Frontiers in Marine Science,
Journal Year:
2024,
Volume and Issue:
11
Published: Nov. 14, 2024
Introduction
Suspended
particulate
matter
(SPM)
is
a
critical
indicator
of
water
quality
and
has
significant
impact
on
the
nearshore
ecological
environment.
Consequently,
quantitative
evaluation
SPM
concentrations
essential
for
managing
environments
planning
marine
resources.
Methods
This
study
utilized
Sentinel-2’s
single
band
index
variables
to
develop
remote
sensing
inversion
model
oceanic
in
estuary
Pinglu
Canal
China.
Six
machine
learning
algorithms
were
employed:
K-nearest
neighbor
regression
(KNNR),
AdaBoost
(ABR),
random
forest
(RF),
gradient
boosting
(GBR),
extreme
(XGBR),
light
generalized
boosted
(LGBM).
The
with
optimal
performance
was
then
selected
further
analysis.
research
applied
established
investigate
spatial-temporal
dynamics
from
2021
2023.
Results
findings
indicated
that
(1)
XGBR
algorithm
exhibited
superior
(R
2
=
0.9042,
RMSE
3.0258
mg/L),
LGBM
=0.8258,
4.0813
mg/L)
GBR
0.823,
4.3477
also
demonstrating
effective
fitting.
However,
ABR,
RF,
KNNR
produced
less
satisfactory
fitting
results.
(2)
Additionally,
revealed
combination
input
more
accurate
than
single-variable
inputs.
(3)
contribution
single-band
surpassed
variables,
B12,
B4,
B11
emerging
as
top
three
influential
model.
(4)
annual
concentration
area
an
overall
increasing
trend,
while
its
spatial
distribution
generally
decreased
toward
Maowei
Sea
Qinzhou
Bay.
Discussion
Sentinel-2
data
shown
good
retrieving
concentration,
providing
new
method
approach
large-scale
estimation
concentration.
Remote Sensing,
Journal Year:
2024,
Volume and Issue:
16(11), P. 1861 - 1861
Published: May 23, 2024
Synthetic
aperture
radar
(SAR)
change
detection
provides
a
powerful
tool
for
continuous,
reliable,
and
objective
observation
of
the
Earth,
supporting
wide
range
applications
that
require
regular
monitoring
assessment
changes
in
natural
built
environment.
In
this
paper,
we
introduce
novel
SAR
image
method
based
on
principal
component
analysis
two-level
clustering.
First,
two
difference
images
log-ratio
mean-ratio
operators
are
computed,
then
fusion
model
is
used
to
fuse
images,
new
generated.
To
incorporate
contextual
information
during
feature
extraction
phase,
Gabor
wavelets
obtain
representation
across
multiple
scales
orientations.
The
maximum
magnitude
all
orientations
at
each
scale
concatenated
form
vector.
Following
this,
cascading
clustering
algorithm
developed
within
discriminative
space
by
merging
first-level
fuzzy
c-means
with
second-level
neighbor
rule.
Ultimately,
combination
changed
unchanged
results
produces
final
map.
Five
datasets
experiment,
show
our
has
significant
advantages
detection.
Remote Sensing,
Journal Year:
2024,
Volume and Issue:
16(20), P. 3804 - 3804
Published: Oct. 13, 2024
The
fusion
of
infrared
and
visible
images
together
can
fully
leverage
the
respective
advantages
each,
providing
a
more
comprehensive
richer
set
information.
This
is
applicable
in
various
fields
such
as
military
surveillance,
night
navigation,
environmental
monitoring,
etc.
In
this
paper,
novel
image
method
based
on
sparse
representation
guided
filtering
Laplacian
pyramid
(LP)
domain
introduced.
source
are
decomposed
into
low-
high-frequency
bands
by
LP,
respectively.
Sparse
has
achieved
significant
effectiveness
fusion,
it
used
to
process
low-frequency
band;
excellent
edge-preserving
effects
effectively
maintain
spatial
continuity
band.
