Optimizing Spatial Relationships in GCN to Improve the Classification Accuracy of Remote Sensing Images

Semantic segmentation of remote sensing images is one of the core tasks of remote sensing image interpretation.With the continuous develop-ment of artificial intelligence technology,the use of deep learning methods for interpreting remote-sensing images has matured.Existing neural networks disregard the spatial relationship between two targets in remote sensing images.Semantic segmentation models that combine convolutional neural networks(CNNs)and graph convolutional neural networks(GCNs)cause a lack of feature boundaries,which leads to the unsatisfactory segmentation of various target feature boundaries.In this paper,we propose a new semantic segmentation model for remote sensing images(called DGCN hereinafter),which combines deep semantic segmentation networks(DSSN)and GCNs.In the GCN module,a loss function for boundary information is employed to optimize the learning of spatial relationship features between the target features and their relationships.A hierarchical fusion method is utilized for feature fusion and classification to optimize the spatial relationship informa-tion in the original feature information.Extensive experiments on ISPRS 2D and DeepGlobe semantic segmentation datasets show that compared with the existing semantic segmentation models of remote sensing images,the DGCN significantly optimizes the segmentation effect of feature boundaries,effectively reduces the noise in the segmentation results and improves the segmentation accuracy,which demonstrates the advancements of our model.

Image Retrieval Based on Objects’ Spatial Orientation Relationships

A new concept of characteristic scanning radial (CSR) is proposed for thesegmented image on the basis of two shape-specific points of its shape-objects. Subsequently, twocharacteristic attribute sequences (CAS) of relative distance and relative direction are derived torepresent the spatial orientation relationships among objects of the image. A novel image retrievalalgorithm is presented using these two CASs. The proposed retrieval approach not only satisfies thetransformational invariance, butalso attains the quantitative comparison of matching. Experimentsidentify the effectiveness and efficiency of the algorithm adequately.

Spatial coupling relationships of gas hydrate formation in the Tibetan Plateau

At present, gas hydrates are known to occur in continental high latitude permafrost regions and deep sea sediments. For middle latitude permafrost regions of the Tibetan Plateau, further research is required to ascertain its potential development of gas hydrates. This paper reviewed pertinent literature on gas hydrates in the Tibetan Plateau. Both geological and ge- ographical data are synthesized to reveal the relationship between gas hydrate formation and petroleum geological evo- lution, Plateau uplift, formation of permafrost, and glacial processes. Previous studies indicate that numerous residual basins in the Plateau have been formed by original sedimentary basins accompanied by rapid uplift of the Plateau. Ex- tensive marine Mesozoic hydrocarbon source rocks in these basins could provide rich sources of materials forming gas hydrates in permafrost. Primary hydrocarbon-generating period in the Plateau is from late Jurassic to early Cretaceous, while secondary hydrocarbon generation, regionally or locally, occurs mainly in the Paleogene. Before rapid uplift of the Plateau, oil-gas reservoirs were continuously destroyed and assembled to form new reservoirs due to structural and thermal dynamics, forcing hydrocarbon migration. Since 3.4 Ma B.P., the Plateau has undergone strong uplift and extensive gla- ciation, periglacier processes prevailed, hydrocarbon gas again migrated, and free gas beneath ice sheets within sedi- mentary materials interacted with water, generating gas hydrates which were finally preserved under a cap formed by frozen layers through rapid cooling in the Plateau. Taken as a whole, it can be safely concluded that there is great temporal and spatial coupling relationships between evolution of the Tibetan Plateau and generation of gas hydrates.

Functional Brain Network Learning Based on Spatial Similarity for Brain Disorders Identification

Functional brain network (FBN) measures based on functional magnetic resonance imaging (fMRI) data, has become important biomarkers for early diagnosis and prediction of clinical outcomes in neurological diseases, such as Alzheimer’s diseases (AD) and its prodromal state (i.e., Mild cognitive impairment, MCI). In the past decades, researchers have developed numbers of approaches for FBN estimation, including Pearson’s correction (PC), sparse representation (SR), and so on. Despite their popularity and wide applications in current studies, most of the approaches for FBN estimation only consider the dependency between the measured blood oxygen level dependent (BOLD) time series, but ignore the spatial relationships between pairs of brain regions. In practice, the strength of functional connection between brain regions will decrease as their distance increases. Inspired by this, we proposed a new approach for FBN estimation based on the assumption that the closer brain regions tend to share stronger relationships or similarities. To verify the effectiveness of the proposed method, we conduct experiments on a public dataset to identify the patients with MCIs from health controls (HCs) using the estimated FBNs. Experimental results demonstrate that the proposed approach yields statistically significant improvement in seven performance metrics over using the baseline methods.

