Journal of

Imaging
Article
A New Pseudo-Spectral Method Using the Discrete
Cosine Transform
Izumi Ito
Information and Communications Engineering, Tokyo Institute of Technology, Tokyo 152-8552, Japan;
[email protected]; Tel.: +81-3-5734-2997




Received: 9 February 2020; Accepted: 24 March 2020; Published: 28 March 2020


Abstract: The pseudo-spectral (PS) method on the basis of the Fourier transform is a numerical
method for estimating derivatives. Generally, the discrete Fourier transform (DFT) is used when
implementing the PS method. However, when the values on both sides of the sequences differ
significantly, oscillatory approximations around both sides appear due to the periodicity resulting
from the DFT. To address this problem, we propose a new PS method based on symmetric
extension. We mathematically derive the proposed method using the discrete cosine transform
(DCT) in the forward transform from the relation between DFT and DCT. DCT allows a sequence to
function as a symmetrically extended sequence and estimates derivatives in the transformed domain.
The superior performance of the proposed method is demonstrated through image interpolation.
Potential applications of the proposed method are numerical simulations using the Fourier based PS
method in many fields such as fluid dynamics, meteorology, and geophysics.

Keywords: pseudo-spectral method; derivatives; discrete cosine transform; discrete sine transform

1. Introduction
The discrete cosine transform (DCT) and discrete sine transform (DST) have been extensively
studied, and they have played a crucial role in science and engineering for decades. For example,
DCT is used for standard image and video compression such as JPEG and MPEG. DST is adopted
for high efficiency video coding (HEVC). DCT and DST are closely related to the discrete Fourier
transform (DFT) [1–3]. All the relevant transforms assume the periodicity of sequences. DFT assumes
circular periodicity, where the left side of a sequence is placed next to the right side, while DCT
and DST assume symmetric circular periodicity, where after a sequence is extended symmetrically,
circular periodicity ensues. There are four types of DCT and DST (Types 1, 2, 3, and 4) corresponding
to different types of symmetry. As is well known, DCT Type 2 (DCT-2) is employed for JPEG and
MPEG, and DCT Type 3 is the inverse transform of DCT-2. Some applications use several types of DCT
and DST, while others just use one type [4,5].
Pseudo-spectral (PS) methods originated from [6,7] and have been studied for solutions of
partial differential equations [8]. The numerical solution is obtained via a finite set of expansion
functions. Generally, for periodic problems, the Fourier series is used as expansion functions, while for
non-periodic problems, orthogonal polynomials (e.g., Jacobi polynomials) are used. Legendre and
Chebyshev polynomials are special cases of Jacobi polynomials, in which derivatives are obtained at
unequally spaced points, such as Legendre–Gauss–Lobatto points and Chebyshev–Gauss–Lobatto
points (also referred to as Chebyshev points) [9–11]. Different expansion functions for PS methods are
chosen depending on the problems of applications. The PS method using Fourier series as expansion
functions is employed for estimating derivatives at equally spaced points, which is common in fluid
dynamics [12,13], meteorology, and geophysics, e.g., a direct numerical simulation for a turbulent
flow [14], wave prediction [15], multibody modeling of wave energy converters [16], and seismogram