Therefore,
combined
with
weighted
sum
eight-neighborhood-based
modified
(WSEML)
bands.
Finally,
inverse
LP
transform
reconstruct
fused
image.
We
conducted
simulation
experiments
publicly
available
TNO
dataset
validate
superiority
our
proposed
algorithm
fusing
images.
Our
preserves
both
thermal
radiation
characteristics
detailed
features
Fractal and Fractional,
Journal Year:
2024,
Volume and Issue:
8(10), P. 554 - 554
Published: Sept. 25, 2024
In
this
paper,
we
introduce
an
innovative
approach
to
multi-focus
image
fusion
by
leveraging
the
concepts
of
fractal
dimension
and
coupled
neural
P
(CNP)
systems
in
nonsubsampled
contourlet
transform
(NSCT)
domain.
This
method
is
designed
overcome
challenges
posed
limitations
camera
lenses
depth-of-field
effects,
which
often
prevent
all
parts
a
scene
from
being
simultaneously
focus.
Our
proposed
technique
employs
CNP
with
local
topology-based
model
merge
low-frequency
components
effectively.
Meanwhile,
for
high-frequency
components,
utilize
spatial
frequency
dimension-based
focus
measure
(FDFM)
achieve
superior
performance.
The
effectiveness
validated
through
extensive
experiments
conducted
on
three
benchmark
datasets:
Lytro,
MFI-WHU,
MFFW.
results
demonstrate
superiority
our
method,
showcasing
its
potential
significantly
enhance
clarity
across
entire
scene.
algorithm
has
achieved
advantageous
values
metrics
QAB/F,
QCB,
QCV,
QE,
QFMI,
QG,
QMI,
QNCIE.
Frontiers in Plant Science,
Journal Year:
2025,
Volume and Issue:
16
Published: March 7, 2025
Glycyrrhiza
uralensis
Fisch.,
a
perennial
medicinal
plant
with
robust
root
system,
plays
significant
role
in
mitigating
land
desertification
when
cultivated
extensively.
This
study
investigates
Dengkou
County,
semi-arid
region,
as
the
research
area.
First,
reflectance
differences
of
feature
types,
and
importance
bands
were
evaluated
by
using
random
forest
(RF)
algorithm.
Second,
after
constructing
G.
vegetation
index
(GUVI),
recognition
accuracy
was
compared
between
RF
classification
model
constructed
based
on
January-December
GUVI
common
indices
set
support
vector
machine
(SVM)
set.
Finally,
spectral
characteristics
other
types
under
2022
analyzed,
historical
distribution
identified
mapped.
The
results
demonstrated
that
blue
near-infrared
are
particularly
for
distinguishing
.
Incorporating
year-round
(January-December)
data
significantly
improved
identification
accuracy,
achieving
producer’s
97.26%,
an
overall
93.00%,
Kappa
coefficient
91.38%,
user’s
97.32%.
Spectral
analysis
revealed
distinct
different
years
types.
From
2014
to
2022,
expanded
from
northeast
County
central
southwestern
regions,
transitioning
small,
scattered
patches
larger,
concentrated
areas.
highlights
effectiveness
models
identifying
,
demonstrating
superior
performance
alternative
sets
or
algorithms.
However,
generalizability
may
be
limited
due
influence
natural
anthropogenic
factors
Therefore,
regional
adjustments
optimization
parameters
necessary.
provides
valuable
reference
employing
remote
sensing
technology
accurately
map
current
regions
similar
environmental
conditions.
Sensors,
Journal Year:
2025,
Volume and Issue:
25(9), P. 2646 - 2646
Published: April 22, 2025
The
fusion
of
infrared
and
visible
images
provides
complementary
information
from
both
modalities
has
been
widely
used
in
surveillance,
military,
other
fields.
However,
most
the
available
methods
have
only
evaluated
with
subjective
metrics
visual
quality
fused
images,
which
are
often
independent
following
relevant
high-level
tasks.
