Data Driven Multispectral Image Registration Framework
Date
2018-08-29
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
Type
Thesis
Degree Level
Masters
Abstract
Multispectral imaging is widely used in remote sensing applications from UAVs and ground-based platforms.
Multispectral cameras often use a physically different camera for each wavelength causing misalignment
in the images for different imaging bands. This misalignment must be corrected prior to concurrent
multi-band image analysis. The traditional approach for multispectral image registration process is to select
a target channel and register all other image channels to the target. There is no objective evidence-based
method to select a target channel. The possibility of registration to some intermediate channel before registering
to the target is not usually considered, but could be beneficial if there is no target channel for which
direct registration performs well for every other channel.
In this paper, we propose an automatic data-driven multispectral image registration framework that determines
a target channel, and possible intermediate registration steps based on the assumptions that 1) some
reasonable minimum number of control-points correspondences between two channels is needed to ensure a
low-error registration; 2) a greater number of such correspondences generally results in higher registration
performance.
Our prototype is tested on five multispectral datasets captured with UAV-mounted multispectral cameras.
The output of the prototype is a registration scheme in the form of a directed acyclic graph (actually a tree)
that represents the target channel and the process to register other image channels. The resulting registration
schemes had more control point correspondences on average than the traditional register-all-to-one-targetchannel
approach. Data-driven registration scheme consistently showed low back-projection error across all
the image channel pairs in most of the experiments. Our data-driven framework has generated registration
schemes with the best control point extraction algorithm for each image channel pair and registering images
in a data-driven approach. The data-driven image registration framework is dataset independent, and it
performs on datasets with any number of image channels. With the growing need of remote sensing and the
lack of a proper evidence-based method to register multispectral image channels, a data-driven registration
framework is an essential tool in the field of image registration and multispectral imaging.
Description
Keywords
Image Registration, Multispectral Image, Image Processing, Control Point, SIFT, SURF, BRISK, ORB, Drone Imaging
Citation
Degree
Master of Science (M.Sc.)
Department
Computer Science
Program
Computer Science