Technium Vol. 9, pp.

41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

Image Morphing Techniques: A Review

Alyaa Qusay Aloraibi

Software Department, College of Computer Science and Mathematics,

University of Mosul, Mosul. Iraq.

E-mail: [email protected]
Abstract
Nowadays image morphing has become one of the important techniques in applications that
require a graphical representation of objects. Morphing tools have become very well known among
users who work on multimedia applications such as art effects, virtual games, photo morphing, and
social media, in addition to scientific and academic fields. There are many algorithms to apply
morphing operations, including the basic and improved techniques, which share some essential
stages, but vary in the algorithm details and the produced image qualities. Morphing techniques, in
general, are based on image features and changing them through a warping process to produce
another image or mixing two images to produce a new combined image. This paper provides an
overview of different morphing techniques explaining how they work and discuss their features in
some terms such as the morph visual quality, technical efficiency, and complexity, which can assist
the researcher in the image morphing field to compare and identify morphing techniques that suit
their working area.
Keywords: Metamorphoses, Image transformation, Warp generation, Feature specification, image
blending.

1. Introduction
Image morphing has become one of the most popular techniques for image transformation that is
used in many fields, especially on multimedia and visual effects including image retransforming and photo
editing tools, this technique is also efficient for graphics, animation, and game design that attract users
whether for computer graphics or mobile applications. Morphing technique for image processing which is
also known as “metamorphosis”, has proven to be a powerful tool for visual effects [1] [2]. Morphing
operation is defined as, a process that changes one image into another through a seamless transition [3].
Appling image metamorphic on two images starts first with the animator to establish the correspondence
between the pairs of features primitive, such as curves, mesh nodes, line segments, and points [1] [2].
Morphing operations include image processing approaches such as warping and cross-dissolving.
[4]. Image Morphing technique is based on mixing the pixels of two images to merge a picture of two
persons (for example), the first image (source) will be transformed into the target image by selecting sets of
points or features of the images, so the result will also be an image but with different features and
characteristics or a new image. Where warping determines how pixels from one image correlate with
corresponding pixels from the other image [4]. The corresponding feature sets are then used to produce a
mapping function, which is the (warp function) [3]. The morphing algorithm's performance depends on
selecting a point from the images to be mixed according to some mathematical equations related to the
applied algorithm. So mathematical morphology is defined as “a field that studies the topological and
structural properties of objects based on their images” [5]. During performing warping and morphing
operations the steps of transforming the original image into another shape can be presented as a sequence
of figures (animation) where the user can see the process of the transformation operation.

41
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

Morphing techniques can be applied to a complete image or by selecting some points or apart from
a picture, selecting human faces from a picture, to apply a morphing operation became one of the important
fields in image manipulation that is called (face morphing), which aims to create a new face image that
resembles the biometric information of two or more face images [6]. This technique became an important
tool for real-time facial animation and 3D video games [7]. Moreover, image-morphing techniques can be
used to reduce memory consumption and preserve the perceptual quality of animations as in [8]. Although
image morphing technique is very famous in entertainment applications, on the other side it can be used to
produce fake images which cause a security issue, especially for face recognition systems thus, face
morphing attacks have received great interest from the biometric community [9], which become an
important topic nowadays, that aims to study morphing techniques and detect morphing attack as in [10]
[11] such studies are very important recently due to the rapid development along with quality improvement
of morphing algorithms that make it difficult to distinguish between origin and the transformed
(manipulated) image.
Due to the importance of morphing algorithms’ effectuation in many applications. This research aims
to present an overview of classic and modified techniques related to morphing algorithms, which will be
listed in section three.

