High Resolution Stereo Vision Using Compressive Sensing System

Autor: Dlab, Nicol
Přispěvatelé: Seršić, Damir
Jazyk: chorvatština
Rok vydání: 2019
Předmět:
Popis: U radu je opisan postupak rekonstruckije dvodimenzionalne scene bazirane na principu sažimajuceg očitavanja koristeći visokorezolucijski projektor i monokromatsku kameru. Nakon što se rekonstrukcija pokazala zadovoljavajucom i potvrđeni su principi CS rekonstrukcije, prešlo se na 3D scenu gdje se koristio visokorezolucijski projektor i dvije monokromatske kamere za akviziciju stereo vida. Zbog problema neplanarne scene u slucaju 3D scene s objektima postavljenim na različitim dubinama, korišteni su principi sažimajuce dualne fotografije i transformacije fundamentalnom matricom kako bi se rekonstruirao visoko rezolucijski stereo vid. Iz rekonstruiranog stereo vida visoke rezolucije je napravljen 3D model snimljene scene. U rekonstukciji su korišteni optimizacijski algoritmi i programsko okruženje MATLAB. This thesis describes a 2D reconstruction setup consisting of a high resolution projector and a monochromatic camera based on compressive sensing. After the reconstruction proved to be satisfactory and the CS reconstruction principles were confirmed, 3D reconstruction was next using a high resolution projector and two monochromatic cameras to acquire stereo vision. Due to the problem of non-planar scene in case of 3D scene with objects set at different depths, principles of dual photography and fundamental matrix transforms were used. From the reconstructed high resolution stereo vision, a 3D model of the scene was made. In the reconstruction the optimization algorithms and the programming environment MATLAB were used.
Databáze: OpenAIRE