The new multistage water adsorption model of Longmaxi Formation shale considering the spatial configuration relationship between organic matter and clay minerals

The water adsorption by shale significantly affects shale gas content and its seepage capacity.However,the mechanism of water adsorption by shale is still unclear due to its strong heterogeneity and complicated pore structure.The relationship between the adsorbed water content at different relative humidities(RHs)and shale compositions,as well as shale pore structure and the spatial configuration relationship between organic matter(OM)and clay minerals,was investigated to clarify the controlling factors and mechanisms of water adsorption by Longmaxi Formation shale from the Southern Sichuan Basin in China.Consequently,the water adsorption process could be generally divided into three different stages from 0%RH to 99%RH.Furthermore,the Johnston’s clay mine ral interlayer pore structure model(JCM),the Freundlich model(FM)and the Dubinin-Astakhov model(DAM)were tested to fit the three water adsorption stages from low RH to high RH,respectively.The fitting results of the JCM and FM at lower RHs were far from good,while the fitting results of DAM at higher RHs were acceptable.Accordingly,two revised models(LRHM and MRHM)considering the spatial configuration relationship between OM and clay minerals were proposed for the two stages with lower RHs,and performed better fitting results indicating the pronounced effect of the spatial configuration relationship between OM and clay minerals on the water adsorption process of Longmaxi Formation shale.The outcomes of this study will contribute to clarifying the water distribution characteristics in the pore network of shale samples with variable water contents.

An Effective Image Retrieval Mechanism Using Family-based Spatial Consistency Filtration with Object Region

How to construct an appropriate spatial consistent measurement is the key to improving image retrieval performance. To address this problem, this paper introduces a novel image retrieval mechanism based on the family filtration in object region. First, we supply an object region by selecting a rectangle in a query image such that system returns a ranked list of images that contain the same object, retrieved from the corpus based on 100 images, as a result of the first rank. To further improve retrieval performance, we add an efficient spatial consistency stage, which is named family-based spatial consistency filtration, to re-rank the results returned by the first rank. We elaborate the performance of the retrieval system by some experiments on the dataset selected from the key frames of ＂TREC Video Retrieval Evaluation 2005 （TRECVID2005）＂. The results of experiments show that the retrieval mechanism proposed by us has vast major effect on the retrieval quality. The paper also verifies the stability of the retrieval mechanism by increasing the number of images from 100 to 2000 and realizes generalized retrieval with the object outside the dataset.

Spatial Mismatch or Not?Evidence from Public Janitors in Xi’an,China

Research on the spatial mismatch experienced by low-income minority residents is US-centric.However,spatial mismatch is not necessarily an appropriate term when considering the situation of low-wage workers in cities of northwestern China where there is higher proximity between jobs and housing and lower levels of residential segregation.This paper empirically examines the jobs-housing spatial relationship for one of the most typical low-wage groups,namely,public janitors,in Xi’an,China.Also,the causes of the jobs-housing spatial relationship are discussed in detail.Individual-level data based on in-depth interviews and questionnaires,as well as the GIS network analysis method,are used to provide baseline analyses of the jobs-housing spatial relationship.Results indicate that there is no jobs-housing spatial mismatch for public janitors in Xi’an.This can be implied from the short commuting distance and time.A basic cause is that most public janitors rent low-cost accommodation in villages-in-the-city,and in old residential quarters,near to their places of work.Other causes lie in off-peak commuting and high sensitivity to commuting distance due to the greater extent of nonmotorized commuting modes.The conclusions,based on a large number of social surveys,are an illuminating analysis of the spatial mismatch issue among low-wage workers in Chinese cities.

A LSA Based Image Classification Framework Utilizing Relative Spatial Arrangement

This paper focuses on the problem of automatic image classification （AIC） by proposing a framework based on latent semantic analysis （LSA） and image region pairs. The novel framework employs relative spatial arrangements for region pairs as the primary feature to capture semantics. The significance of this paper is twofold. Firstly, to the best our knowledge, this is the first study of the influence of region pairs as well as their relative spatial information in latent semantic analysis as applied to automatic image classification. Secondly, our proposed method for using the relative spatial information of region pairs show great promise in improving image semantic classi- fication compared with the classical latent semantic analysis method and 2D string representation algorithm.

PHYLETIC RELATIONSHIP OF PROTEIN STRUCTURES BASED ON SPATIAL PREFERENCE OF RESIDUES

A distance measure that infers to indicate the evolutionary relationship of protein structures has been developed based on spatial preference factors of residues. The spatial preference factor is a reflection of the environment of residues in tertiary structure. Compared with the phyletic relationships derived from sequence homologies and three-dimensional structures, we find that the two lines of evolution are similar in general. This approach is applied to a group of glins here.