2. Morphing algorithms principle

The principle of Image morphing is transforming an image into another image. The transforming
process is widely used in motion image applications and movies that need special effects. Where the work
on image transformation started in the early 1980s [12]. The operation can be linear which means
transforming linear images or morphing an image by segmenting the image into lines and then transforming
them to create the result according to the target image [13].
Prior to the development of morphing techniques, image transformation was commonly carried out
by cross-dissolve techniques e.g., linear interpolation to fade from one image to another [1] [2]. That’s
wherever one image is light out and an alternative image is light in, therefor such a technique wasn't effective
in signifying the metamorphosis [14]. These approaches are generally invariant to the semantics of the
underlying object making them vulnerable to artifacts such as “ghosting” and implausible intermediates
[15]. Therefore, morphing operations were developed to achieve fluid transformation by using warping to
preserve geometric alignment by the cross-dissolve operation [1] [2].
Image warping is defined as “the process of digitally manipulating an image such that any shapes
portrayed in the image have been significantly distorted” [16]. So, morphing techniques became a
combination of two operations, warping i.e. (shape geometric) and cross dissolve i.e. (color photometric)
[13]. Where the warp is specified by a mapping between lines in the first and second images [16]. Warping
techniques are very important to reform the images, using inefficient warping operations or pixel mapping
will produce misfit morphed images. Although, there are many morphing techniques the difference in these
techniques is mainly in how to apply warping [17]. Another important step in the morphing operation is
selecting lines or points ...etc. of the image (features). Feature specification is the most tedious feature of
morph operation [1].
In following figure 1, illustrates morphing and cross-dissolving images between origin images (a)
and (b), where figure (c) shows the intermediate result of the cross-dissolve operation on images (a) and (b),
while image (d) in the figure shows a frame of morphing sequence [5].

42
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

Figure 1. Example of morphing operation of origin images (a and b), the result of the cross-dissolve
operation in c, and a frame of the morphed image in d [5].

3. Morphing and warping techniques

Pioneering morphing algorithms combine corresponding driven bi-directional warping using blend
operations to generate a sequence of images despite the entities in play [15]. Pixel interpolation is a very
important step to generate the morphed image. This process can be applied to the entire two images, or by
selecting some features from them, the morphing process starts with the first image then blend it into the
second image and fades. Morphing techniques are in continuous development to produce a good result, the
techniques differ in the details but still share the basic steps, in general, three main processes are required
for morphing operation: feature specification, warp generation, and transition control [4].
This section reviews 9 warping and morphing techniques, including the basic methods which are:
(Cross-Dissolving method, mesh warping, field morphing, triangulation-based morphing, energy
minimization, multilevel free-form deformation, and work minimization, besides methods for 3D faces
morphing and morphing using structural similarity). The review shows the progress in image morphing
techniques with their specifications and modification. The following subsections discuss these techniques.

3.1 Cross-Dissolving Method

Years ago, when morphing was not developed yet, image transformation operation was achieved
with the help of the cross-dissolve technique [3]. This operator makes a transformation on two images
without considering the transition between the different objects contained in the image [5]. The concept of
cross-dissolving operation is that an image will fade to another image using linear interpolation [4]. In this
process, the color of each pixel is interrelated over time from the first image value to the second image's
corresponding value [16]. An example of a cross-dissolve operation on two pictures (f and g), is crossing
dissolve picture (f) to (g) means taking a sequence of images [5].
The cross-dissolve method can be considered a primitive method of morphing operation, and suffer
from some visual artifact that appears on the resulting morphed image. One of this method’s problems is
that the image at halfway through the morphing process, maybe seemed to be a “ghost” effect or a (gloomy
zigzag) image [3] [13], also the outcomes of the crossing-dissolve approach are poor, having a double-
exposure effect apparent in misaligned regions [2] [3]. Besides when using face image morphing the double
exposure result within the eyes and mouth areas is seen and the morph doesn't look natural [14]. In order to
avoid these issues, warping is used to align the two images before cross-dissolving [12]. A study presented

43
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

in [17] proved that the result of a morphed image is very bad in the case of applying only cross-dissolving
operations without warping, the reason for that is, the input and the destination pictures are not aligning.
Because the warping algorithm keeps the geometric alignment of the areas by the cross-dissolving operation,
so the morph operation will achieve a fluid transformation between the destination and the source image
[3]. Figure 2, presents the progress of cross dissolving operation on five frames, it is clear that in the middle
step, there is double-exposure effect also appears, that the input images will contribute equally to the output
[1] [2]. From the figure, it can be noticed Image overlapping start from the second frame, with blurring and
ghost artifact, especially on the eyes and the hair.

Figure 2. Cross-dissolve sequence over five frames [1].

3.2 Mesh Warping
Mesh warping algorithm is one of the most interesting approaches for morphing operation, it is
considered one of the basic and traditional techniques. The technique pioneered at Industrial Light & Magic
ILM) by Douglas Smythe which is used in a movie named Willow in 1988 and later used successfully in
many subsequent motion pictures [1] [2]. The mesh warping concept works by breaking pictures into little
areas and mapping the pixel from the source to the objective picture. So, no specter lines appear in the
picture [18]. This warping depends on how the mesh is implemented, such as using a rectangular grid of
mesh. To apply it, the edge must coincide to the features of the image without self-intersection between the
grids [19]. Figure 3, present an explained where the mesh morphing algorithm is applied on a source and
destination image at the left and right of the figure, the middle image represents a metamorphosis (morph)
between the two faces at the left and right of the figure [12].

