Monocular vision-based position detection of qr marker using numerical computation

Teoh, Mooi Khee (2023) Monocular vision-based position detection of qr marker using numerical computation. Masters thesis, Universiti Malaysia Sabah.

	Text 24 PAGES..pdf Download (705kB)
	Text FULLTEXT..pdf Restricted to Registered users only Download (6MB)

Abstract

This research presents the development of a monocular vision-based positioning system using single Quick Response (QR) code as a landmark by numerical computation. The system is applicable to an Eye In Hand (EIH) type grasping automation system in which the camera sensor is positioned at the end of the robot arm to take a snapshot of the artificial landmark. Using the system developed, threedimensional (3D) position information such as the depth and orientation can be extracted from a two-dimensional (2D) image that consists of a known-size QR code being used as the marker. The four vertices of the QR code are used as the positioning points to determine the 3D information from the underdetermined system with two known parameters, camera’s focal length and QR code’s dimension. The pinhole imaging theory and similar triangles rule are the fundamental concepts used to analyse the relationship between the 2D image coordinates and the 3D world coordinates of the QR code’s vertices. The 3D coordinates of the QR code marker are then determined using the numerical computation model derived with the appropriate guessing parameters and updating rules, and the orientation is extracted using the rotation matrix. To determine the maximum rotation angle attainable at each point in the coordinate plane, the model is simulated using MATLAB platform with 12 different combinations of rotation around three cardinal axes. The simulation’s precision is set to be no more than two-degree angle error and distance difference of less than five mm. The simulation results show that an average 77.28% of the coordinate plane can achieved converged results within 30 iterations. The simulation's output is then validated experimentally in an indoor environment by implementing the positioning model into a hardware setup built using a Raspberry Pi, a camera module and a fixed-dimension QR code marker. In comparison to the simulation results, the experiment results have been verified with satisfactory outcomes of average 73.15% resemblance was attained among the rotation combinations tested. Based on the model simulation and experimental data, the QR code’s 3D information in terms of distance and orientation extracted from single image captured at one optical point has been validated successfully.

Item Type:	Thesis (Masters)
Keyword:	Monocular vision-based positioning system, Quick Response, Numerical computation
Subjects:	Q Science > QA Mathematics > QA1-939 Mathematics
Department:	FACULTY > Faculty of Engineering
Depositing User:	DG MASNIAH AHMAD -
Date Deposited:	10 Sep 2024 10:22
Last Modified:	10 Sep 2024 10:22
URI:	https://eprints.ums.edu.my/id/eprint/40704

Actions (login required)

View Item