Article in HTML

Author(s): Sushant Bindra1, Mehak Piplani2



    1Department of Computer Science, Manipal University Jaipur, 303007 India.
    2Department of Computer Science, University of Southern California, USA.
    *Corresponding Author: Sushant Bindra

Published In:   Volume - 1,      Issue - 1,     Year - 2021

Cite this article:
Sushant Bindra and Mehak Piplani (2021). Image Fusion using Wavelet Transform: Systematic Literature Review. Spectrum of Emerging Sciences, 1(1), pp.9-15.

  View PDF

Please allow Pop-Up for this website to view PDF file.

Image Fusion using Wavelet Transform: Systematic Literature Review

Sushant Bindraa*, Mehak Piplanib

a*Department of Computer Science, Manipal University Jaipur, 303007 India.

b Department of Computer Science, University of Southern California, USA.                                                                                                                                                                 


*Corresponding Author: Department of Computer Science, Manipal University Jaipur, 303007.

E-mail Address: (Sushant Bindra)

Article available online at:





Original Research Article

Received: 18 July, 2021

Accepted: 2 August, 2021




Image fusion;

Wavelet Transform;

Application; Comparative study



The genesis of an Image fusion is intermingling multiple image of usual features to frame a solitary image which attains all the indispensable characteristics of image. At the moment a great deal of effort is going to be executed on the field of image fusion and used in various applications well like medical imaging and multi spectra sensor image fusing etc. For fusing the image numerous techniques have been proposed like wavelet transform. In present review article the image fusion along with wavelet transform has been discussed with its advantages and disadvantages along with defined parameters like entropy, mutual information, cross entropy, Fusion similarity metric (FSM), etc. It's tough to say which strategy is ideal for a particular application. For the image fusion application, all of the approaches were determined to be reliable.


1.        Introduction

Picture mix is a synergistic instrument that serves to unite distinctive source imagery. The musing is to get two photos of a comparative article under two unmistakable acquisition conditions, and to arrange these two pictures to get a more exact arrive at assessment of sign levels. Picture blend of different sensors in a fantasy system could through and through diminish human/machine screw up in area and affirmation of articles by virtue of the natural overabundance and extended consideration. disillusionment(1).

The essential mark of Image mix (IF) is gathering necessary, similarly as peaceful monotonous information from various pictures to make a merged picture, to giving more complete and exact portrayal. In the space of clinical imaging, joining of different technique pictures of same scene gives such endless advantages(2). It very well may be blend of picture taken at different objective, power and by different systems helps specialist/Radiologists to easily separate or recognize the features or anomalies that may not be consistently clear in single picture. Another advantage of picture mix is that it diminishes the limit cost by taking care of simply the single merged picture, as opposed to the different philosophy pictures(3). This review paper presents execution a bit of the image mix techniques using preliminary data. The image blend strategies used in wavelet change.

2.        Image Fusion Algorithm

Various computations have been made for picture blend to improve the trustworthiness and the show of testing(4). Picture blend methodology can be disengaged into two social affairs: Spatial region mix procedure and Transform space mix. Spatial region blend methodology clearly oversees pixels of data pictures. 

2.1 Wavelet Schemes Wavelet based Methods

 Wavelet methodologies are also a way to deal with separate picture into confined scale unequivocal signs. Wavelet changes are immediate and square vital changes whose premise limits are called wavelets(4). Discrete Wavelet Transform - In the standard wavelet-based mix once the imagery is broken down through wavelet change a composite multi-scale depiction is worked by a decision of the striking wavelet coefficients. The decision can be established on picking the restriction of the incomparable characteristics or a region based most noteworthy energy. The last stage is an opposite discrete wavelet change on the composite wavelet depiction.

3.        Image Fusion Method

Picture mix is a synergistic instrument that serves to combine different source imagery. This assessment is orchestrated to make picture mix techniques for pictures gained with single and different modalities. There are every now and again a couple of issues that should be overseen before the blend can be performed. A huge segment of the photos from various source are slanted. Picture enlistment is routinely used as a groundwork advance in picture mix(5). In this paper, Image Fusion graphical UI is made and execution occasions of mix methods in a part of the applications are presented. The mix results show that the mix reduces the obscurity and improves the constancy of flaw distinguishing proof in both visual and emotional appraisals. The results moreover show that image mix gives an effective system to engage assessment and examination of such data.