Figure 3. Example of mesh morphing [12].

G.Wolberg described in detail the two-pass mesh warping algorithm in [1] [2]. This is summed up
as follows: to apply the mesh algorithm on two images (source and target) their coordinates of control points,
must be specified at the source and target image, to associate the source mesh (Ms) with the target mesh
(MT), which are used to define a spatial transformation that maps all points in the source and target image.
[1][2]. Using mesh tables for the source picture is mapped to a mesh table for the destination picture, then
for each mesh, a calculation is applied, that could be easily computed [20].
Although mesh warping is one of the predominant approaches, it suffers from some problems
including ghosting [21]. besides when the grid size increases the computations will increase also [12].

44
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

Therefore, many researchers modified traditional morphing algorithms to provide efficient results. N. Fish
et al., presented a new approach for the morphing effect that aims to leverage the power of deep neural
networks to learn a shape prior to fitting a given source dataset [15]. Deepalakshmi in [14] developed an
algorithm where polygon meshes square deformation technique is employed, mesh deformation breaks
pictures into a little region and maps pixel to pixel from supply to focus on the image, according to the
author, the technique produced the best visual result due to implement mesh-based image morphing [14].
Another issue in this approach is the control points, which are represented by the line intersection
of the grid, may not be enough in certain regions if needed [21]. Thus, the same resource proposed
improvement of mesh morphing to overcome the problem of control point misalignment that causes the
ghosting effect, by indicating the coordinates of control points [21].

3.3 Field Morphing

The field morph approach was first presented by Beier and Neely [22]. This technique is proposed to
improve mesh warping performance, where image features are considered instead of depending on mesh
nodes. The proposed technique leverage user-defined line segments for the establishment of the
corresponding feature point that is used to distort each endpoint toward the other [15]. The pairs of lines
characterize mapping from one picture to the next where one line can characterize the connection to the goal
picture [18] then warping operation can be applied according to these lines that represent for example the
nose or mouth in the image. According to Beier and Neely’s paper, the simplest scenario of the algorithm is
that the animator must specify the control points of the image and draw lines on it. Then the animator must
take the given points and place them at the correct locations. While the points that are accidentally left, or
points that the animator could not find an associating feature will still be used by the warping algorithm
[22]. The strategy of this approach depends on pixel distance concerning the lines that are drawn on the
origin and the destination [12]. Where the displacement of a point in the original image is a weighted sum
of maps resulting from each line pair, with the weights attributed to distance and line length [1]. Figure 4,
presents an example of drawing lines over a face for a field morphing algorithm with the resulting image as
can be noticed the result is very good and the two images morphed without a ghost or blurring effect [22].

Figure 4. Drawing lines over a face for field morphing algorithm [22].

Feature morphing techniques produce good results for orientation images [17]. Although this
morphing algorithm produces good results, its disadvantage was that all the controls are given to the
animator. Thus, the technique was modified later, the future scope of this technique is automatic feature
selection for morphing [12]. Beier’s algorithm was modified to speed it up, by Hussain Karam et al.
presented in [23] proposed a modified feature-based by applying an endpoints algorithm which modified
Beier’s method for lines features transformation, where just feature line endpoints are morphed and the
remained points are transformed according to these endpoints [3]. Jya-Kai Chang et al. in [24] used field
morphing for face morphing to develop a simple system for the user, that can synthesize facial features, then
ideally create a sequence of fluid transformations between two pictures, the method enables users to draw
features directly and contour lines on the source and destination image. Many animators used both mesh

45
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

and field morphing algorithms to create a good visual effect [16] [24]. Jya-Kai Chang et el. compared this
method with mesh morphing and concluded that field morphing spends most of the time creating more
feature lines for warping, but its performance is better than mesh morphing [24]. Another study in [25].
Presented the limitation of mesh morphing is that quality result is only good if the source and the target
image have a similar shape, while field morphing is simple but tedious. It is obvious that merging morphing
techniques can improve their performance as in [26] proposed hybrid image morphing method by applying
feature-based and mesh warping algorithms, combined with cross-dissolving methods. the author explained
that the proposed algorithm presented an efficient morphed outcome by finding control of the grid and
mapping the related pixels of the source image to the goal image.

