Erinevate tarkvarade võrdlemine laserskaneerimise andmete registreerimisel
Laen...
Kuupäev
2023
Kättesaadav alates
09.09.2023
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Eesti Maaülikool
Abstrakt
Töö eesmärgiks on uurida, kuidas erineb samade andmete täpsus kasutades erinevaid
registreerimise tarkvarasid, põhinedes konkreetsete arendajate poolt välja töötatud
algoritmidel, mis võivad viia erinevate tulemusteni. Punktipilvede registreerimine on
esimene andmetöötluse etapp ning selle tulemusele tuginevad edaspidised joonised ja
mudelid. Seega on oluline tagada kvaliteetne tulemus enne järgmisi etappe.
Töös uuritakse automaatse ja manuaalse registreerimise kiirust ja täpsust. Kui mobiilse
laserskaneerimise ja fotogrammeetriliste toorandmete järeltöötlus on tihti
automatiseeritud, siis maapealse laserskaneerimise puhul on levinud ka manuaalne
registreerimine. Manuaalse ja automaatse registreerimise võrdlemiseks rajati 24
skaneerimisjaama ümber suurema hoone. Seda on tehtud eeldusega, et neid punktipilvi
saab registreerida ühtsesse süsteemi Töös käsitletakse erinevaid automaatregistreerimise
meetodeid ning võrreldakse nende tulemusi manuaalse registreerimisega. Üheks
suuremaks probleemiks laserskaneerimisel ongi andmete kontrollimine ja registreerimine.
Uurimistöö käigus selgitati välja, et tarkvarad suutsid registreerida enamus
skaneerimisjaamadest ühtesse rühma ka automaatselt. Manuaalsete meetodite tulemused
erinevad omavahel vähem võrreldes automaatsete meetoditega. Kõik tarkvarad püüavad
luua võimalikult palju seoseid skaneerimisjaamade vahel, et parandada registreerimise
lõpptulemust.
The aim of this thesis is to investigate the point cloud registration software. How the accuracy of the same data differs using different registration software, based on algorithms developed by company specific developers, which can lead to different results. The registration of point clouds is the first stage of data processing and subsequent drawings and models will be based on its result. It is therefore important to ensure a high-quality result before the next stages. The thesis examines the speed and accuracy of automatic and manual registration. While the post-processing of mobile laser scanning and photogrammetric raw data is often automated, manual registration is still commonly used in terrestrial laser scanning workflow. To compare manual and automatic registration, 24 scan setup around the building were done. This is done accordingly to assumption that all stations will be registered successfully. The thesis deals with different methods of automatic registration and compares its results with the results of manual registration. One of the major problems with laser scanning is data verification and registration. The tests revealed that the software were able to register most of the scan setups into one group automatically. The results of manual registration methods differ less from automatic methods. All software tries to create as many links between scan setups as possible in order to improve the final result of registration.
The aim of this thesis is to investigate the point cloud registration software. How the accuracy of the same data differs using different registration software, based on algorithms developed by company specific developers, which can lead to different results. The registration of point clouds is the first stage of data processing and subsequent drawings and models will be based on its result. It is therefore important to ensure a high-quality result before the next stages. The thesis examines the speed and accuracy of automatic and manual registration. While the post-processing of mobile laser scanning and photogrammetric raw data is often automated, manual registration is still commonly used in terrestrial laser scanning workflow. To compare manual and automatic registration, 24 scan setup around the building were done. This is done accordingly to assumption that all stations will be registered successfully. The thesis deals with different methods of automatic registration and compares its results with the results of manual registration. One of the major problems with laser scanning is data verification and registration. The tests revealed that the software were able to register most of the scan setups into one group automatically. The results of manual registration methods differ less from automatic methods. All software tries to create as many links between scan setups as possible in order to improve the final result of registration.
Kirjeldus
Magistritöö
Geodeesia, kinnisvara- ja maakorralduse õppekaval
Märksõnad
magistritööd, laserskaneerimine, registreerimine, punktipilv, analüüs, tarkvara, veahinnangud