Spatial Segmentation Research Articles

Graph-Based Spatial Segmentation of Areal Data

Smoothing is often used to improve the readability and interpretability of noisy areal data. However, there are many instances where the underlying quantity is discontinuous. For such cases, specific methods are needed to estimate the piecewise constant spatial process. A well-known approach in this setting is to perform segmentation of the signal using the adjacency graph, such as the graph-based fused lasso. However, this method does not scale well to large graphs. A new method is introduced for piecewise constant spatial estimation that (i) is faster to compute on large graphs and (ii) yields sparser models than the fused lasso (for the same amount of regularization), resulting in estimates that are easier to interpret. The method is illustrated on simulated data and applied to real data on overweight prevalence in the Netherlands. Healthy and unhealthy zones are identified, which cannot be explained by demographic or socio-economic characteristics alone. The method is found capable of identifying such zones and can assist policymakers with their health improving strategies.

Deep Learning in Single-cell Analysis

Single-cell technologies are revolutionizing the entire field of biology. The large volumes of data generated by single-cell technologies are high dimensional, sparse, and heterogeneous and have complicated dependency structures, making analyses using conventional machine learning approaches challenging and impractical. In tackling these challenges, deep learning often demonstrates superior performance compared to traditional machine learning methods. In this work, we give a comprehensive survey on deep learning in single-cell analysis. We first introduce background on single-cell technologies and their development, as well as fundamental concepts of deep learning including the most popular deep architectures. We present an overview of the single-cell analytic pipeline pursued in research applications while noting divergences due to data sources or specific applications. We then review seven popular tasks spanning different stages of the single-cell analysis pipeline, including multimodal integration, imputation, clustering, spatial domain identification, cell-type deconvolution, cell segmentation, and cell-type annotation. Under each task, we describe the most recent developments in classical and deep learning methods and discuss their advantages and disadvantages. Deep learning tools and benchmark datasets are also summarized for each task. Finally, we discuss the future directions and the most recent challenges. This survey will serve as a reference for biologists and computer scientists, encouraging collaborations.

Recurrent feature propagation and edge skip-connections for automatic abdominal organ segmentation

Automatic segmentation of abdominal organs in computed tomography (CT) images can support radiation therapy and image-guided surgery workflows. Development of such automatic solutions remains challenging mainly owing to complex organ interactions and blurry boundaries in CT images. To address these issues, we focus on effective spatial context modeling and explicit edge segmentation priors. Accordingly, we propose a 3D network with four main components trained end-to-end including shared encoder, edge detector, decoder with edge skip-connections (ESCs) and recurrent feature propagation head (RFP-Head). To capture wide-range spatial dependencies, the RFP-Head propagates and harvests local features through directed acyclic graphs (DAGs) formulated with recurrent connections in an efficient slice-wise manner, with regard to spatial arrangement of image units. To leverage edge information, the edge detector learns edge prior knowledge specifically tuned for semantic segmentation by exploiting intermediate features from the encoder with the edge supervision. The ESCs then aggregate the edge knowledge with multi-level decoder features to learn a hierarchy of discriminative features explicitly modeling complementarity between organs’ interiors and edges for segmentation. We conduct extensive experiments on challenging abdominal CT datasets with eight annotated organs. Experimental results show that the proposed network outperforms several state-of-the-art models, especially for the segmentation of small and complicated structures (gallbladder, esophagus, stomach, pancreas and duodenum).

ApeTI: A Thermal Image Dataset for Face and Nose Segmentation with Apes

The ApeTI dataset was built with the aim of retrieving physiological signals such as heart rate, breath rate, and cognitive load from thermal images of great apes. We want to develop computer vision tools that psychologists and animal behavior researchers can use to retrieve physiological signals noninvasively. Our goal is to increase the use of a thermal imaging modality in the community and avoid using more invasive recording methods to answer research questions. The first step to retrieving physiological signals from thermal imaging is their spatial segmentation to then analyze the time series of the regions of interest. For this purpose, we present a thermal imaging dataset based on recordings of chimpanzees with their face and nose annotated using a bounding box and nine landmarks. The face and landmarks’ locations can then be used to extract physiological signals. The dataset was acquired using a thermal camera at the Leipzig Zoo. Juice was provided in the vicinity of the camera to encourage the chimpanzee to approach and have a good view of the face. Several computer vision methods are presented and evaluated on this dataset. We reach mAPs of 0.74 for face detection and 0.98 for landmark estimation using our proposed combination of the Tifa and Tina models inspired by the HRNet models. A proof of concept of the model is presented for physiological signal retrieval but requires further investigation to be evaluated. The dataset and the implementation of the Tina and Tifa models are available to the scientific community for performance comparison or further applications.