3.4 Triangle Mesh Warping

Selecting the correct colors, shape, features, shape points, and information of an image is very
important to create a perfect morphing operation. Therefore, the work continues to improve warping
operation with efficient point selection. Thus, a development on mesh warping is proposed by dividing the
image into groups of triangular, Delaunay triangulation is a popular method for optimum triangulation of a
set of points [20]. Delaunay’s triangulation feature maximizes the minimum inner angle of all triangles in
order to avoid thin triangles [17] [21] [27]. A technique proposed by Ruppert [28] used a successive
refinement of the Delaunay triangulation that uses a uniform mesh (all triangles with the same size). Ruppert
refinement algorithm is for bounded aspect ratio triangulation of planar, producing mesh with fewer
triangles and with varying sizes, where the intersection of two triangles can be a common edge, empty set,
or vertex. While the mesh generation technique must guarantee shape measuring, where the mesh is size
optimal within a specific constant factor [28]. Figure 5, shows an example of a triangle-based mesh
morphing method by using 6 lines selected by the animator.

Figure 5. Triangulation morphing with 6 feature points [17].

Using triangles is still interesting to improve point selection Sofie et. El [29] proposed an algorithm
based on triangulation to generate new geometric data objects that partition the given objects, the aim of
this algorithm is independent of the particular choice of the coordinate system. Triangulation-based warping
algorithm is automatic it can produce good effects on the same orientation images [17]. A common problem
of image warping using triangulation-based methods is that foldover may occur easily [17][21][27]. Which
degrades the resulting image quality. Fold over term describes how the overlapping deformations process
occurs, this means that many nonadjacent in the source image are mapped to the same pixel of the output
image [27]. This problem occurs when the corner point orientation changes to any triangles, that is the
triangle is flipped over by the deformation [17].
A comparison study in [17] concludes that morphing algorithms based on triangulation
implementation are complex and take a significant amount of time if compared with morphing using the
feature-based algorithm. However, the technique inspired many researchers in reference [30] to use
automatically specified feature selection for face morphing where feature specification is applied by using
a neural network, edge detection, and medium filter. Then warping is applied by using triangle-based mesh

46
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

morphing. This approach reduces the time of computation since the control points needed for image
morphing are very less [3].
3.5 Energy minimization
The previously mentioned methods have some weaknesses to generate morph or warp images (flood
or ghost artifacts) where the result might be distorted, therefor Energy minimization techniques were
proposed to provide full image warping, presented by Lee et al. [31]. To generating one to one warp function
from a set of point pairs overlaid on two images by using tow dimensional deformation that efficiently
generates (C 1-) continuous with one-to-one deformations of positional constraints, according to the authors,
the method does not restrict a feature set to have any structure such as mesh method, hence allows greater
freedom in warping design. The freedom together with good warps makes it possible to obtain desired in-
between images very effectively [31]. This technique produces features that meet one-to-one
correspondence the energy clause is sequential and the warping function is calculated to minimize the sum
of it [3]. This technique produces a natural warp since it is based on physically meaningful energy terms.
However, the performance of this method is hampered by its high computational cost [1] [2].

3.6 Multilevel free-form deformation

Since each of the previous methods has some drawbacks. A new warp generation was presented by
Lee et al. in [32]. Which is easier and faster than the related energy minimization approach [1] [16] [31].
Lee et al proposed multilevel free-form deformations (MFFD) to achieve C 2- continuous and one-to-one
warps among feature point pairs. The authors adopt a computer vision technique named snakes to reduce
the burden of feature specification. A Snake: is an energy-minimizing spline guided by external constraint
forces and influenced by image forces that pull it toward features such as lines and edges [33]. This method
was adopted since generating snakes from polylines and curves, then the positions of features can be derived
more effectively. The proposed MFFD derives warps as an extension of the free-form deformation (FFD)
method that aims to deform surface geometric models in a free-form manner proposed by Sederberg et al.
in [34]. FFD recomputes control points' positions depending on the surface point(s) movement to ensure
that the resulting surface passes throw newly positioned points [35]. Lee et al showed that the proposed
MFFD technique aims to reduce image morphing feature specification, the problem of warp generation,
besides surface generation for issues with transition control problems, that appeared in the previous
morphing techniques and that the generated warps provide visually pleasing image distortion, as can be seen
in figure 6, One of the MFFD morphing benefits on feature specification is that it is more expensive and
less cumbersome [3].