J. Imaging 2020, 6, 15; doi:10.3390/jimaging6040015 www.mdpi.com/journal/jimaging

J. Imaging 2020, 6, 15 2 of 14

simulation [17]. Generally, DFT is used for the PS method, and in periodic functions, approximations by
the PS method using DFT (PS-DFT) are much more accurate than those by finite difference methods [18].
Moreover, the use of the fast Fourier transform (FFT) accelerates the process. However, conversely, use
of DFT/FFT is problematic due to circular periodicity. If the values on both sides of sequences differ
significantly, oscillatory approximations are observed around both sides, which is already well known
as the Gibbs phenomenon.
Image interpolation by Hermite polynomials is a good example to understand the accuracy of
derivatives. It requires pixel intensities and their derivatives that affect the accuracy of interpolation,
where there is room for the choice of methods for calculating derivatives [19,20]. The combination
with PS-DFT has been shown to outperform conventional interpolation methods such as bicubic and
Lanczos in terms of both accuracy and speed [21]. However, it has not hitherto touched on the adverse
effects by PS-DFT.
In the present paper, we study PS methods based on symmetric extension to address the problem
of the Gibbs phenomenon induced by PS-DFT in image interpolation. We are motivated by the seminal
work [21] combining Hermite polynomials with PS-DFT for image interpolation. We focus on the
symmetric extension of a sequence in the spatial domain, which attenuates the difference in values
at both sides for suppressing the Gibbs phenomenon. The DCT can enable sequences to function as
symmetrical extended sequences. We mathematically derive two types of PS methods using DCT
from the relation between DFT and DCT. We evaluate the proposed methods through combining with
Hermite polynomials for image interpolation. The results testify to the efficacy of the PS method based
on symmetric extension. This paper thus extends earlier results available in the literature [22].
The remainder of the paper is organized as follows. In Section 2, we provide preliminaries in the
form of relevant definitions. We present and derive two PS methods based on symmetric extension in
Section 3. Our evaluations of image interpolation are detailed in Section 4. Finally, conclusions are put
forward in Section 5.
Notations: Let Z and R be the sets of integers and real numbers, respectively. Sequences and
signals in the time domain are represented as lower case letters, and their coefficients in the transformed
domain are denoted as upper case letters. The operator T −1 indicates an inverse transform that assigns
a sequence in the transformed domain to a corresponding sequence in the time domain. The derivatives
of x (n) with respect to n are represented as x 0 (n). We use an asterisk to denote the complex conjugate,
i.e., X ∗ (k ) is the complex conjugate of X (k).

2. Preliminaries
First, we provide the definitions of the relevant transforms and their relations. Then, we briefly
describe the PS method using DFT.

2.1. Definitions of DFT, DCT-1, and DCT-2

Let x (n), n ∈ Z be a sequence of length N. The forward and inverse discrete Fourier transforms
are defined as:
N −1
X (k) = ∑ nk
x (n)WN (1)
n =0

and:
N −1
1
x (n) =
N ∑ −nk
X (k)WN (2)
k =0
√
respectively, where WN = exp(− j2π/N ) and j = −1.
J. Imaging 2020, 6, 15 3 of 14

DCT Type 1 (DCT-1) is defined [23] as:

N −1  
πkn
XC1 (k) = 2 ∑ α(n) x (n) cos
N−1
(3)
n =0

where:
(
1/2, n = 0, N − 1
α(n) = . (4)
1, otherwise

DCT Type 2 (DCT-2) is defined as:

N −1
r
πk(n + 1/2)
 
2
XC2 (k ) = β(k) ∑ x (n) cos (5)
N n =0
N

where:
( √
1/ 2, k=0
β(k) = . (6)
1, otherwise

Note that the inverse DCT-1 corresponds to DCT-1 and that the inverse DCT-2 corresponds to DCT
Type 3.

2.2. Relation between DFT and DCT

The DCT coefficients of an original sequence correspond to the DFT coefficients of the sequence
to which the original sequence is extended symmetrically. Let x (n) be the original sequence of length
N. Figure 1 shows an original sequence and the sequence extended symmetrically depending on the
type of DCT. The difference between DCT-1 and DCT-2 is where the axis of symmetry lies. Specifically,
the axis of symmetry of DCT-1 lies directly between the samples of a sequence, while that of DCT-2
lies at a half-sample between the samples.

x(n)
x̂1 (n)
(a) DCT-1

x(n)
x̂2 (n)
(b) DCT-2
Figure 1. Original sequence and symmetrically extended sequence.
J. Imaging 2020, 6, 15 4 of 14

The symmetrically extended sequence, x̂1 (n), for DCT-1 from x (n), as shown in Figure 1a,
is given as:
(
x (n) n = 0, 1, . . . , N − 1
x̂1 (n) = . (7)
x (2N − 2 − n) n = N, N + 1, . . . , 2N − 3

The relation between DFT coefficients and DCT-1 coefficients is expressed for k = 0, 1, · · · , N − 1 as:

X̂1 (k) = XC1 (k ) (8)

where X̂1 (k) is the DFT coefficients of x̂1 (n) and XC1 (k) is the DCT-1 coefficients of x (n).
The symmetrically extended sequence, x̂2 (n), for DCT-2, as shown in Figure 1b, is given from
x (n) as:
(
x (n) n = 0, 1, . . . , N − 1
x̂2 (n) = . (9)
x (2N − n − 1) n = N, N + 1, . . . , 2N − 1

The relation between DFT coefficients and DCT-2 coefficients is expressed for k = 0, 1, . . . , N − 1 as:
√ 1 −k/2
X̂2 (k) = 2N X (k)W2N (10)
α(k) C2

where X̂2 (k) is the DFT coefficients of x̂2 (n) and XC2 (k) is the DCT-2 coefficients of x (n).
Thus, DCTs can enable sequences to behave as symmetrically extended sequences without the
manual extension of symmetry.

