European Journal of Remote Sensing

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tejr20

An investigation in satellite images based on

image enhancement techniques

R. Ablin , C.Helen Sulochana & G. Prabin

To cite this article: R. Ablin , C.Helen Sulochana & G. Prabin (2020) An investigation in satellite
images based on image enhancement techniques, European Journal of Remote Sensing, 53:sup2,
86-94, DOI: 10.1080/22797254.2019.1673216

To link to this article: https://doi.org/10.1080/22797254.2019.1673216

© 2019 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group.

Published online: 02 Oct 2019.

Submit your article to this journal

Article views: 1596

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tejr20
EUROPEAN JOURNAL OF REMOTE SENSING
2020, VOL. 53, NO. S2, 86–94
https://doi.org/10.1080/22797254.2019.1673216

An investigation in satellite images based on image enhancement techniques

a
R. Ablin , C.Helen Sulochanab and G. Prabinc
a
Arunachala College of Engg., for Women; bSt.Xaviers Catholic College of Engineering; cBSNL, India

ABSTRACT ARTICLE HISTORY

In Satellite Images, enhancement plays a dynamic research topic in image processing. The aim Received 16 April 2019
of enhancement is to process an image so that the result is more suitable than original image Revised 2 September 2019
for specific remote sensing application. Satellite image enhancement techniques provide a lot Accepted 24 September 2019
of choices for improving the visual quality of remotely sensed images. In this research review, KEYWORDS
image fusion plays an important role, since it effectively combines auxiliary image content to Image fusion; satellite image
enhance information contained in the individual datasets. This article provides an overview of enhancement; image
the existing enhancement techniques. There are many techniques which have been proposed processing; visualization;
for enhancing the digital images which may be used for enhancing Satellite images. Here, remote sensing
a survey on various Satellite image enhancement techniques has been performed which
recommends fusion-based enhancement performs superior while comparing with non-fusion-
based enhancement techniques.

Introduction reflected by the targets in the pixel. Thus,

a panchromatic image is interpreted as a black-and-
In Satellite Images, Hyperspectral Image enhancement
white aerial photograph of the object.
plays an active research topic in remotely sensed image
A panchromatic mode sensor gives high spatial reso-
processing. Hyperspectral image classification has
lution image, but it lacks spectral resolution, which does
become a challenging problem due to mixed pixels
not contain any color information. Therefore, fusion-
that would improve by enhancing the hyperspectral
based enhancement technique is used to get the spatial
image and unmixing of classes. Developing and imple-
and spectral information from the image. Since the
menting the enhancement techniques require adequate
panchromatic image does not cover the same spectral
information of the existing problems and idea about the
range as the hyperspectral image, the details extracted
obtained hyperspectral image. Enhancement of hyper-
from the panchromatic image would result in the intro-
spectral images plays a vital role in classifying the pure
duction of spectral distortions with unclear foreground
pixels from the mixed one. Despite many significant
and background information. Therefore, a model is
advances made in the field of enhancement, unmixing
required to add the details of the matte in such a way
of classes from hyperspectral image is a challenging task
that the spatial quality of the hyperspectral image is
due to its high dimensionality, low spatial resolution
improved while its spectral quality remains unchanged
and mixed pixels. This article reviews the most relevant
with sharp edge information. The algorithmic approach
existing enhancement methods.
to obtain high-resolution images is fusion of bands
which have rich entropy to increase the spatial informa-
Review of hyperspectral image enhancement tion. Many sensing platforms are equipped to capture
methods high spectral and low spatial resolution hyperspectral
image as well as low spectral and high spatial resolution
In the last decades, many researchers in the field of
auxiliary image (i.e. panchromatic image).
hyperspectral imaging have developed significant
Generally, the enhancement algorithm is of two
approaches for the enhancement of images in hyper-
methods, namely non-fusion-based and fusion-based.
spectral technology. Hyperspectral imagery is typically
The Various approaches to enhancement methods are
collected and represented as a data cube or image cube
shown in Figure 1.
with spatial information collected in the X-Y plane, and
spectral information is represented in the Z direction.
Most of the sensors operate either in panchromatic Non-fusion-based enhancement methods
mode or hyperspectral mode. A panchromatic image
Non-fusion-based enhancement methods focus on the
consists of only one band. It is usually displayed as a gray
spatial resolution of hyperspectral imaging systems.
scale image, i.e. the displayed brightness of a particular
Three different non-fusion-based enhancement methods
pixel is proportional to the intensity of solar radiation

CONTACT R. Ablin [email protected] Arunachala College of Engg., for Women

© 2019 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
 EUROPEAN JOURNAL OF REMOTE SENSING 87

Hyperspectral image
Enhancement

Non-fusion-based
Fusion-based method
method

Numerical
Spectral
Learning- Matting- Component Multi- & Optimizati
Mixture
based based Substitution resolution Statistical- on -based
Analysis
approach approach approach approach based approach
approach
approach

Figure 1. Various satellite image enhancement approaches.

