Files in this item



application/pdfAlexander_von Alt.pdf (17MB)
(no description provided)PDF


Title:From interest points to map transformation: a discussion of RGB-D SLAM and its applications
Author(s):von Alt, Alexander
Advisor(s):Hutchinson, Seth A.
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Red Green Blue Depth Simultaneous Localization and Mapping (RGB-D SLAM)
Simultaneous Localization and Mapping (SLAM)
Fast Retina Keypoints (FREAK)
Binary Robust Invariant Scalable Keypoints (BRISK)
Red Green Blue Depth (RGB-D)
Abstract:This thesis examines SLAM using a single handheld projected-IR RGB-D camera. It focuses primarily on the use of different interest point detectors, their descriptor extractors, and their matchers in the estimation of pairwise alignment of frames captured using such a sensor. Such estimates give the initial pose estimate of each frame. They are encoded as edge constraints in the pose graph and integral in SLAM's recovery of the optimal trajectory of poses. BRISK and FREAK feature detectors using FLANN and brute-force matchers are analyzed. Their suitability for use in visual odometry estimation is investigated. A method of automatically tuning the BRISK-AGAST detector so as to produce a consistent number of features is developed and studied in the context of RGB-D SLAM. An augmented reality application making use of RGB-D SLAM is also developed. This method uses RGB-D SLAM to create a map of an environment. The map is then distorted using affine transformations to change the dimensions and appearance of the environment. Views of this distorted environment are generated in real time at the camera's current pose. The effectiveness of the approach is considered and example outputs are provided.
Issue Date:2013-05-24
Rights Information:Copyright 2013 Alexander von Alt
Date Available in IDEALS:2013-05-24
Date Deposited:2013-05

This item appears in the following Collection(s)

Item Statistics