1 / 23

Warsaw University of Technology

Warsaw University of Technology. APPLICATION OF COMPUTATIONAL INTELLIGENCE ALGORITHMS IN TOPOLOGY PRESERVING PROCESS OF DTM SIMPLIFICATION. Robert Olszewski. A ssumption data.

whitby
Download Presentation

Warsaw University of Technology

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Warsaw University of Technology • APPLICATION OF COMPUTATIONAL INTELLIGENCE ALGORITHMSIN TOPOLOGY PRESERVINGPROCESS OF DTM SIMPLIFICATION Robert Olszewski

  2. Assumption data • Generalisation of the digital terrain model is an important issue for supplying geographic information systems with data, • The main idea of generalisation of the DTM should be the preservation of its structure (the morphological skeleton), • Simple algorithms of the DTM generalisationallow for relatively low reduction of the structure complexity

  3. The aim of the research • Development of the concept of the multiscale (hierarchical) representation of the terrain relief, • The concept of multirepresentation digital terrain model is a logical supplement of the idea of multirepresentation (MRDB) topographic database and allows to perform common analyses of all topographic components. Hierarchical DTM with monoscales representations of the model at an arbitrary, user-defined level of generalisation

  4. Spatial data generalisation • Distinction: • model generalisation (analysis-oriented), • cartographic generalisation (display-oriented) • Distinction: • DLM (to supply geographic information systems), • DCM (tosupply maps production systems)

  5. Digital Terrain Model - DTM • Generalisation of the DTM is basedon one of the methods (Weibel, 1992): • global filtration, • local filtration (usually multi-stage), • heuristic approach. Generalisation of the DTM (TIN)is understood as model generalisation and not as generalisation of contour lines

  6. The idea of DTM generalisation • combination of two approaches (local weighted filtration & structure lines extraction), • multi iteration approach, • determination of the „skeleton” of the terrain, • dichotomicclassification of source data(mass points vs. structural points), • differential weighting for mass and structural points, • multiscale (hierarchical) TIN model(with monoscale representations), • topology preservation...

  7. The idea of DTM generalisation • combination of two approaches (local weighted filtration & structure lines extraction), • multi iteration approach, • determination of the „skeleton” of the terrain, • dichotomicclassification of source data (mass points vs. structural points), • differential weighting for mass and structural points, • multiscale (hierarchical) TIN model (with monoscale representations), • topology preservation...

  8. Tatra Mountains

  9. The idea of DTM generalisation • combination of two approaches (local weighted filtration & structure lines extraction), • multi iteration approach, • determination of the „skeleton” of the terrain, • dichotomicclassification of source data(mass points vs. structural points), • differential weighting for mass and structural points, • multiscale (hierarchical) TIN model building (with monoscale representations), • topology preservation...

  10. Topology preservation

  11. Spatial data mining and model generalisation • Nowadays, the algorithmic approach may be considered as the dominating tendency in the field of spatial data generalisation, but… • Results of utilisation of computational intelligence and cognitive modelling are also very promising ... • On the contrary to classical expert systems, well known since the eighties of the 20th century, which utilise IF-THEN deterministic rules, the essence of this approach is connected with the use of machine learning (ML) processes (Meng, 1998).

  12. FUZZY • NEURO Inference systems Inference„engines”: • CRISP

  13. The idea of DTM generalisation • combination of two approaches (local weighted filtration & structure lines extraction), • multi iteration approach, • determination of the „skeleton” of the terrain, • dichotomicclassification of source data (mass points vs. structural points), • differential weighting for mass and structural points, • multiscale (hierarchical) TIN model (with monoscale representations), • topology preservation...

  14. Weighted local filtration • In the process of generalisation points are eliminated basing on local evaluation of several criteria: • vertical significance(mass points & structural points), • horizontal significance (density) (mass points & structural points), • the weight of a structural line(structural points only), • the local sinusoity of a structural line(structural points only). Selection of significance of particular factors is fully parameterised, what allows arbitrary assigning of weighting coefficients.

  15. DTM generalisation TIN generalisation

  16. Implementation 2D (MapInfo)

  17. 3D (ArcGIS)

  18. Hierarchical model

  19. 1st level 2nd level 3rd level Levels of TIN topology preservation

  20. Inference engines • Engines already implemented: • CRISP, • FUZZY, • NEURO • Engines to be implemented: • classification and regression trees, • boosted trees, • „random forest”, • MARS (Multivariate Adaptive Regression Splines), • SVM (Support Vector Machines)

  21. Conclusions • The basic feature of generalisation of the terrain model should be the preservation of its structure (the morphological skeleton) - topology preservation, • Utilisation of local weighted filtration algorithms allowfor : • representative selection of points from the source model, • the construction of the multiscale (hierarchical) TIN model with a monoscale representation at an arbitrary, user-defined level of generalisation, • topology preservation..

More Related