An MIU-based deep embedded clustering model for urban functional zoning from remote sensing images and VGI data

Urban functional zoning offers valuable insights into urban morphology and sustainable development. However, the conventional fixed spatial units, such as blocks and grids, cannot easily capture the morphological characteristics inherent in functional union and separation during urban evolution. In this paper, by taking advantage of remote sensing images and geospatial big data, we propose a minimum identification unit (MIU)-based urban functional zoning model. This approach integrates the deep embedded clustering of buildings to generate the spatial unit segmentation, and then identifies the urban function by generating semantic vectors with the Word2Vec model. The effectiveness of the proposed method was tested in the city of Wuhan in China. The results highlight that MIUs provide a more flexible and suitable unit for segmenting urban functional zones compared to traditional street blocks. The proposed method is a feasible way to deal with the semantic redundancy of volunteered geographic information (VGI) data when identifying urban function, and the quality issue only has a significant impact on the minor functional types. Moreover, the building clustering results can effectively reveal the fine-scale urban structure, especially for the administration, manufacturing, and residential types. This demonstrates the potential of our approach in enhancing the understanding of urban morphology and supporting sustainable urban development.

Will you ride the train? A combined home-work spatial segmentation approach

While the influence of land use and transport networks on travel behavior is known, few studies have jointly examined the effects of home and work location characteristics when modelling travel behavior. In this study, a two-step approach is proposed to investigate the combined effect of home and work location characteristics on the intent to use a new public transport service. Using data from the 2019 Montreal Mobility Survey (n=1698), this study examines the intent to use the Réseau Express Métropolitain (REM), a light rail under construction in Montreal, for commuting. A segmentation analysis is first conducted to characterize commuters based on their home and work location characteristics, resulting in six distinct home-work clusters. The clusters are then included in an ordered logistic regression modelling the intent to use the REM, along with socio-economic and attitudinal characteristics. Results from a dominance analysis reveal that the clusters are the third most important determinants of the intent to use the REM, even when controlling for individual characteristics. The addition of the clusters leads to a significant improvement of the model (likelihood of -2388.9 improved from -2400.7, p-value < 0,05). All other clusters have a significantly lower probability (between 32 and 51% less likely) of intent to use the REM than the typical commuters (who commute from the suburbs to downtown, often by transit), at a 95% confidence interval. These findings underscore the implications of pursuing radial public-transport networks, illustrating the ability of the proposed approach to identify which groups are likely to benefit from a public-transport project and to propose recommendations anchored in joint home and work location patterns.

Policy network evaluation of China’s rental housing market from the perspective of text measurement

In recent years, the Chinese government has gradually paid attention to the development of the rental housing market and began to issue a large number of relevant policies by the central and local governments in 2016. In view of the lack of systematic evaluation of rental housing market policy in China, this study constructed an interdependent multi-layer network of ‘policy objective-policy instrument-issuing department-stakeholder’ from the perspective of text measurement, and used network indexes to measure the global features of the four-dimensional sub-policy network and the differential characteristics of the network evolution process, the results are as follows: (i) The four policy instruments in the network of policy instruments, namely, instruction, supply leading, management system and management system reform, are leaders in the network. (ii) The absolute central node in the policy objective network is to increase the effective supply of rental housing. (iii) The central nodes in the issuing department network are housing development department and the land department. (iv) Market subject and public are the central clusters in the stakeholder network. (v) The network nodes of stakeholder policy instruments have the strongest correlation, and there are three clusters with obvious spatial distance segmentation.

Recognition of pseudoinvasion in colorectal adenoma using spatial glycomics.

