PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
July 2018
423
Editorial
“Computer Vision and Big Data Analytics
for Remote Sensing”
The Guest Editors: Vasit Sagan, Sidike Paheding, and Dongodng Wang
Rapid advances in sensor technology, field robotics, unmanned
aerial systems (UAS), and computing power have facilitated ex-
ponential growth of remote sensing applications. Meanwhile,
processing complex, multiscale, and multidimensional data
from UAS, environmental sensors and climate model simula-
tions has become increasingly difficult for both scientists and
public to summarize and visualize the large amount of data for
agricultural and environmental assessments with direct appli-
cations to education, training and decision-making. Data gen-
erated by thousands of images collected in a single UAS flight
is almost impossible to manually analyze. There exist both
numerous opportunities and challenges with broader usage of
multi-sensor remote sensing data, mostly attributed to process-
ing algorithms of ultra-high resolution imagery, translating the
abundant spectral and spatial data to information useful for de-
cision-making. Computer vision, as a multidisciplinary field,
seeks to mimic human visual system for automatic extraction,
analysis, and interpreting useful information from images or a
sequence of video or image frames. And it offers promising an-
alytical solutions for diverse fields of studies concerned with
Big Data. This Special Issue focuses on advancing research in
computer vision and big data analytics for remote sensing ap-
plications to address the state-of-the-art technical and applica-
tion challenges.
The Special Issue: Computer Vision and Big Data Analytics for
Remote Sensing contains four papers related to the hyperspec-
tral image analysis of gold mineralization, detection of buried
remains, oil spill detection, and plant stress detection using
computer vision and machine learning.
The first paper authored by Khan et al., entitled
Characteriza-
tion of gold mineralization in Northern Pakistan using imaging
Spectroscopy,
presents an imaging spectroscopy and geochem-
ical integrated approach to map gold mineralization in north-
ern Pakistan. More specifically, gold mineralization potential
is determined by using the Spectral Angle Mapper (SAM) and
Support Vector Machine (SVM) algorithms. The results show
that SVM yields superior classification accuracy for sulfide
minerals, demonstrating SAM are in a very good agreement
with electron microscope data from laboratory analysis.
The second paper,
Evolutionary approach for detection of bur-
ied remains using hyperspectral images
by Dozal et al. is de-
voted to locate clandestine graves using hyperspectral remote
sensing data. Authors utlize a Genetic Programming technique
called Brain Programming (BP) for automating the design of
Hyperspectral Visual Attention Models (H-VAM), which is pro-
posed as a new method for the detection of buried remains.
The results demonstrate that the proposed method improves
classification accuracy compared to existing approaches and
that images collected after three months from burial are best for
effective detection buried objects.
The next paper by Albalooshi et al.,
Deep belief active con-
tours (DBAC) with its application to oil spill segmentation
from remotely sensed sea surface imagery
, suggests a machine
learning-based oil spill segmentation using aerial images. The
proposed new method, i.e., deep neural network-based object
segmentation – Deep Believe Active Contours (DBAC), outper-
forms the state-of-the-art methods noticeably.
The last paper by Li et al.,
Early drought plant stress detec-
tion with bi-directional long-term memory networks
, presents
a state-of-the art new method in computer vision and machine
learning developments for plant stress detection. Authors pro-
posed a novel deep learning method - the Bidirectional Long
Short-Term Memory (BLSTM) networks for early stress detec-
tion using RGB images, which produces competitive classifica-
tion performance compared to three other deep learning meth-
ods exist in literature.
The four papers published were selected through a rigorous
peer-review process among more than a dozen manuscripts
submitted to this special issue. We hope that the journal readers
will enjoy the papers as they represent novel scientific meth-
ods and state-of-the-art techniques in computer vision, BigDa-
ta and machine learning for remote sensing. We would like to
sincerely thank Dr. Alper Yilmaz, Editor-in-Chief of PE&RS, to
have offered to us the possibility to publish the manuscripts in
his journal, all reviewers for their tremendous contribution to
improve the submitted manuscripts and authors for presenting
their best work to the special issue.
Photogrammetric Engineering & Remote Sensing
Vol. 84, No. 7, July 2018, pp. 423–423.
0099-1112/18/423–423
© 2018 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.84.7.423
