IEEE Geoscience and Remote Sensing Magazine - March 2022 - 4

FROM THE EDITOR
BY JAMES L. GARRISON
Welcome to the March Issue
W
elcome to the March 2022 issue of IEEE Geoscience
and Remote Sensing Magazine! Our theme for
this issue is " Algorithms to Applications. " The 13 features
cover the full breadth of the remote sensing field,
from microwave to optical observations with ground,
airborne, and satellite platforms. They are organized
with a focus on algorithms in the first eight articles,
followed by five that review interesting and important
applications. Several of the features make use of bibliometric
analysis to show the distribution of published
articles within their subject area, a technique that is increasingly
being used in review and survey articles.
Starting off with algorithms, Peng et al. review lowrank
and sparse representations for hyperspectral image
(HSI) processing. Having low dimensionality and high
correlation between channels, HSI data are inherently
sparse. This sparse structure can be exploited by various
methods for denoising, resolution enhancement, and
classification, for example. Issues arising with a limited
number of training samples are also discussed. Ultimately,
the authors hope that their review can provide
guidelines and inspiration to users of HSI and those
working to advance the field.
Our second feature reviews algorithms for land cover
change detection in very-high-resolution (VHR) optical
images. Although the detail provided by VHR images can
be used to detect land cover changes, the results are often
poor due to the limited amount of spectral information.
This article reviews various approaches to improve this
performance. Three pairs of actual VHR images are used in
conducting comparisons between several different methods.
They conclude by identifying directions for future research,
including radiometric corrections, the use of object
topology, and trend analysis using object time series.
The large amount of spectral information captured in
HSI enables the detection of anomalies or targets through
Digital Object Identifier 10.1109/MGRS.2022.3149666
Date of current version: 8 April 2022
4
subtle discrepancies in the spectral signature of ground
objects. This is the subject of the survey by Su et al., which
is organized into seven categories of hyperspectral anomaly
detection methods. Nine representative algorithms
were selected and demonstrated on three typical data sets
(small target, large target, and complex scene). A number
of performance indices are used to evaluate the different
methods. The article concludes with suggestions for future
research and the identification of some key technical
problems needing to be solved.
The next two features are concerned with object
detection and tracking from unpiloted aerial vehicles
(UAVs). Wu et al. look at deep learning (DL) applied to
this problem in three areas: object detection from an
image, object detection from video, and object tracking
from video. A comprehensive survey covers the
current state of research on DL methods as applied to
this problem. Four benchmark data sets are used for
the evaluation and comparison of some representative
methods in terms of their detection success and
computational cost. Prospects for future work are
summarized.
In the second feature on this problem, Fu et al. review
discriminative correlation filter (DCF)-based trackers on
UAVs. The high computational efficiency of these trackers
and their performance in complex environments
make them attractive for onboard implementation.
This article reviews the timeline of DCF development
and presents a generalized framework for DCF-based
tracking. Six benchmarks are used to evaluate different
approaches in performing challenging tracking tasks in
real time from UAVs. Challenges, including inadequate
sampling, fast motion, and changing illumination and
viewpoint, are discussed along with an assessment of
typical failure cases. The article concludes with suggestions
for future work, such as adaptive learning rates,
intelligent search strategies, embedded image processing,
and multimodal trackers. A publicly released code
library is available to help advance progress in this field.
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE MARCH 2022
IEEE Geoscience and Remote Sensing Magazine - March 2022

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2022