are described in the literature, namely Spectral Mixture Learning-based approach

Analysis-based, Learning-based and Matting-based The second classification of non-fusion-based method
method. is termed as learning-based method in which a set of
training images is used for learning the super resolution
Spectral mixture analysis approach attributes. Based on the learning method, it is categor-
Spectral Mixture Analysis (SMA) is a soft- ized into Hopfield Neural Network (HPNN) and Back
classification approach which models the total reflec- Propagation Neural Network (BPNN). A method used
tance in a pixel as the linear combination of reflectance by Gu, Zhang, and Zhang (2008) obtains the abun-
from each class using LMM which predicts the pro- dance map using linear SMA which is based on spatial
portion of each class within each pixel using the correlation of land covers. In this method, they use low-
model. A variety of approaches based on SMA using resolution training images to determine the parameters
LMM have been proposed to address the problem of used in super resolution mapping.
spatial resolution in hyperspectral images. SMA-based Han et al. (2019) have applied a similar approach
sub-pixel processing is performed in which the spatial which uses low-resolution images and their down-
dependencies of materials in mixed pixels are not sampled versions to train BPNN in the learning-
considered thus acting as an initial stage for the spatial based method. A mean filter is used for the down
resolution enhancement of hyperspectral images sampling, and the super resolution is performed by
(Ruescas et al., 2010). considering spatial correlation of different materials
Brown, Gunn, and Lewis (1999) have proposed present in hyperspectral images. Hence, hyperspectral
a linear SVM approach, which estimates land cover images themselves are treated as the training data to
components by sub-pixel processing. This model auto- achieve better coherence between the results of the
matically selects the relevant pure pixels and determines enhanced and the original hyperspectral images.
the number of classes in the region of interest by provid- Zhang and Mishra (2014) have implemented
ing accurate representation of land covers. Atkinson a support vector regression approach which does not
(2005) proposed an algorithm for the enhancement of use explicit formula to describe prior information
hyperspectral images in spatial resolution using sub-pixel about the nonlinear relationships between the coarse
target mapping. High-resolution pixels are placed based fractional pixels and labelled sub-pixels from the best
on spatial correlation using a distance-weighted function. matching high-resolution training data. Due to the
In the algorithm proposed by Villa, Chanussot, above-mentioned limitations, learning-based method
Benediktsson, and Jutten (2011), the spectral unmixing is hardly used in practice.
is performed to determine the proportion of endmem-
bers in each pixel. Sub-pixels in the model are located by Matting-based approach
spatial resolution mapping performed by simulated The third classification of non-fusion-based approach
annealing. However, the limitation is the requirement is termed as matting-based approach which extracts
of high computational load because of the large number the foreground object from an image. Levin, Rav-
of bands present in hyperspectral images. Even though Acha, and Lischinski (2008) have proposed a new
SMA methods provide the abundances of the endmem- alpha matting model in which one assumes the colors
bers within a pixel which is very useful in determining of the foreground and the background to vary linearly
the presence of object in remote sensing applications; inside a small patch. The result is imperfect with
the limitation is that they do not exploit spatial and insufficient user input. Wang and Suter (2007) have
spectral information to their full capacity. introduced a new model in which data terms are based
88 R. ABLIN ET AL.

on color models of the foreground and the back- color distortion because the auxiliary information is not
ground regions. However, the result is worse, that created from the same wavelengths of light as the RGB
some dark-green areas in the image background are image. Therefore, this method has been modified to Fast
semi-transparent layers, i.e. dark green is a mix of dark Intensity-Hue Saturation (FIHS) (Tu, Huang, Hung, &
foreground with green background. Chang, 2004).
Chen, Zou, Zhiying Zhou, Zhao, and Tan (2013) The modification performed in the FIHS method is
proposed a new image matting with local and nonlocal that it extends the IHS method from three bands to
smooth priors. In this method, editing propagation four by incorporating an infrared component because
essentially introduces a nonlocal smooth prior on the the auxiliary information is taken from infrared light,
alpha matte in which the manifold is preserved. Prior in addition to visible wavelength. This modification
from matting the non-locally smooth Laplacian com- allowed the calculated intensity to better match the
plements each other, and hence for natural matting, it auxiliary information, thus causing less color distor-
is combined with a data term from color sampling. The tion in the fused image. The trade-off between spatial
color distribution is similar in foreground and back- improvement and spectral quality loss received a lot of
ground images. It is not easy to set a common window attention and led to the introduction of trade-off para-
size for all test data which is a limitation. This method meters (Tu et al. 2004). To obtain the desired result,
generalizes good results with a small fixed window size these parameters allow a fine tuning by the user.
with the help of nonlocal smoothness constraint. To overcome the spectral quality problems, research-
A matting technique called KNN matting is pro- ers have proposed the Adaptive IHS method (Rahmani,
posed for ordinary image by Chen et al. (2013) with Strait, Merkurjev, Moeller, & Wittman, 2010) which
a closed form solution that can hardness the precondi- adaptively adjusts linear combination of the coefficients
tioned conjugate gradient method and runs in a few of multispectral bands. The weights induced by edge
seconds after accepting very sparse user mark-ups. Xu, injection process in the spatial detail are too large that
Price, Cohen, and Huang (2017) have proposed a deep results in color changes. Thus, this causes spectral dis-
image matting model to obtain high-level context and tortion. In addition to this, weights induced by edges
use high-level features. In this method a neural net- lead to reduction in sharpness of the fused image.
work is capable of capturing higher order features Another improved method which is Improved AIHS
resulting in higher computational complexity. (Leung, Liu, & Zhang, 2014) is designed by a more
Even though the non-fusion-based methods pro- adaptive weighting matrix in the spatial detail injection
vide a good solution to hyperspectral image enhance- step. It performs better than AIHS. Edge of high reflec-
ment by gaining information while extracting the tion area is more prone to distortion. Hence, the overall
foreground from its background images, high spatial spectral distortion is higher.
resolution is not obtained. In order to have high spa- Another technique, namely, Generalized IHS
tial information, fusion-based enhancement is needed. Brovey Transform (BT) Smoothing Filter-based
Intensity Modulation (SFIM) Dehnavi and
Mohammadzadeh (2013) is proposed which incorpo-
Fusion-based enhancement methods
rates Generalized IHS, BT and SFIM by using two
In fusion-based enhancement methods, hyperspectral adjustable parameters. This modulation approach is
images generate the high spatial resolution scene by the most frequently employed approach in which the
fusing a low spatial resolution hyperspectral image spatial and spectral information are controlled. In
with the auxiliary information. The fusion-based addition, it preserves more spectral information, but
enhancement methods are classified into component suffers more spatial information loss.
substitution approach, numerical and statistical-based Hubert et al. (2005) has proposed a novel fusion
approach, multiresolution approach and optimiza- technique which is performed using Principal
tion–based approach. Component Analysis (PCA) approach. This approach
partitions the dataset into sub-groups of bands.
Component substitution (CS) approach Therefore, computational complexity is reduced.PCA
The most popular methods are Intensity-Hue Saturation is applied to each subgroup based on dominant classes.
(IHS) color transformation and Principal Component The spectral signature of a class is used as the transfer
Substitution. A very popular technique is IHS (Malpica, function of matched filter applied to corresponding
2007). Color enhancement, feature enhancement and bands of the dataset. The principal component of each
improvement of spatial resolution are the standard pro- sub-group is used as a component of the final RGB
cedures in analysis of images (Pohl & Van Genderen, image. Since the energy is not uniformly distributed
1998). This technique converts a color image from RGB for each group, the color distortion is low which results
space to the HIS color space. Here, the intensity band is in less visual quality.
replaced by the auxiliary information. Implementing this Qu et al. (2018) proposed a Structure Tensor-based
method is very efficient, but this technique produces algorithm for Hyperspectral and Panchromatic Image
 EUROPEAN JOURNAL OF REMOTE SENSING 89