There are numerous methods that have been developed to perform image fusion(6). Some well-known image fusion methods are listed below:

·     Intensity-tone immersion (IHS) change based combination. 

·     Principal part investigation (PCA) based combination.

·     Multi scale change-based combination.

·     High-pass separating technique.

·     Pyramid strategy.

·     Wavelet changes.

·     Intensity-tone immersion (IHS) change based combination. 

·     Principal part investigation (PCA) based combination.

·     Multi scale change-based combination.

·     High-pass separating technique.

·     Pyramid strategy. 

·     Wavelet changes.


4.        Wavelet Transform:

This wavelet change of picture planning on various repeat channels and this source picture is the first multi-wavelet rot, this amount of sub-picture and the change space, incorporate assurance, making the merged picture in conclusion joined picture amusement by the change. Of late, wavelet change has pulled in coherent thought, it not simply in math has outlined another branch, is ideal mix of utilitarian examination, Fourier assessment, numerical assessment, yet furthermore in planning applications, similar to sign taking care of, picture getting ready, plan affirmation, talk affirmation and mix similarly as various nonlinear science, have a huge effect. Wavelet examination is another advancement of the time scale assessment and the multiresolution examination, possesses incredible limited characteristics in both the time territory and repeat region. A result of the gradually fine reality tread on the high repeat, this can focus in on assessment of optional nuances such as, this brand name is aiming wavelet change the properties of a wavelet change(7). it was poured as a mathematical amplifying focal point. The wavelet weakening of this image has been multi-scale, multi-objective, crumbling of the image, since wavelet is not overabundant(8). Hence the image data after wavelet rot through complete will not augmentation, at the same moment wavelet disintegration owns heading, using this specific brand name may for the characteristic eye to various orientation of this incredible repeat sections with distinct objective of the visual features, the joined picture has been improved outcome in picture blend. This specific design is the MATLAB image employing wavelet transform based fusion, that can realize the processing of various kind of pictures. Particular requirements include:

·     Freedom to select different format picture processing.

·     Can be a variety of associated image processing.

·     Can be fused in a variety of ways.

·     Can be at the fusion of images for this save operation.

·     The whole process has the benefits of simple operation, remarkable man-machine interface.

Comparative study focuses on comparing:

a.    Standard deviation (σ)

b.   Entropy (H)

c.    Spatial frequency (SF), 

d.   Fusion mutual information (FMI), 

e.    Fusion quality index (FQI), 

f.    Fusion similarity metric (FSM), etc.

Image fusion parameters with reference images:

a.    Peak signal to noise ratio (PSNR),  

a.    Correlation coefficient (CC), 

b.   Mutual information (MI), 

c.    Universal quality index (UQI), 

d.   Structural similarity index measure (SSIM)


In view of three boundaries to be specific edge strength, combination factor and combination evenness. The Entropy and combination quality list share the way that they can be viably utilized for intertwining multi see pictures. THP then again is more qualified for multi-center pictures.

5.        Related Work 

Ismail et al. 2017, studied image blend is a broadly discussed subject for improving the information substance of pictures. The essential objective of picture mix computation is to join information from various photos of a scene. The eventual outcome of picture blend is another image which is more workable for human and machine insight for extra picture planning errands like division, feature extraction and article affirmation. This paper explores the opportunity of using the specific wavelet approach in picture blend and denoising. These estimations are considered on cutting edge amplifying instrument pictures. The procedure uses a general change-based picture enlistment followed by wavelet mix. By then the least squares support vector machine based repeat band decision for picture de noising can be combined to diminish the arti real factors. The spaces are to intensify objective, decrease arti real factors and darkening in the last super picture. To accelerate the entire errands, it is proposed to offload the image taking care of estimations to a gear stage as such the presentation can be improved. FPGAs give an astounding stage in executing steady picture dealing with applications, since natural parallelism of the designing can be abused unequivocally. Picture taking care of endeavors executed on FPGAs can be up to 2 critical degrees faster than a similar application on a comprehensively valuable PC(9).