Figure 6. Example of MFFD-based morphing [2].

47
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

3.7 Work Minimization

A technique proposed by Gao and Soldberg named work minimization aims to search for some
measures of (work) that can determine how easily morph [36]. The technique is motivated by polygon shape
blending solving problems [37]. Work minimization algorithm can be applied and reduce human work with
little or no user interaction [2]. The method considered solving cases when some similar images of the
(source and destination) for warping are not precise [16]. Therefore, the proposed method is based on a work
minimization strategy that derives its cost directly from the image intensities and not from user-specified
constraints [2].
This approach is consistent with the choice for the warp equation of C0 bi-linear B-splines and
selects the best shape blend as the one requires the least amount of work to bend and stretch one polygon
into the other [36]. Gao and Soldberg, showed that in case using two images, similar images (such as two
faces) in the origin and the destination image, in general, the algorithm works automatically and the result
will be pleasing. The good feature of this algorithm is that it is time-saving. The work minimization
technique is a similar semi-automatic approach but aims at images with similar colors [38]. According to
the author, the drawback of this method there a slight ghosting artifact may occur. An example of this
method with five anchor points is shown in Figure 7.

Figure 7. Example of work minimization morphing [38].

3.8 Morphing Using Structural Similarity

Knowing that the main factor for accurate image morphing is selecting a map between images in a
way that leads to precise fitness between the images. A new technique for automatic mapping proposed by
Jing et al. [38] by optimizing the compatibility of corresponding warped image neighbors, based on adaptive
structural similarity, the authors exploited the benefits of similarity energy and modified similarity index
presented by Wang et al.in [39]. The modified similarity metric was adopted to measure the neighborhoods
with the same structure if they have similar edge structures, considering distinct color distributions and
geometric distortions. Inspired by Yang et al. work in [40]. Jing et al. technique used an efficient and simple
backward-mapping approach to render intermediate frames for morphing. They proposed a fast interactive
per-pixel search technique to invert the mapping. The method parameterizes the map over a halfway domain
correspondence establishing, it also supplies a technique to handle simple occlusions that do not affect the
optimization process. Also, Liao et al explained a method for motion path improvement by applying a
quadratic path optimization in order to reduce deformation during morph operation. The authors compared
this approach with other morphing techniques, where the resulting image is more accurate than the former
morphing methods. According to the author in [38], the adopted similarity energy can manage more generic
morphs of scenes with various colors, and the result is excellent if compared with the results of the work
minimization approach. As in the previous method, the proposed method does not need mesh rasterization,
since every pixel in an intermediate image is evaluated independently, also, the proposed technique is
capable to generate images that can cover the whole domain of the input, unlike mesh-based approaches
[38]. The next figure 8, presents an example of a morphing sequence using a structural similarity approach.

48
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

The three middle images are the midway morphing images of the input images shown on the figure's left
and right [38].

Figure 8. Morphing result using structural similarity [38].

3.9- 3D Face image Morphing

To keep up with the continuous development on 3 dimensions (3D) graphic technique, Yang et al.
[40] presented a new technique applied to two face images to present 3D face geometry morphing, produced
from the same images of different persons, which is very difficult work because people face is not similar
nor fixed, where the same face differs in each expression or position. This method is inspired by the
regenerative morphing approach [41], which generates automatic morph on images have different objects,
using a traditional warp, where the blended approach is replaced with a regenerative approach and the output
sequence produced by the small pieces of the source images in a patch-based [41]. In opposition to it, the
proposed 3D image optimization works on a frame level instead of a patch base. The developed techniques
operate fully automatically on the same face or two different faces and aim to use two different of different
poses or expressions to generate optimized high-quality face morphing animation.
The proposed technique applies warping independently on two faces and only roughly using a 3D
model-based flow, which is different from the classic warping techniques that are applied for morphing
operations that need accurate correspondence of the source images, so Yang et al. adopted this warping
method aims to overcome common warping like “holes” and fold overs, they applied optical flow based
interpolation to warp the background outside the face area by interpolating per frame flow, the method does
not involve mesh or triangle rasterization, since, each pixel in the intermediate image is evaluated
independently. The first step in the developed system requires, extracting facial landmarks from the input
pictures, and projecting the picture on a subspace learned from an external face shape dataset in order to
recover the 3D face geometry. To improve morphing sequence quality, the authors defined a morphing
energy function, and also employed an iterative optimization approach to minimize it [40]. Figure 9, presents
an example of the optimized 3D face morphing method.