2.3. PS Method Using the DFT

The PS method using DFT (PS-DFT) is a numerical method for estimating the derivatives of
sequences. The underlying theory behind PS-DFT is the time derivative properties of the Fourier
transform.
We assume x (n) are samples of a differentiable continuous-time signal, x a (t), t ∈ R at time nTs ,
i.e., x (n) = x a (nTs ) where Ts represents a sampling period that satisfies the Nyquist sampling theorem.
When the time axis is normalized by a factor of Ts , the samples of the derivative of x a (t) with respect
to t are written as:
∂ ∂
x a (nTs ) = x (n) (11)
∂t ∂n
where 1 = ∂n/∂t from t = nTs and Ts = 1.
From (2), the derivative, x 0 (n), of x (n) with respect to n is expressed as:

N −1
∂ 1 j2πk
x 0 (n) =
∂n
x (n) =
N ∑ N
−nk
X (k)WN . (12)
k =0

Observe that the DFT coefficient, X 0 (k), of x 0 (n) is given in (12) as:

j2πk
X 0 (k) = X ( k ). (13)
N

Note that the coefficients X 0 (k) require complex conjugate symmetry so that x 0 (n) are real numbers.
Therefore, we can obtain the derivatives by the inverse DFT from the coefficients multiplying
the DFT coefficients of the original sequence by the factors with respect to k. The steps involved in
implementing PS-DFT are summarized as follows.
J. Imaging 2020, 6, 15 5 of 14

1. N-point DFT is applied to the sequence of length N to obtain X (k) according to (1).
2. X (k) is multiplied by j2πk/N to obtain X 0 (k) according to (13).
3. The inverse DFT is applied to X 0 (k) according to (2).

3. PS Method Based on Symmetric Extension

Here, we derive two PS methods depending on the type of symmetry extension, considering the
PS-DFT of symmetrically extended sequences. We assume that the symmetrically extended sequence
is the samples that satisfy the Nyquist sampling theorem of a differentiable continuous-time signal
where the time axis is normalized.

3.1. Derivation of PS-DCT1

Let us consider the derivative, x̂10 (n), of the symmetrically extended sequence, x̂1 (n), for DCT-1
in (7).
From (2), x̂10 (n) is given as:
!
2M −1
∂ ∂ 1
x̂10 (n) = x̂ (n) =
∂n 1 ∂n 2M ∑ −nk
X̂1 (k )W2M (14)
k =0

where M = N − 1. With the symmetry properties of DFT, it is developed as:

!
M −1 2M −1
1 ∂
x̂10 (n) =
2M ∂n ∑ −nk
X̂1 (k )W2M + ∑ −nk
X̂1 (k)W2M (15)
k =0 k= M
!
M −1 M −1
1 ∂
=
2M ∂n ∑ −nk
X̂1 (k )W2M + ∑ X̂1∗ (k)W2M
nk
(16)
k =1 k =1
!
M −1 M −1
1 j2πk j2πk ∗
=
2M ∑ 2M
−nk
X̂1 (k)W2M − ∑ nk
X̂ (k)W2M
2M 1
(17)
k =1 k =1

From (8), replacing X̂1 (k) by XC1 (k ), we have:

!
M −1
1 jπk
x̂10 (n) =
2M ∑ M XC1 (k)(W2M −nk
− W2Mnk
) (18)
k =1

1 M−1 πk
 
πnk
M k∑
=− X (k) sin . (19)
=1
M C1 M

Therefore, we define the PS method using DCT-1 in the forward transform as PS-DCT1 together with
−1
the inverse transform, T PS − DCT1 , given by:

N −2  
h i 1 πnk
−1
x̂ 0 1 (n) = T PS 0
− DCT1 XC1 ( k ) = − N −1 ∑ XC0 1 (k) sin
N −1
(20)
k =1

where:
πk
XC0 1 (k) = X ( k ). (21)
N − 1 C1
−1
Note that T PS − DCT1 corresponds to a sign inverted DST Type 1 (DST-1).
J. Imaging 2020, 6, 15 6 of 14

