Punktipilvede registreerimise meetodite mõju staatilise terrestrilise laserskaneerimise täpsusele
Laen...
Kuupäev
2016
Kättesaadav alates
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Abstrakt
Uurimustöö keskendub staatilisele terrestrilisele laserskaneeritud punktipilvede
andmetöötlusele. Probleemina oli tõstatatud küsimus - milline punktipilve registreerimise
meetod on täpsem. Käsitletud on kolme meetodit: skaneeritud punktipilve registreerimine
automaatselt ehk pindade järgi; tähistega punktipilve registreerimine tähiste tsentrite
automaatse tuvastamise järgi; tähistega punktipilve registreerimine tähiste tsentrite manuaalse
määramise järgi.
Antud bakalaureuse töö eesmärgiks oli uurida skaneeritud punktipilvede erinevate
registreerimiste mõju skaneerimise täpsusele. Uurimustöös vajalikud andmed saadi testala
skaneerides. Töö metoodikaks oli laserskaneeritud objekti skaneerimisel saadud punktipilve
modelleerimine programmis Trimble RealWorks, modelleeritud punktipilvest läbilõigete
tegemine ning programmis AutoCad jooniste tegemine.
Töö tulemused saadi modelleeritud punktipilve läbilõigetest saadud pikkuste ja
laserkaugusmõõturiga mõõdetud pikkuste võrdlemisel. Laserkaugusmõõturiga mõõdistatu
võeti õigeks mõõduks. Pindade põhjal registreerimise ja õige mõõdistuse vahe oli 3 mm.
Tähistega punktipilve tähiste automaatse tsentrite tuvastamise meetodi viga oli 2 mm.
Tähistega punktipilve tähiste manuaalse tsentrite määramise meetodi viga oli 6 mm.
Tulemusest järeldati, et antud keskkonnas oli tähistega punktipilve tähiste automaatsete
tsentrite tuvastamise meetod kõige täpsem.
Antud töös oli kasutatud vaid must-valgeid paberist tähiseid, kuid jätkuuuringuna võib
kasutada sfäärilisi tähiseid. Laserskaneeritud punktipilvede ülekattuvus oli 81%, kuid võib
kasutada töös mainitud minimaalset vajalikku ülekatet, milleks on 30 %.
The research focuses on the static terrestrial laser scanning point cloud data processing. The problem was raised by the question - which point clouds registration method is the most accurate. Three different techniques were addressed: scanned point cloud automatic registration also known as surface registration; target based registration – automatic identifying of target center; target based registration – manual target center assignment. This bachelor´s thesis objective was to investigate the impact of different point cloud registration methods on static terrestrial laser scanning accuracy. The necessary data in this research were obtained by scanning the test area. The methodology of this study was laser scanning an object and modelling the scanned point cloud in Trimble RealWorks program. After that cross-sections from modeled point clouds and AutoCad drawings were made. Results were obtained by comparing modeled point cloud cross-section measured lengths with laser rangefinder length. Laser rangefinder length was considered the correct length. Surface based registration and the correct length measuring difference was 3 mm. Target based - automatic identifying of target center error was 2 mm. Target based - manual target center assignment error was 6 mm. It was concluded that, in this environment, the target based - automatic identifying of target center is the most accurate. In this study black-and-white paper targets were used, but as a follow-up study spherical targets could be used. Laser scanning point clouds overlap was 81%, but the necessary minimum 30% overlap could be used.
The research focuses on the static terrestrial laser scanning point cloud data processing. The problem was raised by the question - which point clouds registration method is the most accurate. Three different techniques were addressed: scanned point cloud automatic registration also known as surface registration; target based registration – automatic identifying of target center; target based registration – manual target center assignment. This bachelor´s thesis objective was to investigate the impact of different point cloud registration methods on static terrestrial laser scanning accuracy. The necessary data in this research were obtained by scanning the test area. The methodology of this study was laser scanning an object and modelling the scanned point cloud in Trimble RealWorks program. After that cross-sections from modeled point clouds and AutoCad drawings were made. Results were obtained by comparing modeled point cloud cross-section measured lengths with laser rangefinder length. Laser rangefinder length was considered the correct length. Surface based registration and the correct length measuring difference was 3 mm. Target based - automatic identifying of target center error was 2 mm. Target based - manual target center assignment error was 6 mm. It was concluded that, in this environment, the target based - automatic identifying of target center is the most accurate. In this study black-and-white paper targets were used, but as a follow-up study spherical targets could be used. Laser scanning point clouds overlap was 81%, but the necessary minimum 30% overlap could be used.
Kirjeldus
Märksõnad
bakalaureusetööd, punktipilved, geodeesia, tähised, täpsus