22
1 View- and Scale-Based Progressive Transmission of Vector Data Padraig Corcoran, Peter Mooney, Adam Winstanley and Michela Bertolotto. Department of Computer Science, National University of Ireland Maynooth. School of Computer Science and Informatics, University College Dublin

View - and Scale-Based Progressive Transmission of Vector Data

Embed Size (px)

DESCRIPTION

View - and Scale-Based Progressive Transmission of Vector Data Padraig Corcoran, Adam Winstanley, Peter Mooney - National University of Ireland Maynooth Michela Bertolotto - University College Dublin

Citation preview

Page 1: View - and Scale-Based Progressive Transmission of Vector Data

1

View- and Scale-Based Progressive Transmission of Vector Data

Padraig Corcoran, Peter Mooney, Adam Winstanley and Michela Bertolotto.

Department of Computer Science,National University of Ireland Maynooth.

School of Computer Science and Informatics,University College Dublin

Page 2: View - and Scale-Based Progressive Transmission of Vector Data

2

Introduction

● Web application development is in the middle of a paradigm shift.

● Web-GIS applications still linger behind desktop-GIS in terms of:● Functionality.● Interface.● User Interaction.

● This can be attributed to the manner in which spatial data is transmitted.

Page 3: View - and Scale-Based Progressive Transmission of Vector Data

3

Tile-based Transmission

● Predominant transmission methodology● Vector data converted to raster maps tiles on the

server.● Map tiles transmitted to client.● Used by Google Maps and OpenStreetMap.

Page 4: View - and Scale-Based Progressive Transmission of Vector Data

4

● Advantages:● HTML has native support for images.● Image compression is an advanced science.● All data requests are pre-computed.

Page 5: View - and Scale-Based Progressive Transmission of Vector Data

5

● Disadvantage:● Vector data is not transmitted therefore the client

cannot perform spatial queries or adapt the visualization.

Page 6: View - and Scale-Based Progressive Transmission of Vector Data

6

Vector-Based Transmission

● Can we transmit vector data and maintain the advantages of tile-based transmission?

● Development of such technology is a main goal in the field of Progressive Transmission.

Page 7: View - and Scale-Based Progressive Transmission of Vector Data

7

Progressive Transmission

● For large data sets a trade off exists between:● Transmission of high levels of detail.● Transmission in reasonable time.

● Progressive transmission attempts to optimize this trade-off for vector data.

Page 8: View - and Scale-Based Progressive Transmission of Vector Data

8

● Progressive transmission is characterized by two properties:● Data is transmitted in the form of increments or

refinements.● To reduce redundancy data is not re-transmitted.

Page 9: View - and Scale-Based Progressive Transmission of Vector Data

9

View- and Scale Based Transmission

● In order to structure existing research in this field we propose a classification.

● All methods for progressive transmission may be classified as view- or scale-based.

Page 10: View - and Scale-Based Progressive Transmission of Vector Data

10

View-Based Transmission

● Data is transmitted progressively as a function of changing viewing window.

Time (Progressively Changing View)

Page 11: View - and Scale-Based Progressive Transmission of Vector Data

11

Scale-Based Transmission

● Data is transmitted progressively as a function of changing scale.

Time (Progressively Changing Scale)

Page 12: View - and Scale-Based Progressive Transmission of Vector Data

12

Scale-Based Implementation

● Refinement is the inverse of generalization.● All refinements are actually generalizations and

therefore satisfy the same objectives.

Page 13: View - and Scale-Based Progressive Transmission of Vector Data

13

Fusion View- and Scale-Based

● Both approaches reduce the volume of data transmitted in different ways.

● To maximise efficiency concepts from both must be fused.

● Currently the most advanced fusion method is that of Li et al. 2009

Page 14: View - and Scale-Based Progressive Transmission of Vector Data

14

Li et al. Methodology

● The vector data is divided into tiles.● The subset of tiles a user views is determined.● Each of these tiles is then transmitted using a

scale based transmission strategy.

Page 15: View - and Scale-Based Progressive Transmission of Vector Data

15

● Disadvantages:● Features which span multiple tiles must be

segmented and rejoined.● Such features cannot be generalized.

Page 16: View - and Scale-Based Progressive Transmission of Vector Data

16

Proposed Fusion Methodology

● A transmission method which removes the requirement for tiles is proposed.

● Firstly all features are generalized in a manner which maintains topology (Corcoran et. al, IJGIS 2011).

Page 17: View - and Scale-Based Progressive Transmission of Vector Data

17

● Features are then inserted into an R-tree (spatial indexing method).

● Given a viewing window the features contained within this window are progressively transmitted while maintaining topology (Corcoran. et al, Agile 2011).

Page 18: View - and Scale-Based Progressive Transmission of Vector Data

18

Implementation

● Implemented using client server model.● Server client communication uses HTML 5

WebSocket API.● Client rendering uses HTML 5 Canvas API.

Sequence Diagram

Page 19: View - and Scale-Based Progressive Transmission of Vector Data

19

Transmission Example

Large Scale Map

Page 20: View - and Scale-Based Progressive Transmission of Vector Data

20

Page 21: View - and Scale-Based Progressive Transmission of Vector Data

21

Quantitative Results

Comparison of data volume transmitted.

Page 22: View - and Scale-Based Progressive Transmission of Vector Data

22

Conclusions

● We provide an analysis and propose a framework to classify existing progressive transmission methods.

● Subsequently, a new fusion method is proposed.

● Request are computed on the fly; future work will aim to reduce computational complexity.