3.2. Derivation of PS-DCT2

In a similar manner, PS-DCT2 is derived. From the definition in (2), the derivative, x̂20 (n), of the
symmetrically extended sequence, x̂2 (n), for DCT-2 in (9) is expressed with the DFT coefficients, X̂2 (n),
of x̂2 (n) as:
!
2N −1
∂ ∂ 1
x̂20 (n) =
∂n
x̂2 (n) =
∂n 2N ∑ −nk
X̂2 (k )W2N (22)
k =0

which is developed with the symmetry properties of DFT as:

!
N −1 2N −1
1 ∂
x̂20 (n) =
2N ∂n ∑ −nk
X̂2 (k )W2N + ∑ −nk
X̂2 (k )W2N (23)
k =0 k= N
!
N −1 N −1
1 ∂
=
2N ∂n ∑ −nk
X̂2 (k )W2N + ∑ X̂2∗ (k)W2N
nk
(24)
k =0 k =0
!
N −1 N −1
1 jπk jπk ∗
=
2N ∑ N
−nk
X̂2 (k )W2N − ∑ nk
X̂ (k)W2N
N 2
(25)
k =0 k =0

From (10), we have:

N −1  
1 jπk −k jπk k
x̂20 (n) = √ ∑
2N k=0 N
−kn
XC2 (k)W2N2 W2N −
N
XC2 (k)W2N
2 kn
W2N (26)

2 N −1 πk
r
πk(n + 1/2)
 

N k∑
=− X (k) sin . (27)
=0
N C2 N

Therefore, we define the PS method using DCT-2 in the forward transform as PS-DCT2 together with
−1
the inverse transform, T PS − DCT2 , given by:

N −1
r
πk(n + 1/2)
 
h i 2
x̂ 0 2 (n) = −1
T PS − DCT2 XC0 2 (k) =−
N ∑ XC0 2 (k) sin
N
(28)
k =0

where:
πk
XC0 2 (k) = X ( k ). (29)
N C2
−1
Note that T PS − DCT2 corresponds to a sign inverted DST Type 2 (DST-2).

3.3. Implementing PS-DCT1/PS-DCT2

The steps involved in implementing PS-DCT1/PS-DCT2 are summarized as follows.

1. DCT-1/DCT-2 is applied to a sequence of length N to obtain XC1 (k)/XC2 (k) according to (3)/(5).
2. XC1 (k)/XC2 (k) is multiplied by a factor, πk/( N − 1) or πk/N, to obtain XC0 1 (k )/XC0 2 (k) according
to (21)/(29).
−1 −1 0 0
3. T PS − DCT1 / T PS− DCT2 is applied to XC1 ( k )/XC2 ( k ) according to (20)/(28).

3.4. Extension to Second and Higher Derivatives

The second and higher derivatives are obtained by differentiating the first derivative repeatedly.
For example, from (28), the second derivatives based on the symmetric extension for DCT-2 are
given as:
J. Imaging 2020, 6, 15 7 of 14

N −1
r
πk (n + 1/2)
 
∂ 0 2
∂n
x̂ 2 (n) = −
N ∑ XC002 (k) cos
N
(30)
k =0

where:
 2
πk
XC002 (k ) = XC2 (k). (31)
N

4. Application to Image Interpolation

We focus on whole-image interpolation for applications such as image registration and motion
correction, rather than interpolation for a small portion of an image. We evaluate the proposed
methods combined with Hermite polynomials for image interpolation in terms of accuracy and speed.
Interpolation by Hermite polynomials is referred to as Hermite interpolation.

4.1. Hermite Interpolation

Let yi and yi0 be the nodal value and the nodal derivative at ti ∈ Z, (i = 0, 1, . . . , I), respectively.
The Hermite polynomial f (t), t ∈ R, of degree 2I + 1, is given [19] by:

I I
f (t) = ∑ yi ai (t) + ∑ yi0 bi (t) (32)
i =0 i =0

where:

p 0 ( ti ) pi ( t ) pi ( t )
 
ai ( t ) = 1 − ( t − ti ) i , bi ( t ) = ( t − t i ) , (33)
pi ( ti ) pi ( ti ) pi ( ti )
p i ( t ) = ( t − t 0 )2 ( t − t 1 )2 · · · ( t − t i −1 )2 ( t − t i +1 )2 · · · ( t − t I )2 . (34)

when I = 1 (degree=3), f (t) in (32) requires two nodal values and two nodal derivatives nearest the
location t for estimating the value at t, which is referred to as cubic Hermite interpolation (CHI).