fusion. In this algorithm, an image enhancement the coefficients from the source image which have
approach is utilized to sharpen the spatial information higher energy in the local region.
of panchromatic image and the spatial details of the MRA-based approaches decompose images into
Hyperspectral image which is obtained by using an many number of channels depending on the local
adaptive weight method. The structure tensor is intro- frequency content (Nunez et al., 1999). The pyramid
duced to extract spatial details of the enhanced pan- is used to represent the multi-scale models for the
chromatic image. In order to avoid artifacts at the original image. With increasing level, the original
boundaries, a guided filter is applied to the integrated image is approximated at coarser spatial resolution.
spatial information image. To reduce spectral and spa- In between the individual pyramid levels, the trans-
tial distortion, an injection matrix is constructed. This form is performed using wavelet and Curvelet trans-
algorithm provides more spatial details while preserving forms. The wavelet transform approach is based on
the spectral information. Xie et al. (2019) proposed an substitution and addition. In substitution approach,
enhancement algorithm using multispectral and selected multispectral wavelet planes are substituted
Hyperspectral fusion model, based on the observation by the planes of the corresponding panchromatic
models. In this method, all parameters can be learned images. In addition approach, the decomposed pan-
from the training data and spatial, spectral response chromatic planes are added to the multispectral bands.
operators are discovered. This algorithm provides Garzelli, Nencini, Alparone, and Baronti (2005) have
color and brightness much closer to the low- proposed a fusion based on multiresolution analysis to
resolution Hyperspectral image. Jayanth, Kumar, and describe how high-pass information is modelled from
Koliwad (2018) proposed an enhancement algorithm panchromatic image. The basic wavelet transform sub-
using regionally weighted principal component analysis stitution methods are low-low, low-high, high-low and
and wavelet algorithm. In this algorithm spectral infor- high-high. These decompositions exist to form the pyr-
mation is preserved with improvement in spatial quality amid at several levels. The obtained fused image is
and good clarity. Parveen, Kulkarni, and Mytri (2018) inverse transformed. In practice, the wavelet transform
proposed an image enhancement algorithm for low- function and scaling are not explicitly derived
resolution satellite images. This algorithm improves (Amolins, Zhang, & Dare, 2007). They are described
the interpretation and makes the image visually clear. by coefficients which are fused by the different rules of
Component substitution-based approaches focus on fusion to produce the resultant image Better results are
making an ideal image intensity and a high-frequency obtained when the fusion process is context-driven
injection model to preserve spectral information. But (Aiazzi et al. 2002). This process is to make the fused
incurse more spatial information loss sharpness reduc- bands the most similar to the narrow band multispec-
tion and increasing spectral distortion. Some algorithms tral sensor image with the same resolution as the broad-
can be applied only to a specific sensor, although a few band image sensing the single panchromatic band. In
commercially available fusion software tools have pro- order to achieve gain equalization, the higher frequency
ven to be suitable for all available optical panchromatic coefficients taken from the high-resolution image are
and multispectral images. In addition, these tools have selected based on statistical congruence and weighted
a greater potential to improve the spectral quality, by a space-varying factor. Ringing artifacts are comple-
although they only show visually prominent results. tely moderate. Here, the spectral signatures of small size
may be restored (Aiazzi, Alparone, Barducci, Baronti, &
Pippi, 2001), even though a heavily smeared image is
Multi-resolution approach obtained.
Multi-Resolution Approach (MRA) merges the spa- Pradhan, King, Younan, and Holcomb (2006) have
tial information from a high-resolution image with proposed a multiresolution analysis which is extended
the radiometric information from a low-resolution to discrete function. The space is conserved and deter-
image. The process is to sharpen the low-resolution mines the best possible number of decomposition
image. In recent years, the powerful MRA technique levels required for merging images with a particular
such as wavelets, curvelets and others have become resolution ratio. If the resolution ratio is high, the
popular because of increase in computational power decomposition levels are more to produce better
and availability of algorithms in commercial remote results. Computational complexity is more due to
sensing software. Fusion based on multiresolution more decomposition levels. Recently, Contourlet
contourlet transform has been proposed by Miao Transform (CT) have been proposed by Metwalli
and Wang (2006). In this approach, first directional et al. (2014). This transform captures and links dis-
image pyramids up to certain levels using the con- continuity points into linear structures. The ability is
tourlet decomposition are obtained. The low- to have different number of directions at each scale of
frequency coefficients which are at the top of the multiresolution decomposition. The non-subsampled
image pyramids are fused using the average-based CT works on a non-subsampled pyramid and pro-
rule. At the remaining levels, the fusion rule selects duces better results. This technique can also be
90 R. ABLIN ET AL.