D. Hemasree et al. 2019, An Image blend is the headway of amalgamating in any event two image of essential brand name to outline a lone picture which secures all of the crucial features of interesting picture. Currently lot of work will be executed on the area of picture mix and moreover used in numerous applications like clinical imaging, multi spectra sensor picture entwining, etc. For interweaving the image, a variety of systems have been proposed by distinct makers, for instance, the wavelet change, IHS and PCA put together methods thus with respect to in this paper composing of an image blend with the wavelet change has been inspected with advantages and blames(10).

Mahyoub et al. 2019, reported about the image blend subject to the wavelet change and examination of picture mix major head, methodology and benefit. The essential objective of the picture blend is to solidify an information received from various images of a comparable picture reliant upon a particular estimation, the delayed consequence of picture mix is another consequent that can be more appropriate for human as well as machine. This present day's image blend advancement has been for the most part applied in various fields including distant distinguishing, automata affirmation, PC vision, clinical picture dealing with. This report designs and comprehends the strategy for picture computation which relies upon wavelet change(11). 

Fajaryati et al. 2020, explained multi-focus picture mix infers merging an absolutely clear picture with a lot of photos of a comparable arena and under comparative imaging circumstances along with varied focus centers. To attain a sensible picture which contains entire relevant things around there, the multi-focus picture blend computation is proposed subject to wavelet change. First thing, the multi-centered pictures had been crumbled because of wavelet change. Likewise, these wavelet coefficients of an approximant as well as comprehensive sub-pictures have been merged independently subject to the blend policy. Ultimately, the interlaced picture had been gotten by utilizing the contrary wavelet change. Among these, for the low-repeat as well as high-repeat coefficients, we introduce a blend rule subject to the weighted extents and these weighted points with the enriched edge disclosure chairman. These preliminary outcomes address that proposed computation has been amazing to hold the ordered pictures(12). 

The general need of an image merging measure is to save all considerable and important information from the source pictures, while at the same time it should not present any bowing in resultant interlaced picture. Execution measures are used significant for measure the expected benefits of blend and besides used to differentiate results got and different estimations.  Standard deviation: It portrays the degree of dissipating between the value of each pixel and the ordinary worth of picture. When in doubt, the more conspicuous the standard deviation regard, the more dispersive the transport of overall greyscale will be, the more imperative picture contrast it will present.

The acclaimed creator of information theory, Shannon, suggested that the possibility of Entropy can address how much information is contained in signals. It's in like manner comprehensively used to show the ordinary proportion of information of pictures in picture getting ready field. For an image, grayscale worth of every pixel can be considered as shared self-sufficient.

Vidhya et al. 2019, studied the combination of pictures is the way toward consolidating at least two pictures into a solitary picture holding significant highlights from each. Combination is a significant procedure inside numerous dissimilar fields like far off detecting, advanced mechanics and clinical applications. Wavelet based combination procedures have been sensibly powerful in joining perceptually significant picture highlights. Shift invariance of the wavelet change is significant in guaranteeing vigorous sub-band combination. In this manner, the unusual use of the shift invariant as well as directionally particular Dual Tree Complex Wavelet Transform (DT-CWT) to picture combination has been presently presented (13). 

Yadav et al. 2014, In different applications, picture combination assumes a significant part. Picture combination is only consolidating at least two pictures into a solitary picture by removing significant highlights from every one of the pictures. The combination of pictures is regularly needed to meld pictures that are caught from instrument. Complex Wavelet based combination strategies have been utilized in consolidating perceptually significant highlights. A tale picture combination method dependent on double tree complex wavelet change is introduced in this paper. Double tree CWT is an augmentation to discrete wavelet change (DWT). Our methodology depends on an inclination space strategy that jam significant neighborhood perceptual highlights which evades numerous issues, for example, ghosting, associating and haloing (14).   