4.2. Methods and Environment

We performed geometrical transformation by interpolation to evaluate the proposed methods.
The derivatives required for CHI are obtained by PS-DCT1, PS-DCT2, PS-DFT, first-order forward
difference (1-FD), second-order central finite difference (2-CLFD), fourth-order central finite difference
(4-CLFD), and fourth-order compact finite difference (4-CTFD) [24]. In addition to these results,
we provide results according to the cubic natural spline (CNS) to facilitate comparison with other
interpolation methods.
We used a total of 12 grayscale images of size 256 × 256 in standard image data base (SIDBA) [25].
The resulting images are evaluated by the signal-to-noise ratio (SNR) given as:
!
SNR = 10log10 ∑ g(n)2 / ∑(r(n) − g(n))2 (35)
n n

where g(n) and r (n) represent the ground truth sample and the final resulting sample, respectively.
All algorithms were implemented in MATLAB and run on a macOS system with a 2.3 GHz Intel Core
i5 and 8 GB memory.
J. Imaging 2020, 6, 15 8 of 14

4.3. Image Translation

The image was shifted by 0.05 pixels in the horizontal direction in each step. The output of the
previous step was used for the input. After 20 steps, the image was shifted by one pixel.
Figure 2 shows a portion on the leftmost side of the final output after successive image translation
of the image “Barbara”. The results from PS-DFT suffered severe artifacts compared to those from the
other methods. Figure 3 shows the absolute errors of the final output, where black corresponds to zero
error, values greater than 20 are white, and values between zero and 20 are intermediate shades of gray.
It can be clearly discerned that there is less error associated with PS-DCT1 and PS-DCT2 compared to
the other methods. The errors resulting from PS-DCT1, PS-DCT2, and PS-DFT are concentrated on
the both sides of the image. Table 1 summarizes the SNRs between the final output and the image
shifted by one pixel (ground truth). The highest SNR for each image is in bold. It can be observed
that the SNRs of the results from PS-DCT1 and PS-DCT2 were greater than the SNRs of PS-DFT.
Accordingly, it was discerned that symmetric extension in the PS method was effective. On average,
CNS yielded the highest SNR. However, in several images, PS-DCT2 was better than CNS.

(a) ground truth (b) PS-DCT1 (c) PS-DCT2 (d) PS-DFT

(e) 1-FD (f) 2-CLFD (g) 4-CLFD (h) 4-CTFD

Figure 2. A portion on the leftmost side of the final output after successive image translation of the
image “Barbara”. PS, pseudo-spectral; FD, forward difference; CLFD, central finite difference; CTFD,
compact finite difference.

Table 1. SNRs (dB) of the final output after successive image translation.

Interpolation CHI CNS

Derivative PS-DCT1 PS-DCT2 PS-DFT 1-FD 2-CLFD 4-CLFD 4-CTFD –
airplane 22.87 27.00 21.40 23.99 25.40 25.14 27.05 30.77
Barbara 26.05 28.36 19.95 18.18 18.21 20.54 23.74 24.68
boat 25.29 27.91 20.29 23.49 24.21 24.73 27.00 29.82
bridge 25.25 27.08 21.17 19.63 18.34 19.45 21.04 21.35
building 24.43 26.86 21.41 23.46 25.55 25.36 27.50 32.70
cameraman 25.68 27.46 20.31 21.20 21.26 22.30 24.22 25.40
girl 25.14 27.20 22.35 23.71 24.94 24.97 26.79 29.81
LAX 23.45 22.55 21.01 15.24 13.61 14.69 16.20 16.41
Lenna 25.31 27.56 21.74 23.19 24.71 25.15 27.35 30.84
lighthouse 25.03 28.19 21.28 22.36 22.50 23.37 25.28 26.63
text 25.72 28.49 21.45 22.52 23.95 25.01 27.19 29.50
woman 20.01 26.69 25.73 23.08 24.08 24.19 26.29 29.31
average 24.52 27.11 21.51 21.67 22.23 22.91 24.97 27.27
J. Imaging 2020, 6, 15 9 of 14

(a) PS-DCT1 (b) PS-DCT2 (c) PS-DFT (d) 1-FD

(e) 2-CLFD (f) 4-CLFD (g) 4-CTFD (h) CNS

Figure 3. The absolute errors of the final output after successive image translation of the image
“Barbara”. CNS, cubic natural spline.