Figure 9. Example of 3D Morphing using aware appearance optimization [40].

49
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

It should be noted 3d image morphing is also used to blend 3D human faces into different structures
such as animals, but the problem with this morphing is their faces structures and feature are totally different.
Later Yan et al. presented a new technique for building semantic-adaptive correspondences between human
and animal faces which helps preserve human features better [42]. The proposed Alignment-aware 3D Face
Morphing framework applies morphing using an alignment-aware controller mesh using controller-based
mapping, which builds multi-density correspondences between the source controller and the target controller
according to the importance of semantic information [42]. In fact, 3D image morphing techniques have
developed rapidly in the last few years, Egger et al. in [43] presented a detailed survey of 3d image morphing
techniques and challenges over 20 years starting from the first proposed method. Although developing
efficient 3d image morphing increased the resulted image quality, it recently caused a critical security issue,
especially if an efficient morphing attack is used, as explained in [44] the authors presented a novel
morphing attack, aiming to improve the image visual fidelity, the authors also measured the effectiveness
of morphing attack detector, demonstrating that their method is difficult to detect. Thus, detecting forged
images became a very important field of study in [45] the authors proposed a technique for detecting a single
morphing image attack, using patterns and analyzing the principal component, which produced a good
result. Good image segmentation algorithms may also be useful to classify the object or spaces in the image
as explained in [46], that can help to detect forged image. So, the competition between the attacker and the
detector is contiguous using different techniques including image morphing.

4- Conclusions
Morphing algorithms have become an important technique not only for graphic and animation
design but for various applications, especially those that require predictable results. However, there is a lack
of research that surveys morphing techniques, thus a sincere attempt is made in this research to review
different morphing algorithms and highlight their character with the advantages and limitations. In general
morphing, algorithms share basic steps such as selecting image features, warping, and interpolation
operation. The classic morphing techniques, cross-dissolving mesh warping, triangulation, and field
morphing can be considered as the classic methods that became almost the base or essential for the newly
developed morphing techniques, where modern techniques get advantages of these methods to be developed
or merged with other classic or proposed technique to produce better results with less noticeable visual
artifacts. energy minimization algorithms opened the way to innovative new methods such as MFFD, that
overcome the drawback of the classic algorithms. While work minimization introduces a new technique for
image morphing that was inspired by many types of research to design very efficient morphing algorithms,
such as using structural similarity to present the best result. From the reviewed method, it is clear that 3D
morphing will be dominant in the next few years.
Morphing algorithms vary in the resulting image quality in addition to their computation complexity
that affects the execution time, thus the designer or researcher must compare these methods’ specifications
(strength point and limitation) and adopt the most compatible algorithm for the required application. All the
presented algorithms can be emerged with other morphing algorithms to generate a new hybrid method
using deep learning for example to produce a good warp and morphing for future works. After viewing so
many papers it is clear that more study in this field is required to investigate improving an object’s
transformation process and simultaneously strengthen manipulated image detection techniques.

5. Acknowledgement
The author would like to extend deep thanks and gratitude to the University of Mosul/Iraq for its
support.