MVCE-Net: Multi-View Region Feature and Caption Enhancement Co-Attention Network for Visual Question Answering

Visual question answering(VQA)requires a deep understanding of images and their corresponding textual questions to answer questions about images more accurately.However,existing models tend to ignore the implicit knowledge in the images and focus only on the visual information in the images,which limits the understanding depth of the image content.The images contain more than just visual objects,some images contain textual information about the scene,and slightly more complex images contain relationships between individual visual objects.Firstly,this paper proposes a model using image description for feature enhancement.This model encodes images and their descriptions separately based on the question-guided coattention mechanism.This mechanism increases the feature representation of the model,enhancing the model’s ability for reasoning.In addition,this paper improves the bottom-up attention model by obtaining two image region features.After obtaining the two visual features and the spatial position information corresponding to each feature,concatenating the two features as the final image feature can better represent an image.Finally,the obtained spatial position information is processed to enable the model to perceive the size and relative position of each object in the image.Our best single model delivers a 74.16%overall accuracy on the VQA 2.0 dataset,our model even outperforms some multi-modal pre-training models with fewer images and a shorter time.

Content-based retrieval based on binary vectors for 2-D medical images

In medical research and clinical diagnosis, automated or computer-assisted classification and retrieval methods are highly desirable to offset the high cost of manual classification and manipulation by medical experts. To facilitate the decision-making in the health-care and the related areas, in this paper, a two-step content-based medical image retrieval algorithm is proposed. Firstly, in the preprocessing step, the image segmentation is performed to distinguish image objects, and on the basis of the ...

Narratives in Mamluk architecture： Spatial and perceptual analyses of the madrassas and their mausoleums

Mamluk sultans were known for their patronage of the arts and architecture. Their educational institutions were among the wide array of architectural projects that linked them as ruling elites to the religious scholars of their times. Their tombs were placed in a mausoleum attached to their educational-religious complexes to attest to their legacy. The evolution of their buildings such that both educational and memorial functions are integrated with the dense surroundings is scrutinized through chronological-spatial analysis. The configuration of the built form, the disposition of its boundaries, its patterns of accessibility, and its visual properties are the features that present the buildings to one＇s experience and bring certain perceptions into play. In this study, various spatial descriptor tools of space syntax are employed to analyze the data of 14 Mamluk examples （1260-1517A.D.） and capture the differences in the experience where the expression is preserved. Analyses of the configurational characteristics, axial attributes, visibility structures, and isovists highlight how the spatial and formal properties of the layouts were used to express certain representational relationships. The advantages of combining different spatial investigations allows for understanding historical design principles and how the geometry of forms could hide in its abstract rules, conceptual and perceptual qualities.

Spatial relationship-assisted classification from high-resolution remote sensing imagery

Spatial information remains to be an important topic in geographic information system and in remote sensing fields,and spatial relationships have been increasingly incorporated into the image classification processes.Previous studies have employed multiple occurrences of spatial features(shape,texture,etc.,)to improve classification results.However,less attention has been focused on using higher-level spatial relationships for image classification.In this study,two novel spatial relationships,namely,maximum spatial adjacency(MSA)and directional spatial adjacency(DSA),were proposed to assist in image classification.The proposed methods were implemented to extract buildings,beach,and emergent vegetation land-cover classes according to their spatial relationships with their corresponding reference classes.The promising results obtained from this study suggest that the proposed MSA and DSA spatial relationships can be valuable information in defining rule sets for a more reasonable and accurate classification.

Development of Suburban New Towns in Shanghai: Jobs-Housing Spatial Relationship Analysis

This paper identifies the employment and housing locations of residents in Shanghai based on mobile phone signaling data, so as to obtain the employment density and commuting data and analyze the development of nine suburban new towns from the perspective of jobs-housing spatial relationship. Firstly, the paper defines employment-intensive areas and gets the average employment density of each new town according to the employment density data. Then it marks out the scope of the employment influence through analyzing the sources of workers in each new town in accordance with the commuting data. Finally, it analyzes the jobs-housing balance of each new town using independence index, finding that suburban new towns in Shanghai have become main clusters of economic activities, while the scope of employment influence in each new town is still concentrated in its administrative area, with less attraction to residents in other areas. The independence index demonstrates a law that the suburban new town which is farther from the central city sees a higher degree of jobs-housing balance. Among them, new towns located in the outer suburbs with a low independence index indicate their special development situation, the reason of which is worth further study.