Pseudoinvasion (PI) is a benign lesion in which cancer is mimicked in the colon by misplacement of dysplastic glands in the submucosa. Although there are morphological clues, the discrimination of PI from true invasion can be a challenge during pathological evaluation of colon adenomas. Both overdiagnosis and underdiagnosis can result in inadequate clinical decisions. This calls for novel tools to aid in cases where conventional methods do not suffice. We performed mass spectrometry imaging (MSI)-based spatial glycomics analysis on a cohort of formalin-fixed paraffin-embedded tissue (FFPE) material from 16 patients who underwent polypectomy. We used this spatial glycomic data to reconstruct the molecular histology of the tissue section using spatial segmentation based on uniform manifold approximation and projection for dimension reduction (UMAP). We first showed that the spatial glycomic phenotypes of the different morphological entities separated as distinct clusters in colon tissues, we separated true invasion from the other morphological entities. Then, we found that the glycomic phenotype in areas with suspected PI in the submucosa was strongly correlating with the corresponding glycomic phenotype of the adenomatous colon epithelium from the same tissue section (Pearson correlation distance average = 0.18). These findings suggest that using spatial glycomics, we can distinguish PI as having a molecular phenotype similar to the corresponding surface epithelium and true invasion as having a different phenotype even when compared to high-grade dysplasia. Therefore, when a novel molecular phenotype is found in the deepest submucosal region, this may be used as an argument in favor of true invasion.

Contextualized Trajectory Parsing with Spatio-Temporal Graph

This work investigates how to automatically parse object trajectories in surveillance videos, which aims at jointly solving three subproblems: 1) spatial segmentation, 2) temporal tracking, and 3) object categorization. We present a novel representation spatiotemporal graph (ST-Graph) in which: 1) Graph nodes express the motion primitives, each representing a short sequence of small-size patches over consecutive images, and 2) every two neighbor nodes are linked with either a positive edge or a negative edge to describe their collaborative or exclusive relationship of belonging to the same object trajectory. Phrasing the trajectory parsing as a graph multicoloring problem, we propose a unified probabilistic formulation to integrate various types of context knowledge as informative priors. An efficient composite cluster sampling algorithm is employed in search of the optimal solution by exploiting both the collaborative and the exclusive relationships between nodes. The proposed framework is evaluated over challenging videos from public datasets, and results show that it can achieve state-of-the-art tracking accuracy.

Spatial attributes-based segmentation and topological attributes-based recognition algorithm for Myanmar OCR

Spatial attributes-based segmentation and topological attributes-based recognition algorithm for Myanmar OCR

Auto-LIA: The Automated Vision-Based Leaf Inclination Angle Measurement System Improves Monitoring of Plant Physiology.

Plant sensors are commonly used in agricultural production, landscaping, and other fields to monitor plant growth and environmental parameters. As an important basic parameter in plant monitoring, leaf inclination angle (LIA) not only influences light absorption and pesticide loss but also contributes to genetic analysis and other plant phenotypic data collection. The measurements of LIA provide a basis for crop research as well as agricultural management, such as water loss, pesticide absorption, and illumination radiation. On the one hand, existing efficient solutions, represented by light detection and ranging (LiDAR), can provide the average leaf angle distribution of a plot. On the other hand, the labor-intensive schemes represented by hand measurements can show high accuracy. However, the existing methods suffer from low automation and weak leaf-plant correlation, limiting the application of individual plant leaf phenotypes. To improve the efficiency of LIA measurement and provide the correlation between leaf and plant, we design an image-phenotype-based noninvasive and efficient optical sensor measurement system, which combines multi-processes implemented via computer vision technologies and RGB images collected by physical sensing devices. Specifically, we utilize object detection to associate leaves with plants and adopt 3-dimensional reconstruction techniques to recover the spatial information of leaves in computational space. Then, we propose a spatial continuity-based segmentation algorithm combined with a graphical operation to implement the extraction of leaf key points. Finally, we seek the connection between the computational space and the actual physical space and put forward a method of leaf transformation to realize the localization and recovery of the LIA in physical space. Overall, our solution is characterized by noninvasiveness, full-process automation, and strong leaf-plant correlation, which enables efficient measurements at low cost. In this study, we validate Auto-LIA for practicality and compare the accuracy with the best solution that is acquired with an expensive and invasive LiDAR device. Our solution demonstrates its competitiveness and usability at a much lower equipment cost, with an accuracy of only 2. 5° less than that of the widely used LiDAR. As an intelligent processing system for plant sensor signals, Auto-LIA provides fully automated measurement of LIA, improving the monitoring of plant physiological information for plant protection. We make our code and data publicly available at http://autolia.samlab.cn.