Next, we changed the evaluation area. Figure 4 shows the SNRs for each image when the
evaluated area was narrowed by m columns on each side of the image. It can be seen that the SNRs of
PS-DCT1, PS-DCT2, and PS-DFT increased significantly when m = 1 and then continued to increase
slowly for m > 1, while the SNRs of the other methods slightly increased when m = 1, then plateaued
for m > 1. The results imply that there are large adverse effects by PS-DCT1 and PS-DCT2 within
one column on both sides of the results. That is, we can obtain superior results by excluding only
one pixel at each side. Table 2 summarizes the SNRs of the final output when m = 1. Compared to
Table 1, the SNRs increased across all methods. Notably, the SNRs of PS-DCT1, PS-DCT2, and PS-DFT
increased significantly. PS-DCT2 performed best, except with respect to the image “LAX”.

Table 2. SNRs (dB) of the final output after successive image translation (m = 1).

Interpolation CHI CNS

Derivative PS-DCT1 PS-DCT2 PS-DFT 1-FD 2-CLFD 4-CLFD 4-CTFD –
airplane 27.46 44.24 34.28 27.16 27.74 29.57 31.86 31.98
Barbara 37.58 42.89 29.93 18.62 18.43 21.24 24.84 24.86
boat 35.18 46.02 31.21 25.56 25.51 27.58 30.48 30.56
bridge 28.49 32.42 28.76 20.04 18.52 19.82 21.44 21.44
building 39.77 44.92 33.99 26.45 28.28 31.09 34.54 34.85
cameraman 36.95 38.01 30.47 22.20 21.80 23.58 25.58 25.60
girl 38.43 42.45 35.13 26.27 26.75 28.52 30.57 30.66
LAX 28.60 28.19 27.72 15.52 13.73 14.94 16.46 16.46
Lenna 40.30 44.73 33.60 25.27 26.31 28.80 31.67 31.79
lighthouse 30.84 39.16 31.48 23.59 23.17 24.88 26.87 26.90
text 33.58 42.63 31.88 23.79 24.95 27.42 30.01 30.08
woman 22.30 41.11 42.62 25.74 25.83 27.61 30.15 30.24
average 33.29 40.57 32.59 23.35 23.42 25.42 27.87 27.95
J. Imaging 2020, 6, 15 10 of 14

Figure 4. SNRs of the final output when excluding m columns on both sides of the image.

4.4. Image Rotation

The image was rotated by 24 degrees in each step. The output of the previous step was used for
the input. After 15 steps, the image was fully rotated. We made each output image the same size as the
input, cropping and zero-padding the rotated image to fit. As a result, the part with pixel intensities
other than zero of the final output became circular.
Figure 5 shows a portion near the circumference of the final output after successive rotation of
the image “Lenna”. The results from PS-DCT1, PS-DCT2, and PS-DFT were clearer than those from
the other methods, but there were minor artifacts in the homogeneous area of the results. Figure 6
shows the absolute errors of the final output. Again, black denotes zero error; values greater than
20 are white; and values between zero and 20 are intermediate shades of gray. There was less error
associated with PS-DCT1, PS-DCT2, and PS-DFT compared to the other methods.
Table 3 summarizes the SNRs of the final output when the evaluation area was narrowed by
one pixel of the inner the circle, where the highest SNR for each image is in bold. It can be seen that
PS-DCT2 yielded the best SNRs of all the methods. However, compared to Table 2, there were no
J. Imaging 2020, 6, 15 11 of 14

significant differences among PS-DCT1, PS-DCT2, and PS-DFT. One reason for this was that the rotated
image for the next input in successive rotation was zero-padded so that the image should be square,
which attenuated the Gibbs phenomenon. Another reason was that errors were not concentrated in
one place, due to rotation.

(a) ground truth (b) PS-DCT1 (c) PS-DCT2 (d) PS-DFT

(e) 1-FD (f) 2-CLFD (g) 4-CLFD (h) 4-CTFD

Figure 5. A portion near the circumference of the final output after successive rotation of the image
“Lenna”.

(a) PS-DCT1 (b) PS-DCT2 (c) PS-DFT (d) 1-FD

(e) 2-CLFD (f) 4-CLFD (g) 4-CTFD (h) CNS

Figure 6. The absolute errors of the final output after successive rotation of the image “Lenna”.