Comparison between CBR and CA methods for estimating land use change in Dongguan, China

Many studies on land use change （LUC）, using different approaches and models, have yielded good results. Applications of these methods have revealed both advantages and limitations. However, LUC is a complex problem due to influences of many factors, and variations in policy and natural conditions. Hence, the characteristics and regional suitability of different methods require further research, and comparison of typical approaches is required. Since the late 1980s, CA has been used to simulate urban growth, urban sprawl and land use evolution successfully. Nowadays it is very popular in resolving the LUC estimating problem. Casebased reasoning （CBR）, as an artificial intelligence technology, has also been employed to study LUC by some researchers since the 2000s. More and more researchers used the CBR method in the study of LUC. The CA approach is a mathematical system constructed from many typical simple components, which together are capable of simulating complex behavior, while CBR is a problem-oriented analysis method to solve geographic problems, particularly when the driving mechanisms of geographic processes are not yet understood fully. These two methods were completely different in the LUC research. Thus, in this paper, based on the enhanced CBR model, which is proposed in our previous research （Duet aL 2009）, a comparison between the CBR and CA approaches to assessing LUC is presented. LUC in Dongguan coastal region, China is investigated. Applications of the improved CBR and the cellular automata （CA） to the study area, produce results demonstrating a similarity estimation accuracy of 89% from the improved CBR, and 70.7% accuracy from the CA. From the results, we can see that the accuracies of the CA and CBR approaches are both 〉70%. Although CA method has the distinct advantage in predicting the urban type, CBR method has the obvious tendency in predicting non-urban type. Considering the entire analytical process, the preprocessing workload in CBR is less than that of the CA approach. As such, it could be concluded that the CBR approach is more flexible and practically useful than the CA approach for estimating land use change.

Item Ownership Relationship Semantic Learning Strategy for Personalized Service Robot

In order to satisfy the robotic personalized service requirements that can select exclusive items to perform inference and planning according to different service individuals,the service robots need to have the ability to independently obtain the ownership relationship between humans and their carrying items.In this work,we present a novel semantic learning strategy for item ownership.Firstly,a human-carrying-items detection network based on human posture estimation and object detection model is used.Then,the transferred convolutional neural network is used to extract the characteristics of the objects and the back-end classifier to recognize the object instance.At the same time,the face detection and recognition model are used to identify the service individual.Finally,on the basis of the former two,the active learning of ownership items is completed.The experimental results show that the proposed ownership semantic learning strategy can determine the ownership relationship of private goods accurately and efficiently.The solution of this problem can improve the intelligence level of robot life service.

The quantitative classification of granites and their metallogenetic relations in West Qinling,Gansu Province,China

The quantitative classification of granite and their metallogenetic relations have never been discussed.The Q-system clustering analysis and discriminant analysis methods were alternately used to quantitatively analyze the 11 oxide data in granite samples from the West Qinling area of Gansu Province,and then to construct the quantitative classification series models of granite(oxide).The granites samples are divided into three categories and eight subcategories.The classification of granites is biased toward prospecting.According to the spatial correlation between eight types of granites and copper deposits,lead and zinc deposits,gold deposits,etc.(within 3 km of the intrusion)in the West Qinling area in Gansu Province,the“metallogenic related intrusions”are sought,and the prospecting target areas are defined.Furthermore,they provide reliable basis for regional geological prospecting.

血缘共同体规定下的传统手工业村落适应性变迁——山西省新绛县西庄石匠村个案研究

As an important witness to the rural industry in the agricultural era,traditional handicraft villages have attracted renewed attention in the contemporary context of intangible cultural heritage protection.The dual characteristics of their traditional cultural heritage and modern economic benefits make these villages an important carrier of the dynamic inheritance and development of the heritages under the background of rural revitalization.This paper takes a stonemason village of Xizhuang in Xinjiang County,Shanxi Province,as an example,and reveals the evolutionary characteristics of traditional handicraft villages in the process of transition from agricultural to industrial society from the“social-spatial”perspective.In the agricultural era,the kinship,production organization,and religious beliefs of traditional handicraft villages were closely linked under the blood relationship in a clan-based structure,forming a“social-spatial”trinity composed of“residence,production,and spirit.”In the industrial era,the development of handicraft economy results in the transformation of the“social-spatial”form of the settlement,the social structure shrank from clan control to nuclear family domination,the production space began to be stripped away from villages,the living space changed from gathered clan living to separated-family homes,and the spiritual place was transformed into public space.However,due to the inter-generational transmission of handicrafts based on blood relationship,the consanguinity community has always been the main carrier for the continuity of skills,and has become the stable foundation for maintaining the“social-spatial”form of the village.Therefore,the adaptive transition of traditional handicraft villages in the context of socio-economic transformation continues the interconnection among society,economy,and space,and presents a unique vitality and comparative stability.