Transportation resilience under Covid-19 Uncertainty: A traffic severity analysis

Transportation systems are critical lifelines and vulnerable to various disruptions, including unforeseen social events such as public health crises, and have far-reaching social impacts such as economic instability. This paper aims to determine the key factors influencing the severity of traffic accidents in four different stages during the pre- and the post Covid-19 pandemic in Illinois, USA. For this purpose, a Random Forest-based model is developed, which is combined with techniques of explainable machine learning. The results reveal that during the pandemic, human perceptual factors, notably increased air pressure, humidity and temperature, play an important role in accident severity. This suggests that alleviating driver anxiety, caused by these factors, may be more effective in curbing crash severity than conventional road condition improvements. Further analysis shows that the pandemic leads notable shifts in residents' daily travel time and accident-prone spatial segments, indicating the need for increased regulatory measures. Our findings provide new insights for policy makers seeking to improve transportation resilience during disruptive events.

Rapid survey method for large-scale outdoor surveillance cameras using binary space partitioning

Video surveillance systems have been widely used in public security and urban management. Following the increasing demand for video surveillance in various countries worldwide, the number of outdoor surveillance cameras is rising. Their unified management and layout optimization have become a great challenge in the construction of video surveillance. Although there is a large number of outdoor surveillance cameras, there exists a lack of spatial location information, leading to difficulties regarding their quantification, space management, and construction status analysis. Hence, a method for estimating the location and orientation of outdoor surveillance cameras based on spatial segmentation and the use of a rapid survey device was proposed. Using the position information obtained from the movement of a designed marker and surveillance images simultaneously captured by each camera, we constructed an initial camera space and segmentation plane using binary space partitioning. By recursive segmentation, the real location of each camera is gradually approximated, and the azimuth angle of the camera orientation is calculated, achieving the rapid estimation of the spatial location and orientation of all surveillance cameras. A city area spanning 2.8 km2 was selected as an experimental area to verify the method proposed here, yielding an average error of 1.78 m between the estimated and actual locations of the cameras and a maximum error of 3.87 m. The average offset between the estimated and actual orientation was 3.13°, with a maximum offset of 6°. Thus, the target of the surveillance camera's estimated deviation is still in the visible range of the operator. The proposed method represents a new way for the rapid investigation of outdoor surveillance cameras and provides a basis for their unified management and layout optimization.

A quantitative evaluation model of ancient military defense efficiency based on spatial strength—take Zhejiang of the Ming Dynasty as an example

In the study of the defense capacity of ancient military settlements, most existing studies have evaluated the defense capacity of citadels qualitatively from a historiographical perspective. There are fewer quantitative studies on defense efficiency in a specific regional context. This paper takes the coastal defense citadels system of Zhejiang in the Ming Dynasty of China as the research object. It adopts a combination of digital historical data mining and quantitative research to study the defense efficiency of the military defense area in Zhejiang in the Ming Dynasty. This paper first analyzes the defense pattern of Zhejiang's sea defense citadels system in the Ming Dynasty. It determines the control points of defense groups, quantifies the defense group area with a Voronoi diagram, and classifies and assigns the entropy value to the deployment data of guard forts in historical data. Finally, the assigned forces are superimposed on the defense area to obtain the quantified defense efficiency of each defense group, and the defense results are verified. This paper innovatively proposes an "area-force" perspective. It introduces spatial segmentation and quantitative research methods to study the defense efficiency of a specific regional defense group system in ancient times. It proves that this method is feasible and can be applied to the study of other ancient military defense areas. It also expands the scientific perspective of ancient military fortification systems, which can contribute to systematically preserving ancient military settlement heritage on a larger scale.