6.        Information Measure for Performance ofImage Fusion on Wavelet Transform

 Picture combination plans to blend at least two pictures to create another picture that is superior to the first ones. A picture combination framework takes as an info at least two source pictures and delivers one melded picture as a yield(15). Picture combination execution measures rely basically upon assessing the measure of data moved from both source pictures into the subsequent melded picture (16). 

The different picture combination boundaries reference picture for estimation performs and measure investigation based on MATLAB examination given underneath: 

Common data: Mutual data estimates the stretch between the joint factual dispersions for two irregular factors i.e., X and Y from the case on the off chance that they are absolutely free. It utilizes cross entropy in between the joint conveyance X Y and the prime case appropriation of being completely independent random variables as follows:


It is symmetric and reaches zero if X and Y are totally independent where p XY (x, y) ¼ p X(x) p Y(y) which leads to



7.        Problem with Image Fusion

Mutual information measure to estimate the joint information between source images  and the fused image  as follows Figure 1 and 2:


Where,  the entropy of variable image fusion



Where H(X), H (Y) are the entropies of X and Y, respectively.


Where H(X), H (Y) and H(F) are the entropies of X, Y and F, respectively.

Figure 1. Image right side and left side.

The results show that MI records fundamentally higher contrasts as H(X)2H (Y) increments. 

·     Classic common data (MI) is one-sided in relation to the source picture with the most noteworthy entropy (17). The test runs 3 picture combination calculations on enrolled source pictures highlighting visual and infrared data (18).  


Figure 2. Fused image of combined both images.

That exemplary shared data (MI) is one-sided towards the source picture with the most elevated entropy.  

·     The results unveil that MI records fundamentally higher contrasts as H(X)- H (Y) increments  

·     Results show how the mistake between utilizing exemplary shared data and the standardized form increments as the distinction between entropies of source pictures increment (19).


This design is the MATLAB image using wavelet transform based fusion, which can realize the processing of a variety of pictures (20).  

Specific requirements include:

Freedom to choose various format picture processing:

·     Can be a variety of related image processing  

·     Can be fused in different ways          

·     Can be on the fusion of images for the save operation  

·     The whole process has the advantages of simple operation, good man-machine interface



8.        Conclusion

It can be observed that the fusion rule described in the research can supply more picture information while also producing crisper fusion images after being evaluated using entropy, peak signal-to-noise ratio, and average gradient.

Table 1 Evaluated various result.

Fusin Rules

Information Entropy

Mutual Information



Average Gradate

Peak Signal to Noise Ratio

Running Tie (S)