50
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

6. References
[1] Wolberg, G. “Recent advances in image morphing”, Proceedings of CG International'96, pp.64-71,1996
[2] Wolberg, G.”Image morphing: a survey”, The visual computer, vol.14, no.8, pp.360-372. 1998
[3] Zope, B. and Zope, S.B., “A Survey of Morphing Techniques”, International Journal of Advanced
Engineering, Management and Science, vol.3, no.2, pp.82-87, 2017.
[4] Singh, H., Kumar, A., and Singh, G., “Image Morphing: A Literature Study”, International Journal of
Computer Applications Technology and Research, vol.3, no.11, pp.701-705, 2014.
[5] Gomes J, Velho L., “Image processing for computer graphics”, Springer Science & Business Media,
pp.149-273,1997.
[6] Scherhag, U., Rathgeb, C., Merkle, J., Breithaupt, R., & Busch, C., “Face recognition systems under
morphing attacks: A survey”, IEEE Access, vol. 7, pp23012-23026, 2019.
[7] Chanho, J. and Kim, M., “Three Methods for Making of Character Facial Animation based on Game
Engine”, International Journal of Asia Digital Art and Design Association, vol.18, no.4, pp.68-73, 2014.
[8] Yuksel, K., Yucebilgin, A., Balcisoy, S., and Ercil, A., “Real-time feature-based image morphing for
memory-efficient impostor rendering and animation on GPU”, The Visual Computer, vol.29, no.2, pp.
131–140, 2013.
[9] Venkatesh S, Ramachandra R, Raja K, Busch C., “Face morphing attack generation and detection: A
comprehensive survey”, IEEE transactions on technology and society, vol.2, no.3, pp.128-45,2021.
[10] Pikoulis EV, Ioannou ZM, Paschou M, Sakkopoulos E., “Face morphing, a modern threat to border
security: Recent advances and open challenges”, Applied Sciences, vol.11, no.7, p.3207, 2021.
[11] Ferrara M, Franco A, Maltoni D, Busch C., “Morphing Attack Potential”, arXiv preprint
arXiv:2204.13374, 2022.
[12] Bagade AM, Talbar SN, “A review of image morphing techniques”, Elixir Electr Eng J, vol.70,
pp.24076–24079, 2014.
[13] Rahman MT, Al-Amin MA, Bakkre JB, Chowdhury AR, Bhuiyan MA, “A novel approach of image
morphing based on pixel transformation”, In10th international conference on computer and information
technology, IEEE, pp.1-5, 2007.
[14] Deepalakshmi, R. “Image Morphing Using Hybrid Mesh-klt Algorithm”, International Journal of
Advanced Research in Computer Science, vol. 9, no. 2,2018.
[15] Fish N, Zhang R, Perry L, Cohen‐Or D, Shechtman E, Barnes C, “Image morphing with perceptual
constraints and STN alignment”, In Computer Graphics Forum, vol. 39, no. 6, pp. 303-313, 2020.
[16] Soni R, Chaudhari R, Shah V., “an overview on image morphing and its techniques”, International
Journal of Advance Engineering and Research Development (IJAERD), vol.1, no.10, 2014.
[17] Bhatt, Ms. Bhumika G. “Comparative Study of Triangulation based and Feature-based Image
Morphing”, Signal & Image Processing: An International Journal (SIPIJ), vol. 2, no.4, 2011.
[18] [19] Gautam A, Verma RS, “Survey of Existing Techniques of Image Morphing”, International Journal
of Computer Applications, vol.161, no.9, pp.28-30, 2017.

51
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

[19] Islam M.B, Inam M.T, Kaliyaperumal B., “Overview and challenges of different image morphing
algorithms”, International Journal of Advanced Research in Computer Science and Electronics
Engineering (IJARCSEE), vol.2, no.4, pp.410-414, 2013.
[20] Lokhande, S.V., and Patil, S.B., “Morphing techniques for facial images-a review”, International
Journal of Engineering, vol.2, no.12, pp.1106-1110, 2013.
[21] Taha, M.A., “Improved Mesh Based Image Morphing”, Journal of University of Babylon for Pure and
Applied Sciences, vol.25, no.5, pp.1608-1617, 2017.
[22] Beier, T. and Neely, S., “Feature-based image metamorphosis”, ACM SIGGRAPH computer
graphics, vol.26, no.2, pp.35-42, 1992
[23] Karam, H., Hassanien, A. and Nakajima, M., “Feature-based image metamorphosis optimization
algorithm”, Proceedings Seventh International Conference on Virtual Systems and Multimedia, IEEE,
pp. 555-564, 2001.
[24] Chang, J.K., Hsieh, W.F., Chen, S.C., Chen, L.H. and Takama, Y., “A Feature-Based Facial Image
Morphing System”, Asian Journal of Computer and Information Systems, vol.2, no.6, 2014.
[25] Rohra, A.I. and Kulkarni, R.K., November, “Survey on recent trends in image morphing techniques”.
In 2019 International Conference on Smart Systems and Inventive Technology (ICSSIT), IEEE, pp. 19-
23, 2019.
[26] Panchal, J.B., Shah, K.R., Sanghani, N.J. and Jhaveri, R.H., “An Implementation of Enhanced Image
Morphing Algorithm using Hybrid Approach”, International Journal of Computer
Applications, vol.66, no.20, 2013.
[27] Patel A, Lapsiwala P. “Image morphing algorithm: A survey”, International Journal of Computer
Application”, vol.5, no.3, pp.156-60, 2015.
[28] Ruppert J. “A Delaunay refinement algorithm for quality 2-dimensional mesh generation”, Journal of
algorithms, vol.18, no.3,548-85, 1995.
[29] Haesevoets S, Kuijpers B, Revesz PZ. “Affine-invariant triangulation of Spatio-temporal data with an
application to image retrieval”, ISPRS International Journal of Geo-Information, vol.6, no.4, p.100,
2017
[30] Karungaru, S., Fukumi, M., Akamatsu, N. and Takuya, A., “Automatic human faces morphing using
genetic algorithms-based control points selection”, International Journal of Innovative Computing,
Information, and Control, vol.3, no.2, pp.1-6, 2007.
[31] Lee, S.-Y., Chwa, K.-Y., Hahn, J. and Shin, S. Y., “Image Morphing Using Deformation Techniques”,
The Journal of Visualization and Computer Animation, vol. 7, no.1, pp.3-23, 1996.
[32] S.-Y. Lee, K.-Y. Chwa, S. Y. Shin, and G. Wolberg, “Image metamorphosis using snakes and free-
form deformations”, Computer Graphics (Proc. SIGGRAPH ’95), pp. 439–448, 1995.
[33] Kass, M., Witkin, A. and Terzopoulos, D., “Snakes: Active contour models”, International journal of
computer vision, vol,1, no.4, pp.321-331, 1988.
[34] T. W. Sederberg and S. R. Parry, “Free-form deformation of solid geometric models”, Computer
Graphics (Proc. SIGGRAPH ’86), vol.20, no.4, pp.151–160, 1986.
[35] Prochazkova, J. and Novak, J., “The solution of free-form deformation problem using
pseudoinverse”, International Journal of Applied Mathematics, vol.31, no.6, pp.831, 2018.