MFAGNet: multi-scale frequency attention gating network for land cover classification

ABSTRACT As a classic problem of high-resolution remote sensing image tasks, land cover classification has many challenges, for instance, variable scales and low-edge discrimination. Traditional spatial semantic segmentation methods based on single-scale feature extraction are challenging to model multi-scale features. They cannot effectively fit high-resolution remote sensing images with variable scales and are prone to misclassification issues. On the other hand, spatial multi-scale semantic segmentation methods lack edge constraints, which can easily lead to edge segmentation ambiguity issues. We propose a segmentation network based on a multi-scale frequency-domain attention gating mechanism (MFAGNet) to solve the above problems. Specifically, we obtain multi-scale features through multi-scale input and design the discrete cosine transform channel attention module (DCTCAM). In DCTCAM, we extract global low-frequency features and effectively alleviate misclassification caused by scale changes. The edge features of the image are obtained through the high-frequency feature extraction module (HFEM) by using edge detection operators to enhance the model’s ability to recognize edges. Additionally, no further pooling operations are employed except for a single max pooling layer in the backbone network. In this paper, we design a new gating mechanism to dynamically control the extraction ratio of various frequency features so that it can adapt to input images of different scales. This paper conducts comparative experiments on three high-resolution land cover datasets: GID, Vaihingen, and Potsdam. The new algorithm proposed compares with classic semantic segmentation models such as DANet, PSPNet, SegNet, UNet, DeepLabv3+, FcaNet, BiSeNet, AFN, and FastFCN. The experimental results show that MFAGNet is superior to the comparison models. With no significant increase in computing overhead, MFAGNet obtains MIoU of 69.05%, 73.40%, and 79.42% on the standard data sets GID, Vaihingen, and Potsdam, respectively.

Classification of Speed Change and Unstable Flow Segments Using Geohash-Encoded Vehicle Big Data

Precise and detailed speed information is indispensable for ensuring safe and efficient transportation. This is particularly true within unstable flow (UF) segments, which are especially prone to accidents due to the significant speed variations between vehicles and across lanes, and in the context of evolving transportation systems, where autonomous and non-autonomous vehicles are increasingly mixing. To address the limitations of existing methods in providing such data, this study aims to improve the detail, accuracy, and granularity of road information for micro-segments by leveraging individual vehicle big data. The proposed approach utilizes the geohash algorithm for spatial segmentation and introduces a novel criterion for identifying UF segments based on the relationship between space mean speed (SMS) and time mean speed (TMS). The presented strategy was validated through a comprehensive analysis of DTG (Digital Tachograph) data from freight vehicles on Expressway No. 50 in the Gyeonggi region in the Republic of Korea. As a result, a total of 301 segments were identified, including 178 eastbound and 123 westbound segments. UF segments corresponded to partitions falling beyond the reference standard deviation range. Compared with VDS (Vehicle Detection System) and conzone speeds, the proposed method provided more precise and continuous speed information, surpassing those obtained from conventional link-based approaches.

An automatic extraction method for geothermal radiation sources based on an LST retrieval algorithm and semantic network

Geothermal resources are efficient, renewable and clean energy sources, and their reservoirs are usually closely associated with high-temperature regions of the land surface. Current exploration methods primarily involve migrating traditional geological techniques, which fail to fully use the unique features of geothermal radiation characteristics. Thermal infrared remote-sensing imaging technology can capture and present areas with distinctive surface thermal radiation features, providing considerable significance as a guide for localization prior to field exploration. In this study, we propose a deep learning–based method for intelligently identifying and segmenting geothermal radiation sources from thermal infrared remote-sensing images, including data preparation and model training. To improve the localization drift and anomalous interference caused by the high complexity of the Earth's surface environment, this study uses a surface temperature retrieval algorithm to calculate the land surface temperature in the research area. The retrieval results are used to train the semantic segmentation model. In addition, a pixel-level geothermal spatial segmentation network (PGSSNet) is proposed to suppress the diffuse thermal radiation and reduce the broad and blurred white areas of images to exact locations. Once the training is completed, the model directly segments and extracts the actual range of thermal radiation sources from subsequent thermal infrared remote-sensing images without temperature retrieval and/or manual calibration.

