Chen Chongcheng

An Algorithm for Spatial Outlier Detection Based on Delaunay Triangulation

Spatial outliers represent spatial objects whose non-spatial attributes are significantly different from their spatial neighborhoods. Spatial outlier detection can provide user with unexpected, interesting and useful spatial patterns for further analysis and has received a lot of attention. However, many existing methods for spatial outlier mining use the k-neighborhood method to determine spatial neighborhood which depends on a priori parameter k, and don?t consider spatial autocorrelation. As a result, it usually violates the true situation. So we propose a similarity measurement between spatial objects that based on Delaunay triangulation (DT_SOF), which captures spatial correlation and spatial neighborhood from the dataset itself. Furthermore, the measure takes in account the local behavior of a spatial object in its neighborhood. Finally, experimentations on a synthetic and a real-world ecological geochemical dataset demonstrate that our approach can effectively detect spatial outliers with a lower disturbance by human intervention.

Geometrical shapes and swaying movements of realistic tree: design and implementation

Trees and plants are important elements of outdoor scenes, therefore, it makes demands of research involved in computer graphics on methods to realize the natural-looking trees and plants in various environmental condition, such as windy or stormy weather, to contribute more facticity to the virtual reality. The paper starts from introducing the general background. Then, an approach adopting a series of statistical curves as inputted parameters of tree morphogenesis and motion depiction, which represent ramification modes, phyllotaxes or motion profiles of leaves and stems. Finally, some experimental models in virtual scene were showed. We focus on the faithful morphogenesis of tree, swaying motion in wind and the appropriate rendering speed. Aside, all the inputted statistical curves can be modified interactively to integrate the research work of botanists and other experts, which vary with species and districts. Within this methodology, various plant models can be created easily and rapidly without any botanic background, and faithful motion of tree in wind can be simulated easily with proper engineering accuracy and effective computation.Trees and plants are important elements of outdoor scenes, therefore, it makes demands of research involved in computer graphics on methods to realize the natural-looking trees and plants in various environmental condition, such as windy or stormy weather, to contribute more facticity to the virtual reality. The paper starts from introducing the general background. Then, an approach adopting a series of statistical curves as inputted parameters of tree morphogenesis and motion depiction, which represent ramification modes, phyllotaxes or motion profiles of leaves and stems. Finally, some experimental models in virtual scene were showed. We focus on the faithful morphogenesis of tree, swaying motion in wind and the appropriate rendering speed. Aside, all the inputted statistical curves can be modified interactively to integrate the research work of botanists and other experts, which vary with species and districts. Within this methodology, various plant models can be created easily and rapidly without any botanic background, and faithful motion of tree in wind can be simulated easily with proper engineering accuracy and effective computation.

Interactive Physical Based Animation of Tree Swaying in Wind

Lots of researches have been done on computer animation of tree swaying in wind. Some of them employed artificial simplification methods which could not achieve faithful swaying motion, others adopted complicated physical principles which would involve vast computation and fail to get an appropriate rendering speed. This paper presents a series of convenient interactive physical based methods for animation of trees swaying in wind. We focus on both the faithful swaying motion and the appropriate rendering speed. Therefore, a lot of statistical curves, representing wind profile loaded in tree or stem displacement profile and so on, are introduced for swaying motion calculation. All those curves can be modified interactively to get best match with the researches of botanists and other experts. These methods can be adjusted quickly and conveniently to simulate the animation of faithful tree swaying in wind with proper engineering accuracy and effective computation.

Research on grid based spatial outliers mining and its implementation

Integrated and deep-seated applications call for the intelligentization of spatial data processing. Spatial data mining is becoming more and more important to satisfy practical application and business intelligence. With the rapid development of distributed sharing infrastructures such as grid and web services, providing knowledge services on grid environment is becoming the trend of knowledge discovery and application technologies. This paper introduces an architecture of geographical knowledge grid platform based on globus (GeoKS-Grid), then puts forward a spatial data integration scheme by the combination of GML and WFS specification, finally, taking spatial outliers mining based on MST clustering as an example to illustrates the design and realization of knowledge service. Experimentation finds that it is feasible and practicable to extend the grid environment by adding more and more distributed high performance knowledge discovery services implemented based on data grid mechanisms and globus toolkit, to meet the requirements of application. For the application in real system, a task-oriented architecture based on KSDM is proposed and an implementation of knowledge service is introduced.