found as hybrid component together with PCA and comprised in panchromatic image. This method was
IHS (Xiao-Hui 2008). extended recently by Duran, Coll, and Sbert (2013) to
Proportional Additive Wavelet and Laplacian-based adapt the process considering local relationships of
context-based decision method have been considered as neighbouring pixels which have the denoising effect.
the good image fusion approaches during the Data- In commercialization purpose, one of the statistical
fusion contest, which perform better than CS-based approach algorithms performed is Fuze Go known as
method (Alparone, Aiazzi, Baronti, Garzelli, & pansharpening algorithm. It uses a least square fit
Nencini, 2006). However, in MRA-based fusion, spatial between the gray values of the input bands. The output
distortions may occur, because of aliasing effects and values are estimated with statistical methods (Xu et al
the blurring of textures, and spatial enhancement is not 2014). The strength is that the fully automated process
satisfactory compared with CS – based methods (Aiazzi, also allows inexperienced users to achieve good
Alparone, Baronti, Garzelli, & Selva, 2006). MRA results, and the fact that the input images are treated
(Mallat, 1989) have provide effective tools, like wavelets individually to find the best match (Zhang & Mishra,
and pyramid, to carry out image merging tasks. 2014). Devika and Parthasarathy (2018) proposed
However, in the case of high pass detail injection in an a fuzzy statistics-based technique for enhancing the
image, spatial distortions, aliasing effects, originating satellite images. This algorithm results an efficient and
shifts or blur of contours and textures may occur accurate fuzzy clustering. Therefore, to enhance the
(Yocky, 1996). These disadvantages, which may be as contrast, techniques which jointly perform the combi-
much annoying as spectral distortions, are emphasized nation operation with a solution are required.
by mis-registration between MS and Panchromatic
data, especially if the MRA underlying detail injection Optimization-based approach
is not shift-invariant (Aiazzi et al. 2002, González- The maximization or minimization of a real function
Audícana, Saleta, Catalán, & García, 2004). Wenyan, by choosing inputs within an allowable set for deter-
Zhenhong, Yu, Yang, and Kasabov (2018) proposed mining the resultant of the function is termed as
an enhancement algorithm based on equal weight optimization technique. In order to solve optimiza-
image fusion which improves the accuracy of change tion-based problems algorithms or iterative methods
detection with less visuality. To avoid these problems, are used that converge to a finite solution. An optimi-
Numerical and statistical-based approach is proposed zation-based approach is used for fusion for multi-
which gives more efficient outputs. exposure optical images by Raman and Chaudhuri
(2007) where a set of images has been fused for the
Numerical and statistical-based approach purpose of enhancing the dynamic range in the output
The simplest and earliest methods used in remote sen- image. However, due to smoothness incorporation in
sing are mathematical combinations of different images. the resultant image in cost function, it results in
Addition, subtraction, multiplication and division a smooth solution. A fast approach for fusion of
approaches play an important role in earth observation. hyperspectral images through redundancy elimination
One of the approaches is called subtractive resolution was proposed by Kotwal and Chaudhuri (2010). In
merge influence of user and predefined calculated band this method, a specific set of image bands selected is
weights in subtractive resolution without difference in mutually correlated, and most of the information is
result have been analyzed by Ashraf, Brabyn, and Hicks retained in the data. As only a fraction of the entire
(2013). A classical technique is BT based on the spectral data is being fused, this method is computationally
modelling that reaches a normalization of the input much faster. A new approach for visualization-based
band through subtraction and addition. A major draw- fusion of hyperspectral image bands was proposed by
back is the distortion in color induced by BT. Kotwal and Chaudhuri (2012). Here, the geological
A modification of BT is the colour normalized spectral input data have a very small value of intrinsic contrast
sharpening (Vrabel, 2000). This method groups the and is difficult to visualize.
input bands into spectral segments and is, therefore, Xu, Zhang, Li, and Ding (2015) have proposed a Gram
an adaptive approach which improves the spectral qual- Schmidt approach which generates a simulated lower
ity of the fused images. resolution pan image through weighted sum of Green,
Another modification is modified BT which is based Blue and Red and near-infrared multispectral bands. In
on local modulation of the multispectral image by the spatial and spectral evaluations, results are blurred in all
ratio of the new intensity and initial intensity compo- the band combinations and inobvious color distortion
nents (Chibani, 2007). The variational model is formu- and strange artifacts are introduced. In addition, this
lated by Ballester, Caselles, Igual, Verdera, and Rougé transformation is computationally intensive, and hence
(2006) which describe the relationship between lower it takes more time in generating output images. Wang
resolution multispectral image and high-resolution et al. (2013) proposed a projected gradient approach
panchromatic image using subsampling and filtering. based on unmixing-based non-negative matrix factoriza-
It assumes that multispectral image with its geometry is tion. This method produces the fused image with high
 EUROPEAN JOURNAL OF REMOTE SENSING 91