4.5. Computational Complexity

PS-DCT1 requires an N-point DCT-1 and an N-point DST-1 for the forward and inverse transforms,
respectively, and N multiplications for the derivatives in the transformed domain. Likewise, PS-DCT2
needs an N-point DCT-2, an N-point DST-2, and N multiplications. PS-DFT requires an N-point
DFT, an N-point inverse DFT, and N multiplications. Table 4 summarizes the arithmetic operations
of PS-DCT1, PS-DCT2, and PS-DFT for a sequence of length N. The operations of PS-DCT1 and
PS-DCT2 were based on Wang’s algorithm [26]. The operations of PS-DFT were based on the FFT [27],
J. Imaging 2020, 6, 15 12 of 14

which were converted to operations in real numbers in which one multiplication of complex numbers
was converted to three multiplications and three additions according to Nakayama’s method [28] and
one addition of complex numbers was converted to two additions. Figure 7 shows the comparison of
the arithmetic operations based on Table 4.
We measured the execution time for image interpolation. All algorithms were executed on
a macOS system with a 2.3GHz Intel Core i5 and 8GB memory. A total of 12 images of size 256 × 256
were used. Table 5 summarizes the mean execution time for successive image translation and successive
image rotation. The proposed methods ran slightly more slowly than PS-DFT, but the execution time
was acceptable. Together with the above qualities, the results testified to the efficacy of the proposed
methods.

Table 3. SNRs (dB) of the final output after successive image rotation.

Interpolation CHI CNS

Derivative PS-DCT1 PS-DCT2 PS-DFT 1-FD 2-CLFD 4-CLFD 4-CTFD –
airplane 30.15 31.66 30.85 25.54 25.51 27.16 28.69 24.83
Barbara 27.14 27.82 27.41 18.80 18.65 20.55 22.86 19.16
boat 31.42 33.55 32.37 26.71 26.69 28.30 30.07 25.60
bridge 22.36 22.58 22.46 18.21 18.14 19.07 20.01 18.33
building 30.02 31.86 30.68 24.67 24.64 26.73 28.58 24.14
cameraman 26.83 27.52 27.14 21.29 21.25 22.76 24.16 21.42
girl 29.27 30.34 29.43 24.79 24.77 26.41 27.77 24.60
LAX 18.42 18.49 18.46 14.47 14.42 15.20 16.02 14.73
Lenna 30.30 31.79 30.90 24.91 24.88 26.70 28.49 24.50
lighthouse 24.55 24.95 24.72 19.60 19.53 20.65 21.72 19.68
text 25.86 26.49 26.03 20.34 20.26 21.83 23.23 20.40
woman 30.93 32.28 31.44 26.46 26.46 27.99 29.43 25.90
average 27.27 28.28 27.66 22.15 22.10 23.61 25.09 21.94

Table 4. Arithmetic operations for PS-DCT1, PS-DCT2, and PS-DFT in real numbers.

Method Multiplications Additions

N 7
PS-DCT1 2 (3 log2 N − 10) + 4 log2 N + 6 + N 2 N (log2 N − 2) + 4 log2 N +6
PS-DCT2 2N ( 34 log2 N − 1) + 6 + N N ( 27 log2 N − 4) + 6
PS-DFT 3( N log2 N + N ) 4N log2 N + 3( N log2 N + N )

(a) multiplications (b) additions

Figure 7. Comparison of the arithmetic operations based on Table 4.

J. Imaging 2020, 6, 15 13 of 14

Table 5. Mean execution time (s) for successive image translation and successive image rotation. CHI,
cubic Hermite interpolation.

Method Translation Rotation

CHI-PS-DCT1 0.058 0.436
CHI-PS-DCT2 0.052 0.417
CHI-PS-DFT 0.037 0.391
CHI-1-FD 0.019 0.357
CHI-2-CLFD 0.021 0.362
CHI-4-CLFD 0.024 0.372
CHI-4-CTFD 1.455 3.207
CNS 3.399 92.866

5. Conclusions
We proposed PS methods based on symmetric extension to attenuate the oscillatory approximation
that occurs with DFT. Using DCT, derivatives can be estimated in the transformed domain with
sequences behaving as symmetrically extended sequences without a manual extension of symmetry.
We derived two PS methods, PS-DCT1 and PS-DCT2, from the relation between DFT and DCT.
Through image interpolation by Hermite polynomials, we showed that the proposed methods
outperform PS-DFT. DCT can be calculated in real numbers rather than in the complex numbers
that are required for DFT. Moreover, DCT has fast algorithms, as well as DFT. Therefore, the proposed
methods are advantageous in terms of accuracy and validity compared to PS-DFT.
The potential applications of the proposed method are numerical simulations/modeling using
the Fourier based PS method to solve several equations in many fields such as fluid dynamics,
meteorology, and geophysics. Furthermore, the proposed method may be applicable to numerical
methods where the Fourier based PS method has not been used before, such as those relevant to sensor
networks and radar imaging [29–31]. The future work of this paper will study potential applications
with numerical analysis.

