined in the form of texts (oral and written ones). The article deals with metonymy from the point of view of langue / parole: lexicalized metonymy in langue is a semantic transposition mechanism on contiguity and carries out a terminological nomination; discursive metonymy in parole becomes the result of syntagmatic contiguity of syntactic constructions. Linguistic metonymic terms are grouped by types of knowledge: declarative and procedural ones. The shifts of meaning between the logical terms “object”, “subject”, “general” and “specific”, “abstract” and “concrete”, “form”, “content”, etc., directed towards each other, are observed in metonymic terms of declarative type. Metonymy can reflect the processes due to the causality between adjacent objects. Transitional phenomena between lexicalized (linguistic) and discursive (speech) metonymy reflect those models that contain onyms; they are related to the designation of the subject of knowledge (linguist) and his scientific discovery.

Online Multiple Object Tracking Using a Novel Discriminative Module for Autonomous Driving

Multi object tracking (MOT) is a key research technology in the environment sensing system of automatic driving, which is very important to driving safety. Online multi object tracking needs to accurately extend the trajectory of multiple objects without using future frame information, so it will face greater challenges. Most of the existing online MOT methods are anchor-based detectors, which have many misdetections and missed detection problems, and have a poor effect on the trajectory extension of adjacent object objects when they are occluded and overlapped. In this paper, we propose a discrimination learning online tracker that can effectively solve the occlusion problem based on an anchor-free detector. This method uses the different weight characteristics of the object when the occlusion occurs and realizes the extension of the competition trajectory through the discrimination module to prevent the ID-switch problem. In the experimental part, we compared the algorithm with other trackers on two public benchmark datasets, MOT16 and MOT17, and proved that our algorithm has achieved state-of-the-art performance, and conducted a qualitative analysis on the convincing autonomous driving dataset KITTI.

Building Multi-Feature Fusion Refined Network for Building Extraction from High-Resolution Remote Sensing Images

Deep learning approaches have been widely used in building automatic extraction tasks and have made great progress in recent years. However, the missing detection and wrong detection causing by spectrum confusion is still a great challenge. The existing fully convolutional networks (FCNs) cannot effectively distinguish whether the feature differences are from one building or the building and its adjacent non-building objects. In order to overcome the limitations, a building multi-feature fusion refined network (BMFR-Net) was presented in this paper to extract buildings accurately and completely. BMFR-Net is based on an encoding and decoding structure, mainly consisting of two parts: the continuous atrous convolution pyramid (CACP) module and the multiscale output fusion constraint (MOFC) structure. The CACP module is positioned at the end of the contracting path and it effectively minimizes the loss of effective information in multiscale feature extraction and fusion by using parallel continuous small-scale atrous convolution. To improve the ability to aggregate semantic information from the context, the MOFC structure performs predictive output at each stage of the expanding path and integrates the results into the network. Furthermore, the multilevel joint weighted loss function effectively updates parameters well away from the output layer, enhancing the learning capacity of the network for low-level abstract features. The experimental results demonstrate that the proposed BMFR-Net outperforms the other five state-of-the-art approaches in both visual interpretation and quantitative evaluation.