Low frequency













High frequency




























Proposed fusion algorithm












  1.  Staniforth S, Wallert A, Hermens E, Peek M, Hermens E. Historical Painting Techniques, Materials, and Studio Practice. Studies in Conservation. 1999;44(2):142.
  2. Bai X, Zhang Y, Zhou F, Xue B. Quadtree-based multi-focus image fusion using a weighted focus-measure. Information Fusion. 2015 Mar 1;22:105–18.
  3. Cai J-H, Hu W-W. Feature Extraction of Gear Fault Signal Based on Sobel Operator and WHT. Shock and Vibration. 2013;20(3):551–9.
  4. Chen W, Hu M, Zhou L, Gu H, Zhang X. Fusion Algorithm of Multi-focus Images with Weighted Ratios and Weighted Gradient Based on Wavelet Transform. Journal of Intelligent Systems. 2019 Oct 1;28(4):505–16.
  5. A SAR image compression algorithm based on Mallat tower-type wavelet decomposition [Internet]. [cited 2021 Aug 31]. Available from:
  6. Yazid K, Zin MRM, Sayuti SB. Development of image fusion techniques :11.
  7.  Liu Y, Jin J, Wang Q, Shen Y, Dong X. Region level based multi-focus image fusion using quaternion wavelet and normalized cut. Signal Processing. 2014 Apr 1;97:9–30.
  8. Multi-focus image fusion using HOSVD and edge intensity | Journal of Visual Communication and Image Representation [Internet]. [cited 2021 Aug 31]. Available from:
  9.  Ismail MP bin, Sani S bin, Masenwat NA bin, Mohd S, Sayuti S, Ahmad MRB, et al. Radiation attenuation on labyrinth design bunker using Iridium-192 source. AIP Conference Proceedings. 2017 Jan 6;1799(1):050011.
  10.  Hemasree D, Reddy DSN, Rajeswari VR. Fusion of Panchromatic and Multispectral Image using PCA and Wavelet Transform. International Journal of Engineering Research & Technology
  11. Mahyoub S, Fadil A, Mansour EM, Rhinane H, Al-Nahmi F. Fusing of Optical and Synthetic  Aperture (SAR) Remote Sensing Data: A systematic literature review. Int Arch Photogramm Remote Sens Spatial Inf Sci. 2019 Feb 21;XLII-4/W12:127–38.
  12. Fajaryati N, Budiyono, Akhyar M, Wiranto. The Employability Skills Needed To Face the Demands of Work in the Future: Systematic Literature Reviews. Open Engineering. 2020 Jan 1;10(1):595–603.
  13. Patil DrS. An Efficient MRI Brain Image Registration and Wavelet Based Fusion. International Journal of Recent Technology and Engineering. 2019 Nov 1;8:10209–18.
  14. Yadav N, Shrivastri S. A Review Of Image Fusion Using IHS With Wavelet Transform.
  15. Ramesh C, Ranjith T. Fusion performance measures and a lifting wavelet transform based algorithm for image fusion. In: Proceedings of the Fifth International Conference on Information Fusion FUSION 2002 (IEEE CatNo02EX5997). 2002. p. 317–20 vol.1.
  16. Abd-el-kader A, El-Din Moustafa H, Rehan S. Performance measures for image fusion based on wavelet transform and curvelet transform. In: 2011 28th National Radio Science Conference (NRSC). 2011. p. 1–7.
  17.  Morabito FC, Simone G, Cacciola M. 15 - Image fusion techniques for non-destructive testing and remote sensing applications. In: Stathaki T, editor. Image Fusion. Oxford: Academic Press; 2008. p. 367–92. Available from:
  18. Mitianoudis N, Stathaki T. 12 - Enhancement of multiple sensor images using joint image fusion and blind restoration. In: Stathaki T, editor. Image Fusion [Internet]. Oxford: Academic Press; 2008. p. 299–326. Available from:
  19.   Silverman E. Convolutional Neural Networks for Cellular Automata Classification. In MIT Press; 2019 [cited 2021 Sep 1]. p. 280–1. Available from:
  20. Anoop Suraj A, Francis M, Kavya TS, Nirmal TM. Discrete wavelet transform based image fusion and de-noising in FPGA. Journal of Electrical Systems and Information Technology. 2014 May 1;1(1):72–81.

Related Images:

Recomonded Articles:

Author(s): Binod Shrestha, Sambridhi Shah, Khagendra Chapain, Rajendra Joshi, Rajesh Pandit

DOI: 10.55878/SES2022-2-1-7         Access: Open Access Read More

Author(s): Sushant Bindra; Mehak Piplani

DOI:         Access: Open Access Read More

Author(s): Anand Pathak; Saurabh Deshmukh

DOI:         Access: Open Access Read More

Author(s): Reena Rawat

DOI:         Access: Open Access Read More

Author(s): Bhupendra Kande, Prachi Parmar

DOI: 10.55878/SES2022-2-1-3         Access: Open Access Read More

Author(s): Vania Munjar

DOI: 10.55878/SES2021-1-1-12         Access: Open Access Read More

Author(s): Shubhangi Jha, Pragya Kulkarni and Anamika Sharma

DOI: 10.55878/SES2022-2-2-3         Access: Open Access Read More

Author(s): Roli Jain

DOI: 10.55878/SES2022-2-1-6         Access: Open Access Read More

Author(s): Shathya Pranav Sujithra Rajesh Kannan

DOI: 10.55878/SES2022-2-2-1         Access: Open Access Read More

Author(s): Manish Kumar, Keshav Shishodiya, Yogendra Singh Rajawat, Seema Nayak

DOI: 10.55878/SES2023-3-1-6         Access: Open Access Read More

Author(s): Madhav Kapoor

DOI: 10.55878/SES2022-2-1-5         Access: Open Access Read More