spectral and spatial resolution. It improves the spatial resolution multispectral image. Gewali et al. (2018)
resolution without losing much of its color information. proposed an algorithm for Hyperspectral image analysis
Rajathurai A and Chellakon H S (2018) proposed a KNN based on machine learning. This analysis algorithm
matting model which has a closed form solution that extracts the desired information from intrinsic spectral
leverages the existing approaches by producing efficient variation while ignoring the extrinsic variation and
visualization multilayer extraction results with reduced intrinsic variation caused by unrelated factors.
computational complexity. Li et al. (2019) proposed an enhancement algorithm
Ben Abbes, Bounouh, Farah, de Jong, and Martínez which offers large scale degraded underwater images.
(2018) compare three satellite image using time series This algorithm is highly desirable in which effective
decomposition methods for vegetation change detec- non-reference underwater image quality evaluation
tion. This results of the comparative analysis show the metrics are calculated. Maselli, Chiesi, and Pieri (2016)
better performance of image fusion techniques when proposed a novel approach for the enhancement of
compared to non-fusion based techniques. Kaplan spatial properties which produces NDVI image series.
(2018) proposed a weighted intensity hue saturated The statistical method is applied to improve the spatial
transform algorithm for image enhancement. In this features of the abundance images based on the end
technique, the intensity component is obtained by members. Tiede, Baraldi, Sudmanns, Belgiu, and Lang
weighting function which preserves more information (2017) proposed an architecture and prototypical
from the bands of the input image so that the visual implementation of a semantic querying system for big
and quantitative comparisons give superior results. earth observation image bases, which enhances the
Hashimoto et al. (2011) proposed a multispectral vision of the photographic images. Gavankar and
image enhancement algorithm which gives an effective Ghosh (2018) proposed an automatic building foot
visualization. In this method, the user can specify the print extraction from high-resolution satellite image
spectral band to extract the spectral feature and the using mathematical morphology. In this approach,
color for visualizing independently so that the desired buildings can be detected from different size and
feature is enhanced in spectral domain in the specified shapes. This method eliminates false-detected build-
color. Mozgovoy, Hnatushenko, and Vasyliev (2018) ings. Lal and Anouncia (2016) proposed an enhanced
proposed an algorithm for automated recognition of dictionary-based sparse representation fusion for multi-
vegetation, waterbodies and the territory in satellite temporal remote sensing images. A locally adaptive
images. This algorithm provides a significant increase dictionary is created such that the dictionary contains
of the efficiency and reliability while updating maps of patches extracted from images. This technique pre-
large cities which reduces the financial cost and also serves spectral information, errors, color and visual
the human errors are minimised. quality of the fused product.
Guo, Ma, Bao, and Wang (2018) proposed an algo-
rithm for fusing panchromatic and short wave infrared
bands based on convolutional neural network. This Conclusion
method effectively enhances the spatial information by
separating the basic architecture into three layers. This article summarizes the review of Satellite
Yadav and Agrawal (2018) proposed an enhancement image enhancement methods. Nonfusion-based
algorithm using road network identification and extrac- enhancement method provides low spatial informa-
tion in satellite imagery using otsu’s method. This tion with higher computational complexity. In
method detects and extracts the road network from order to improve the spatial information and
high-resolution satellite images which enhances the reduce the computational complexity, fusion-based
contrast of the image. Md Noor, Ren, Marshall, and method is preferred. Fusion-based enhancement
Michael (2017) proposed an enhancement algorithm results in high spectral distortion, low spatial reso-
for corneal epithelium injuries in Hyperspectral images lution, reduced contrast and sharpness. Therefore,
this algorithm improves the interpretability of data into it is necessary to have a fusion technique to
clinically relevant information to facilitate diagnostics. enhance the Satellite images for better visualization
Gunlu (2014) proposed an enhancement algorithm and classification accuracy.
for the prediction of stand parameters using pan-
sharpened IKONOS satellite image. Multiple stepwise
regression analysis is used to estimate these stand para- Disclosure statement
meters. It gives high accurate measurements, with No potential conflict of interest was reported by the authors.
higher cost and time. Qifal Wang, Jia, Qin, Yang, and
Hu (2011) proposed an enhancement technique for
multispectral and panchromatic image fusion. This ORCID
method obtains a high spatial resolution multispectral
image with high similarity referenced true high- R. Ablin http://orcid.org/0000-0001-8116-9868
92 R. ABLIN ET AL.