52
 Technium Vol. 9, pp.41-53 (2023)
ISSN: 2668-778X
www.techniumscience.com

[36] Gao, P. and Sederberg, T.W., “A work minimization approach to image morphing”, The Visual
Computer, vol.8, no.14, pp.390-400, 1998.
[37] Sederberg TW, Greenwood E., “A physically based approach to 2D shape blending”, In Catmull EE
(ed), Computer Graphics (SIGGRAPH '92 Proceedings), vol.26, no.2, pp.25-34, 1992.
[38] Liao, J., Lima, R.S., Nehab, D., Hoppe, H., Sander, P.V. and Yu, J., “Automating image morphing
using structural similarity on a halfway domain”, ACM Transactions on Graphics (TOG), vol.33, no.5,
pp.1-12, 2014.
[39] Z. Wang, A. Bovik, H. Sheikh, and E. Simoncelli, “Image quality assessment: From error visibility to
structural similarity”, IEEE Trans, Image Process. vol.13, no.4, pp.600–612. 2004.
[40] F. Yang, E. Shechtman, J. Wang, L. Bourdev, and D. Metaxas, “Face morphing using 3D-aware
appearance optimization”, In Proceedings of the Graphics Interface Conference, pp.93–99, 2012.
[41] E. Shechtman, A. Rav-Acha, M. Irani, and S. Seitz. “Regenerative morphing”. In Proceedings of CVPR,
San-Francisco, CA, 2010. DOI: 10.1109/ CVPR. 2010. 5540159
[42] Yan, X., Yu, Z., Ni, B. and Wang, H., “Cross-Species 3D Face Morphing via Alignment-Aware
Controller”, The Thirty-Sixth AAAI Conference on Artificial Intelligence (AAAI-22), 2022.
[43] Egger, B., Smith, W.A., Tewari, A., Wuhrer, S., Zollhoefer, M., Beeler, T., Bernard, F., Bolkart, T.,
Kortylewski, A., Romdhani, S. and Theobalt, C., “3D morphable face models—past, present, and
future”, ACM Transactions on Graphics (TOG), 39, no.5, pp.1-38, 2020.
[44] Blasingame, Z., & Liu, C., “Diffusion Models For Stronger Face Morphing Attacks”, arXiv preprint
arXiv:2301.04218, 2023.
[45] Dargaud, L., Ibsen, M., Tapia, J., & Busch, C., “A Principal Component Analysis-Based Approach for
Single Morphing Attack Detection”. In Proceedings of the IEEE/CVF Winter Conference on
Applications of Computer Vision, pp. 683-692, 2023.
[46] Jasim, M. S., & Younis, M. C., “Object-based Classification of Natural Scenes Using Machine Learning
Methods”. Technium vol. 6, pp.1-22, 2023.

53