A Web3D Rendering Optimization Algorithm for Pipeline BIM Models

BIM (building information modeling) plays a pivotal role in the construction industry. BIM technology tailored for pipelines offers in-depth semantic information and spatial data, bolstering the utility and implementation of digital twin-associated technologies in both architecture and urban planning. This paper introduces a rendering optimization algorithm rooted in the BSP Tree (Binary Space Partitioning Tree). The algorithm is used to address the challenges of slow loading and poor rendering quality of pipeline BIM models when displayed on the web, which stem from large amounts of model data and complex geometric configurations. Initially, the algorithm delves into the geometric distribution traits of the pipeline BIM model from multiple perspectives, pinpointing the spatial division dimension. Subsequently, it employs an adaptive step size technique for spatial segmentation, harmonizing it with real-world application contexts. Concurrently, any superfluous data that emerge are refined to uphold the structural wholeness of the BIM model. This algorithm is adept at systematically arranging and overseeing the BIM model data. Trial outcomes reveal that the AKDT (Adaptive K-Dimensional Tree) algorithm significantly trims the browser’s initial rendering duration while maintaining the model’s accuracy and semantic uniformity. Moreover, it excels in areas such as rendering frame rate, user interaction responsiveness, and data transmission duration. In essence, the algorithm stands out for its efficiency and precision in rendering pipeline BIM models on web platforms, achieving the desired optimization results.

Impact of Spatial Segmentation on the Assessment of Coastal Vulnerability—Insights and Practical Recommendations

Coastal areas are dynamic multidimensional systems challenged by the complex interactions between natural, environmental, and human-induced pressures, as well as the ever-changing climate. A comprehensive evaluation of their spatial and temporal features enables the development of effective practices required to apply integrated coastal zone management (ICZM) policies. ICZM seeks to address the vulnerability of coastal areas in an attempt to mitigate their weaknesses and increase their resilience. Hence, coastal vulnerability assessment is a prerequisite to proceed with optimal adaptation or upgrading actions. Currently, assessments are performed by considering different approaches related to dividing coastal areas into segments to observe the spatial variations of vulnerability. The present research seeks to investigate the impact of the spatial segmentation of coastal areas on the assessment of their vulnerability. To achieve this, a case study of the coastal zone of the Municipality of Thebes, located in the Northeastern Corinthian Gulf, Greece, is examined. Five segmentation approaches are applied in terms of a physical-based vulnerability assessment for two different time horizons, (a) the present and (b) the future, by incorporating the climate change impacts. This study allows for optimizing practices to estimate vulnerability parameters and obtain reliable results for practical applications while reducing time-consuming analyses.

Mitigation of spatial nonstationarity with vision transformers

Spatial nonstationarity, the location variance of features’ statistical distributions, is ubiquitous in many natural settings. For example, in geological reservoirs rock matrix porosity varies vertically due to geomechanical compaction trends, in mineral deposits grades vary due to sedimentation and concentration processes, in hydrology rainfall varies due to the atmosphere and topography interactions, and in metallurgy crystalline structures vary due to differential cooling. Conventional geostatistical modeling workflows rely on the assumption of stationarity to be able to model spatial features for geostatistical inference. Nevertheless, this is often not a realistic assumption when dealing with nonstationary spatial data and this has motivated a variety of nonstationary spatial modeling workflows such as trend and residual decomposition, cosimulation with secondary features, and spatial segmentation and independent modeling over stationary subdomains. The advent of deep learning technologies has enabled new workflows for modeling spatial relationships. However, there is a paucity of demonstrated best practice and general guidance on mitigation of spatial nonstationarity with deep learning in the geospatial context. We demonstrate the impact of two common types of geostatistical spatial nonstationarity on deep learning model prediction performance and propose the mitigation of such impacts using self-attention (vision transformer) models. We demonstrate the utility of vision transformers for the mitigation of nonstationarity with relative errors as low as 10%, exceeding the performance of alternative deep learning methods such as convolutional neural networks. We establish best practice by demonstrating the ability of self-attention networks for modeling large-scale spatial relationships in the presence of commonly observed geospatial nonstationarity.