References Photogrammetry, Remote Sensing and Spatial

Information Sciences, 1(3), 139–145. doi:10.5194/isprsarc-
Aiazzi, B., Alparone, L., Barducci, A., Baronti, S., & Pippi, I. hives-XL-1-W3-139-2013
(2001). Information-theoretic assessment of sampled Devika, G., & Parthasarathy, S. (2018). Fuzzy statistics-based
hyperspectral imagers. IEEE Transactions on Geoscience affinity propagation technique for clustering in satellite
and Remote Sensing, 39(7), 1447–1458. doi:10.1109/ cloud image. European Journal of Remote Sensing, 51(1),
36.934076 754–764. doi:10.1080/22797254.2018.1482731
Aiazzi, B., Alparone, L., Baronti, S., & Garzelli, A. Duran, J., Coll, B., & Sbert, C. (2013). Chambolle’s projection
(2002). Context-driven fusion of high spatial and spec- algorithm for total variation denoising. Image Processing
tral resolution images based on oversampled multi- on Line, 2013, 311–331. doi:10.5201/ipol.2013.61
resolution analysis. IEEE Transactions on Geoscience Garzelli, A., Nencini, F., Alparone, L., & Baronti, S. (2005,
and Remote Sensing, 40(10), 2300–2312. doi:10.1109/ July). Multiresolution fusion of multispectral and pan-
TGRS.2002.803623 chromatic images through the curvelet transform. In
Aiazzi, B., Alparone, L., Baronti, S., Garzelli, A., & Selva, M. Proceedings. 2005 IEEE International Geoscience and
(2006). MTF-tailored multiscale fusion of high-resolution Remote Sensing Symposium, 2005, Netherland.
MS and Pan imagery. Photogrammetric Engineering & IGARSS’05. (Vol. 4, pp. 2838–2841). IEEE.
Remote Sensing, 72(5), 591–596. doi:10.14358/ Gavankar, N.L., & Ghosh, S.K. (2018). Automatic building
PERS.72.5.591 footprint extraction from high-resolution satellite image
Alparone, L., Aiazzi, B., Baronti, S., Garzelli, A., & using mathematical morphology. European Journal of
Nencini, F. (2006, July). A new method for MS+ Pan Remote Sensing, 51(1), 182–193. doi:10.1080/
image fusion assessment without reference. In 2006 22797254.2017.1416676
IEEE International Symposium on Geoscience and Gewali, U.B., Monteiro, S.T., & Saber, E. (2018). Machine
Remote Sensing (pp. 3802–3805). IEEE. learning based hyperspectral image analysis: A survey.
Amolins, K., Zhang, Y., & Dare, P. (2007). Wavelet based arXiv Preprint arXiv, 1802.08701.
image fusion techniques—An introduction, review and González-Audícana, M., Saleta, J.L., Catalán, R.G., &
comparison. ISPRS Journal of Photogrammetry and García, R. (2004). Fusion of multispectral and panchro-
Remote Sensing, 62(4), 249–263. doi:10.1016/j. matic images using improved IHS and PCA mergers
isprsjprs.2007.05.009 based on wavelet decomposition. IEEE Transactions on
Ashraf, S., Brabyn, L., & Hicks, B.J. (2013). Alternative Geoscience and Remote Sensing, 42(6), 1291–1299.
solutions for determining the spectral band weights for doi:10.1109/TGRS.2004.825593
the subtractive resolution merge technique. International Gu, Y., Zhang, Y., & Zhang, J. (2008). Integration of spatial–
Journal of Image and Data Fusion, 4(2), 105–125. Spectral information for resolution enhancement in
doi:10.1080/19479832.2011.607473 hyperspectral images. IEEE Transactions on Geoscience
Atkinson, P.M. (2005). Sub-pixel target mapping from and Remote Sensing, 46(5), 1347–1358. doi:10.1109/
soft-classified, remotely sensed imagery. TGRS.2008.917270
Photogrammetric Engineering & Remote Sensing, 71(7), Günlü, A., Ercanlı, İ., Sönmez, T., & Başkent, E.Z. (2014).
839–846. doi:10.14358/PERS.71.7.839 Prediction of some stand parameters using pan-sharpened
Ballester, C., Caselles, V., Igual, L., Verdera, J., & Rougé, B. IKONOS satellite image. European Journal of Remote
(2006). A variational model for P+ XS image fusion. Sensing, 47(1), 329–342. doi:10.5721/EuJRS20144720
International Journal of Computer Vision, 69(1), 43–58. Guo, M., Ma, H., Bao, Y., & Wang, L. (2018). Fusing pan-
doi:10.1007/s11263-006-6852-x chromatic and swir bands based on cnn-a preliminary
Bayarri, M.J., Berger, J.O., Paulo, R., Sacks, J., Cafeo, J.A., study over worldview-3 datasets. International Archives of
Cavendish, J., . . . Tu, J. (2007). A framework for valida- the Photogrammetry, Remote Sensing & Spatial
tion of computer models. Technometrics, 49(2), 138–154. Information Sciences, 42, 3.
doi:10.1198/004017007000000092 Han, X., Yu, J., Luo, J., & Sun, W. (2019). Hyperspectral and
Ben Abbes, A., Bounouh, O., Farah, I.R., de Jong, R., & multispectral image fusion using cluster-based multi-
Martínez, B. (2018). Comparative study of three satellite branch bp neural networks. Remote Sensing, 11(10), 1173.
image time-series decomposition methods for vegetation Hashimoto, N., Murakami, Y., Bautista, P.A.,
change detection. European Journal of Remote Sensing, 51 Yamaguchi, M., Obi, T., Ohyama, N., . . . Kosugi, Y.
(1), 607–615. doi:10.1080/22797254.2018.1465360 (2011). Multispectral image enhancement for effective
Brown, M., Gunn, S.R., & Lewis, H.G. (1999). Support visualization. Optics Express, 19(10), 9315–9329.
vector machines for optimal classification and spectral doi:10.1364/OE.19.009315
unmixing. Ecological Modelling, 120(2–3), 167–179. Hubert, M., & Rousseeuw, P. J., & Vanden Branden, K.
doi:10.1016/S0304-3800(99)00100-3 (2005). ROBPCA: a new approach to robust principal
Chen, X., Zou, D., Zhiying Zhou, S., Zhao, Q., & Tan, P. component analysis. Technometrics, 47(1), 64–79
(2013). Image matting with local and nonlocal smooth Jayanth, J., Kumar, T.A., & Koliwad, S. (2018). Fusion of
priors. In 2013 Proceedings of the IEEE Conference on multispectral and panchromatic data using regionally
Computer Vision and Pattern Recognition, China, (pp. weighted principal component analysis and wavelet.
1902–1907). Current Science, 115(10), 1938. doi:10.18520/cs/v115/
Chibani, Y. (2007). Integration of panchromatic and SAR i10/1938-1942
features into multispectral SPOT images using the ‘a Kaplan, N.H. (2018). Weighted intensity hue saturation
tròus’ wavelet decomposition. International Journal of transform for image enhancement and pansharpening.
Remote Sensing, 28(10), 2295–2307. doi:10.1080/ Turkish Journal of Electrical Engineering and Computer
01431160600606874 Science, 26(1), 204–219. doi:10.3906/elk-1704-43
Dehnavi, S., & Mohammadzadeh, A. (2013). A new devel- Kotwal, K., & Chaudhuri, S. (2010, December). A fast
oped GIHS-BT-SFIM fusion method based on edge and approach for fusion of hyperspectral images through
class data. ISPRS-International Archives of the redundancy elimination. In Proceedings of the seventh
 EUROPEAN JOURNAL OF REMOTE SENSING 93

