Kiire kaardiliides mahukate vektorandmetega
Date
2013
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Tartu Ülikool
Abstract
Tänapäeva maailmas esineb sageli vajadus kujundada ja arendada igasuguseid veebipõhiseid geograafilisi infosüsteeme (GIS), praktilistel ja teaduslikel eesmärkidel. Tihti juhtub, et geograafiliste andmete maht, mida infosüsteem peab samaaegselt kasutajale näitama ja edastama on nii suur, et veebilehitsejatel võivad tekkida probleemid andmete visualiseerimisega ja kiire esitamisega.
Peamine töö eesmärk on uurida kuidas olemasolevad geoandmete visualiseerimise vahendid toetavad suuri geograafilisi andmekogumeid (rohkem kui 100000 objekte), nende kodeeringut ja kujutamist kaardil. Tähelepanu pööratakse ülevaatele, kus kirjeldatakse kõige sagedamini kasutatavad meetodit ja tehnoloogiad, mis töötavad väga suuremahuliste ja mitmedimensionaalsete andmetega ja võimaldavad määrata kuvatavate andmete hulka erinevatel suumiastmetel. Töös võrreldakse visualiseerijaid (SVG, canvas), serveri- ja kliendipoolseid klasterdamise tehnikaid ning detsimeerimise printsiipi. Lisaks uuritakse geoandmete visualiseerimise kasutusmugavust. Väga suuremahuliste andmete kuvamisel ei ole kasutajale tihtipeale arusaadav, mida on kaardile visualiseeritud.
Lõpptulemusena luuakse väike GIS veebirakendus, kus rakendatakse mõningaid optimeerimis- ja visualiseerimistehnikad, et näha kuidas veebilehitsejad saavad hakkama väga suuremahuliste andmemahtudega. Rakenduse testimisel saadud tulemused korrastati ja analüüsiti, et hinnata valitud meetodite töökindlust ja jõudlust.
The main purpose of the current paper was to study which features could provide benefits for fast representation and generalization of large volumes of spatial data. Work concentrates on vector data type and provides the overview of the most common technologies used to speed-up the process of rendering objects on the map. The main emphasis was on the practical implementation of several optimization methods, namely, clustering, canvas renderer, and decimation principle. Map readability and geovisualization usability issues were also discussed. Prototype created as a practical part of the thesis was used as the input for measuring the effectiveness of a particular technology, where the time needed to render vector objects as well as map readability were considered as main criteria. The results were computed for various cases, using different number of objects. The data obtained during the experiment indicate that none of the examined approaches allows to solve alone the performance problem - each optimization method has its advantages and disadvantages. If the goal of the GIS web application is to provide an efficient vector web mapping in satisfying time and allow manipulation with the data, then several optimization and rendering approaches should be used together.
The main purpose of the current paper was to study which features could provide benefits for fast representation and generalization of large volumes of spatial data. Work concentrates on vector data type and provides the overview of the most common technologies used to speed-up the process of rendering objects on the map. The main emphasis was on the practical implementation of several optimization methods, namely, clustering, canvas renderer, and decimation principle. Map readability and geovisualization usability issues were also discussed. Prototype created as a practical part of the thesis was used as the input for measuring the effectiveness of a particular technology, where the time needed to render vector objects as well as map readability were considered as main criteria. The results were computed for various cases, using different number of objects. The data obtained during the experiment indicate that none of the examined approaches allows to solve alone the performance problem - each optimization method has its advantages and disadvantages. If the goal of the GIS web application is to provide an efficient vector web mapping in satisfying time and allow manipulation with the data, then several optimization and rendering approaches should be used together.