Comparative spatio-spectral heterogeneity analysis using multispectral and hyperspectral airborne images

Knowledge of spatio-spectral heterogeneity within multisensor remote sensing images across visible, near-infrared and short wave infrared spectra is important. Till now, little comparative research on spatio-spectral heterogeneity has been conducted on real multisensor images, especially on both multispectral and hyperspectral airborne images. In this study, four airborne images, Airborne Thematic Mapper, Compact Airborne Spectrographic Imager, Specim AISA Eagle and AISI Hawk hyperspectral airborne images of woodland and heath landscapes at Harwood, UK, were applied to quantify and evaluate the differences in spatial heterogeneity through semivariogram modelling. Results revealed that spatial heterogeneity of multisensor airborne images has a close relationship with spatial and spectral resolution and wavelength. Within the visible, near-infrared spectra and short wave infrared spectra, greater spatial heterogeneity is generally observed from the relatively longer wavelength in short wave infrared spectra. There are dramatic changes across the red and red edge spectra, and the peak value is generally examined in the red middle or red edge wavelength across the visible and near-infrared spectra for vegetation or non-vegetation landscape respectively. In all, for real multisensor airborne images, the change in spatial heterogeneity with spatial resolution will accord with the change of support theory depending on whether dramatic change exists across the corresponding wavelength. Besides, if with close spatial resolution, the spatial heterogeneity of multispectral images might be far from the overall integration of these bands from the hyperspectral images involved. A comparative assessment of spatio-spectral heterogeneity using real hyperspectral and multispectral airborne images provides practical guidance for designing the placement and width of a spectral band for different applications and also makes a contribution to the understanding of how to reconcile spatial patterns generated by multisensors.

PDDL based workflow auto-modeling for Spatial Data Mining

Spatial Data Mining (SDM) has grown a lot during the last years. More and more data mining operators are brought forward to the SDM domain. A SDM problem can be typically confronted by separating into several steps that could be organized into a workflow which is consisted of several core SDM operators and some assistant operators. However, the large numbers of operators makes it difficult to organize a tailored workflow by a manual workflow modeling. In this paper, an Artificial Intelligence (AI) Planning method is introduced into SDM workflow modeling, while the critical processes and elements for modeling a complete SDM workflow is abstracted and analyzed, and the corresponding elements are formally described by using Planning Domain Definition Language (PDDL). After the PDDL description, a JavaFF planner is used to drive the SDM workflow automodeling. A demonstration application of workflow automodeling for spatial outlier mining is proposed as well, to test the feasibility and availability of automated workflow construction achieved by the presented AI Planning method. The goal of this work is to present an attempt for AutoModeling of SDM workflow, and to help users out of previously cumbersome manual design of workflow and meanwhile to lower the threshold by AI Planning.

Terrain rendering technology based on vertex texture

According to the Vertex Texture Fetch technology in Shader Model 3.0, we designed a terrain rendering algorithm based on vertex texture. The algorithm storages the data of terrain as vertex texture, and get elevation value through the vertex shaders. During rendering phase a group of nested grids are built centeredly around current viewpoint. Elevation values are sampled from the vertex texture according to two-dimensional coordinates of the grids. Then the whole data are sent to GPU for rendering. The algorithm can reduce the CPU load and make full use of the powerful float computing and programmability feature.

Effect of spatial scale on regional land-use pattern analysis in different classification system and data sources

Study of spatial scale effect on regional land-use pattern analysis for construction of spatial model of regional land-use change entails consideration of regional land-use classification system selected and data source used. The effect characteristics of spatial scale on the results of regional land-use pattern analysis was investigated for two different classification systems and different data sources in terms of spatial autocorrelation by empirical study in the Fujian province of China. Fujian province and Longhai County, are selected representing two hierarchical management levels. At County level, two different data sources are used: the 1: 100,000 land use map interpreted from Landsat TM of 2000 and 1:10,000 land use surveying map of 2000. Results showed that all land uses in our study area shows strong positive spatial autocorrelation and VMI of most land uses are scale-dependent. VMI of most land uses decrease with increasing grain size except in Fujian province and VMI of regional land use from more coarse scale land use map decreases more sharply with increasing grain size. Characteristics of fractal dimension are similar to that of spatial autocorrelation in both Counties to a certain extent. When two or three land uses are combined into one, the fractal dimension of it is comparatively larger than the original ones but No general laws in terms of VMI were seen. There may be more variation of VMI of land uses in more detailed classification system. Reducing number of land-use for one classification system may diminish the loss of information of land use area across the range of scale and the effect of spatial scale on regional land-use analysis in certain extent.

Map-based BitSet association rule mining of remote sensing images

This paper presents the MBSA algorithm, which uses TreeMap class and a compressed BitSet class in Java. Boolean values are stored in the Bitset. MBSA algorithm scans the transaction database once and further database scans can be replaced by Bitset logical AND operation, which efficiently improves the computation. The MBSA algorithm was applied to crop remote sensing images for association rules mining. Interesting association rules discovered from remote sensing images composed of red, green and blue bands and crop yield is useful for improving crop production.