Indian conference on computer vision, graphics and Image Pohl, C., & Van Genderen, J.L. (1998). Review article multi-
processing (pp. 506–511). ACM. doi:10.1177/ sensor image fusion in remote sensing: Concepts, meth-
1753193409347495 ods and applications. International Journal of Remote
Kotwal, K., & Chaudhuri, S. (2012). An optimization-based Sensing, 19(5), 823–854. doi:10.1080/014311698215748
approach to fusion of hyperspectral images. IEEE Journal Pradhan, P.S., King, R.L., Younan, N.H., & Holcomb, D.
of Selected Topics in Applied Earth Observations and W. (2006). Estimation of the number of decomposi-
Remote Sensing, 5(2), 501–509. doi:10.1109/ tion levels for a wavelet-based multiresolution multi-
JSTARS.2012.2187274 sensor image fusion. IEEE Transactions on Geoscience
Lal, A.M., & Anouncia, S.M. (2016). Enhanced dictionary and Remote Sensing, 44(12), 3674–3686. doi:10.1109/
based sparse representation fusion for multi-temporal TGRS.2006.881758
remote sensing images. European Journal of Remote Qu, J., Lei, J., Li, Y., Dong, W., Zeng, Z., & Chen, D. (2018).
Sensing, 49(1), 317–336. doi:10.5721/EuJRS20164918 Structure tensor-based algorithm for hyperspectral and
Leung, Y., Liu, J., & Zhang, J. (2014). An improved adaptive panchromatic images fusion. Remote Sensing, 10(3), 373.
intensity–Hue–Saturation method for the fusion of remote doi:10.3390/rs10030373
sensing images. IEEE Geoscience and Remote Sensing Rahmani, S., Strait, M., Merkurjev, D., Moeller, M., &
Letters, 11(5), 985–989. doi:10.1109/LGRS.2013.2284282 Wittman, T. (2010). An adaptive IHS pan-sharpening
Levin, A., Rav-Acha, A., & Lischinski, D. (2008). Spectral method. IEEE Geoscience and Remote Sensing Letters, 7
matting. IEEE Transactions on Pattern Analysis and (4), 746–750. doi:10.1109/LGRS.2010.2046715
Machine Intelligence, 30(10), 1699–1712. doi:10.1109/ Rajathurai, A., & Chellakkon, H.S. (2018). Improved visua-
TPAMI.2008.168 lization using a fusion technique based on KNN matting
Li, C., Guo, C., Ren, W., Cong, R., Hou, J., Kwong, S., & of remotely sensed images. Journal of the Indian Society of
Tao, D. (2019). An underwater image enhancement Remote Sensing, 46(2), 179–187. doi:10.1007/s12524-017-
benchmark dataset and beyond. arXiv Preprint arXiv, 0693-7
1901.05495. Raman, S., & Chaudhuri, S. (2007, October). A matte-less,
Mallat, S.G. (1989). A theory for multiresolution signal variational approach to automatic scene compositing. In
decomposition: The wavelet representation. IEEE 2007 IEEE 11th International Conference on Computer
Transactions on Pattern Analysis & Machine Intelligence, Vision (pp. 1–6), Bombay. IEEE.
11(7), 674–693. doi:10.1109/34.192463 Ruescas, A.B., Sobrino, J.A., Julien, Y., Jiménez-Muñoz, J.C.,
Malpica, J.A. (2007). Hue adjustment to IHS pan-sharpened Sòria, G., Hidalgo, V., . . . Mattar, C. (2010). Mapping
IKONOS imagery for vegetation enhancement. IEEE sub-pixel burnt percentage using AVHRR data.
Geoscience and Remote Sensing Letters, 4(1), 27–31. Application to the Alcalaten area in Spain. International
doi:10.1109/LGRS.2006.883523 Journal of Remote Sensing, 31(20), 5315–5330.
Maselli, F., Chiesi, M., & Pieri, M. (2016). A novel approach doi:10.1080/01431160903369592
to produce NDVI image series with enhanced spatial Tiede, D., Baraldi, A., Sudmanns, M., Belgiu, M., & Lang, S.
properties. European Journal of Remote Sensing, 49(1), (2017). Architecture and prototypical implementation of
171–184. doi:10.5721/EuJRS20164910 a semantic querying system for big Earth observation
Md Noor, S., Ren, J., Marshall, S., & Michael, K. (2017). image bases. European Journal of Remote Sensing, 50(1),
Hyperspectral image enhancement and mixture 452–463. doi:10.1080/22797254.2017.1357432
deep-learning classification of corneal epithelium Tu, T.M., Huang, P.S., Hung, C.L., & Chang, C.P. (2004).
injuries. Sensors, 17(11), 2644. doi:10.3390/s17050968 A fast intensity-hue-saturation fusion technique with
Metwalli, M.R., Nasr, A.H., Faragallah, O.S., El-Rabaie, spectral adjustment for IKONOS imagery. IEEE
E.S.M., Abbas, A.M., Alshebeili, S.A., & Abd El- Geoscience and Remote Sensing Letters, 1(4), 309–312.
Samie, F.E. (2014). Efficient pan-sharpening of satellite doi:10.1109/LGRS.2004.834804
images with the contourlet transform. International Villa, A., Chanussot, J., Benediktsson, J.A., & Jutten, C.
Journal of Remote Sensing, 35(5), 1979–2002. (2011). Spectral unmixing for the classification of hyper-
doi:10.1080/01431161.2013.873832 spectral images at a finer spatial resolution. IEEE Journal
Miao, Q., & Wang, B. (2006, April). The contourlet trans- of Selected Topics in Signal Processing, 5(3), 521–533.
form for image fusion. In Multisensor, Multisource doi:10.1109/JSTSP.2010.2096798
Information Fusion: Architectures, Algorithms, and Vrabel, J. (2000). Multispectral imagery advanced band
Applications 2006 (Vol. 6242, p. 62420Z). International sharpening study. Photogrammetric Engineering and
Society for Optics and Photonics, Florida. Remote Sensing, 66(1), 73–80.
Mozgovoy, D.K., Hnatushenko, V.V., & Vasyliev, V.V. Wang, H., & Suter, D. (2007). A consensus-based method
(2018). Automated recognition of vegetation and water for tracking: Modelling background scenario and fore-
bodies on the territory of megacities in satellite images of ground appearance. Pattern Recognition, 40(3),
visible and ir bands. ISPRS Annals of Photogrammetry, 1091–1105. doi:10.1016/j.patcog.2006.05.024
Remote Sensing & Spatial Information Sciences, 4, 3. Wang, Q., Jia, Z., Qin, X., Yang, J., & Hu, Y. (2011). A new
Nunez, J., Otazu, X., Fors, O., Prades, A., Pala, V., & technique for multispectral and panchromatic image
Arbiol, R. (1999). Multiresolution-based image fusion fusion. Procedia Engineering, 24, 182–186. doi:10.1016/j.
with additive wavelet decomposition. IEEE Transactions proeng.2011.11.2623
on Geoscience and Remote Sensing, 37(3), 1204–1211. Wang, Y.X., & Zhang, Y.J. (2013). Nonnegative matrix fac-
doi:10.1109/36.763274 torization: A comprehensive review. IEEE Transactions
Parveen, R., Kulkarni, S., & Mytri, V.D. (2018). Automated on Knowledge and Data Engineering, 25(6), 1336–1353.
extraction and discrimination of open land areas from doi:10.1109/TKDE.2012.51
IRS-1C LISS III imagery. International Journal of Wenyan, Z., Zhenhong, J., Yu, Y., Yang, J., & Kasabov, N.
Computers and Applications, 1–10. doi:10.1080/ (2018). SAR image change detection based on equal
1206212X.2018.1558937 weight image fusion and adaptive threshold in the NSST
94 R. ABLIN ET AL.