Moreover,
as
a
useful
technique
especially
low-light
scenarios,
effect
conditions
on
result
not
well-addressed
yet.
To
address
these
challenges,
decoupled
semantic
segmentation-driven
image
network
is
proposed
this
paper,
connects
downstream
task
to
drive
be
optimized.
Firstly,
cross-modality
transformer
module
designed
learn
rich
hierarchical
feature
representations.
Secondly,
semantic-driven
developed
enhance
key
features
prominent
targets.
Thirdly,
weighted
strategy
adopted
automatically
adjust
weights
different
modality
features.
This
effectively
merges
thermal
characteristics
detailed
images.
Additionally,
we
design
refined
loss
function
that
employs
decoupling
constrain
pixel
distributions
produce
more-natural
evaluate
robustness
generalization
method
practical
challenge
applications,
Maritime
Infrared
Visible
(MIV)
dataset
created
verified
for
maritime
environmental
perception,
will
made
soon.
experimental
results
public
datasets
practically
collected
MIV
highlight
notable
strengths
best-ranking
among
its
counterparts.
Of
more
importance,
achieved
over
96%
target
detection
accuracy
dominant
high
mAP@[50:95]
value
far
surpasses
all
competitors.
Mathematics,
Journal Year:
2023,
Volume and Issue:
11(18), P. 3803 - 3803
Published: Sept. 5, 2023
Multi-focus
image
fusion
is
a
popular
technique
for
generating
full-focus
image,
where
all
objects
in
the
scene
are
clear.
In
order
to
achieve
clearer
and
fully
focused
effect,
this
paper,
multi-focus
method
based
on
parameter-adaptive
pulse-coupled
neural
network
fractal
dimension
nonsubsampled
shearlet
transform
domain
was
developed.
The
pulse
coupled
network-based
rule
used
merge
low-frequency
sub-bands,
dimension-based
via
multi-scale
morphological
gradient
high-frequency
sub-bands.
inverse
reconstruct
fused
coefficients,
final
generated.
We
conducted
comprehensive
evaluations
of
our
algorithm
using
public
Lytro
dataset.
proposed
compared
with
state-of-the-art
algorithms,
including
traditional
deep-learning-based
approaches.
quantitative
qualitative
demonstrated
that
outperformed
other
as
evidenced
by
metrics
data
such
QAB/F,
QE,
QFMI,
QG,
QNCIE,
QP,
QMI,
QNMI,
QY,
QAG,
QPSNR,
QMSE.
These
results
highlight
clear
advantages
fusion,
providing
significant
contribution
field.
Frontiers in Neurorobotics,
Journal Year:
2024,
Volume and Issue:
18
Published: Nov. 11, 2024
With
the
rapid
development
of
Industrial
Internet
Things
(IIoT)
technology,
various
IIoT
devices
are
generating
large
amounts
industrial
sensor
data
that
spatiotemporally
correlated
and
heterogeneous
from
multi-source
multi-domain.
This
poses
a
challenge
to
current
detection
algorithms.
Therefore,
this
paper
proposes
an
improved
long
short-term
memory
(LSTM)
neural
network
model
based
on
genetic
algorithm,
attention
mechanism
edge-cloud
collaboration
(GA-Att-LSTM)
framework
is
proposed
detect
anomalies
facilities.
Firstly,
established
real-time
process
amount
at
edge
node
in
real
time,
which
reduces
time
uploading
cloud
platform.
Secondly,
overcome
problem
insufficient
important
features
input
sequence
traditional
LSTM
algorithms,
we
introduce
adaptively
adjust
weights
model.
Meanwhile,
algorithm
optimized
hyperparameters
transform
anomaly
into
classification
effectively
extract
correlation
time-series
data,
improves
recognition
rate
fault
detection.
Finally,
method
has
been
evaluated
publicly
available
database.
The
results
indicate
accuracy
99.6%,
F1-score
84.2%,
precision
89.8%,
recall
77.6%,
all
exceed
performance
five
machine
learning
methods.