domain. European Journal of Remote Sensing, 51(1), Xu, Q., Zhang, Y., Li, B., & Ding, L. (2015).
785–794. doi:10.1080/22797254.2018.1491804 Pansharpening using regression of classified MS and
Xiao-Hui, Y.A.N.G., & Li-Cheng, J.I.A.O. (2008). Fusion pan images to reduce color distortion. IEEE Geoscience
algorithm for remote sensing images based on nonsub- and Remote Sensing Letters, 12(1), 28–32. doi:10.1109/
sampled contourlet transform. Acta Automatica Sinica, LGRS.2014.2324817
34(3), 274–281. doi:10.3724/SP.J.1004.2008.00274 Yadav, P., & Agrawal, S. (2018). Road network identification
Xie, Q., Zhou, M., Zhao, Q., Meng, D., Zuo, W., & Xu, Z. and extraction in satellite imagery using otsu’s method
(2019). Multispectral and hyperspectral image fusion by and connected component analysis. International
MS/HS fusion net. In Proceedings of the IEEE Archives of the Photogrammetry, Remote Sensing &
Conference on Computer Vision and Pattern Spatial Information Sciences, XLII-5, 91–98. doi:10.5194/
Recognition (pp. 1585–1594), China. isprs-archives-XLII-5-91-2018
Xu, N., Price, B., Cohen, S., & Huang, T. (2017). Deep image Yocky, D.A. (1996). Multiresolution wavelet decomposition
matting. In Proceedings of the IEEE Conference on I me merger of landsat thematic mapper and SPOT pan-
Computer Vision and Pattern Recognition (pp. 2970–2979), chromatic data. Photogrammetric Engineering & Remote
China. Sensing, 62(9), 1067–1074.
Xu, Q., Zhang, Y., & Li, B. (2014). Recent advances in Zhang, Y., & Mishra, R.K. (2014). From UNB PanSharp to
pansharpening and key problems in applications. Fuze Go–The success behind the pan-sharpening algo-
International Journal of Image and Data Fusion, 5(3), rithm. International Journal of Image and Data Fusion, 5
175–195. doi:10.1080/19479832.2014.889227 (1), 39–53. doi:10.1080/19479832.2013.848475