Funding: This research received no external funding.

Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; nor in the
decision to publish the results.

References
1. Rao, K.R.; Yip, P. Discrete Cosine Transform; Academic Press. Inc.: Boston, MA, USA, 1990.
2. Martucci, S.A. Symmetric convolution and the discrete sine and cosine transforms. IEEE Trans. Signal Process.
1994, 42, 1038–1051. [CrossRef]
3. Reeves, R.; Kubik, K. Shift, scaling and derivative properties for the discrete cosine transform. Signal Process.
2006, 86, 1597–1603. [CrossRef]
4. Long, B.; Reinhard, E. Real-time fluid simulation using discrete sine/cosine transforms. IEEE Trans. Pattern
Anal. Mach. Intell. 2001, 23, 643–660.
5. Bhat, P.; Curless, B.; Cohen, M.; Zitnick, C.L. Fourier analysis of the 2D screened poisson equation for
gradient domain problems. In European Conference on Computer Vision; ECCV Part II LNCS; Springer:
Berlin/Heidelberg, Germany, 2008; Volume 5303, pp. 114–128.
6. Kreiss, H.-O.; Oliger, J. Comparison of accurate methods for the integration of hyperbolic equations. Tellus
1972, 24, 199–215. [CrossRef]
7. Orzag, S.A. Comparison of pseudospectral and spectral approximation. Appl. Math. 1972, L1, 253–259.
[CrossRef]
8. Fornberg, F. A Practical Guide to Pseudospectral Methods; Cambridge University Press: Cambridge, UK, 1998.
9. Elnagar, G.N.; Kazemi, M.A.; Razzaghi, M. The pseudospectral Legendre method for discretizing optimal
control problems. IEEE Trans. Autom. Control 1995, 40, 1793–1796. [CrossRef]
J. Imaging 2020, 6, 15 14 of 14

10. Fahroo, F.; Ross, I.M. Costate estimation by a Legendre pseudospectral method. J. Guidance Control Dyn.
2001, 24, 270–277. [CrossRef]
11. Elnagar, G.N.; Kazemi, M.A. Pseudospectral Chebyshev optimal control of constrained nonlinear dynamical
systems. Comput. Optim. Appl. 1998, 11, 195–217. [CrossRef]
12. Roache, P.J. A pseudo-spectral FFT technique for non-periodic problems. J. Comput. Phys. 1978, 27, 204–220.
[CrossRef]
13. Canuto, C.; Hussaini, M.Y.; Quarteroni, A.; Zang, T.A. Spectral Methods in Fluid Dynamics; Springer: New York,
NY, USA, 1988.
14. Takata, M.; Yamamoto, Y.; Shouno, H.; Kunugi, T.; Joe, K. Parallelization and evaluation of a direct numerical
simulation for a turbulent flow using pseudo spectral method. IPSJ J. 2003, 44, 45–54. (In Japanese)
15. Yoon, S.; Kim, J.; Choi, W. An explicit data assimilation scheme for a nonlinear wave prediction model based
on a pseudo-spectral method. IEEE J. Ocean. Eng. 2016, 41, 112–122.
16. Paparella, F.; Bacelli, G.; Paulmemo, A.; Mouring, S.E.; Ringwood, J.V. Multibody modelling of wave energy
converters using pseudo-spectral methods with application to a three-body hinge-barge device. IEEE Trans.
Sustain. Energy 2016, 7, 966–974. [CrossRef]
17. Das, S.; Chen, X.; Hobson, M.P. Fast GPU-based seismogram simulation from microseismic events in
marine environments using heterogeneous velocity models. IEEE Trans. Comput. Imaging 2017, 3, 316–329.
[CrossRef]
18. Smith, G.D. Numerical Solution of Partial Derivative Equations: Finite Difference Methods; Oxford University
Press: Oxford, UK, 1986.
19. Szegö, G. Orthogonal Polynomials; American Mathematical Society: Providence, RI, USA, 1939.
20. Seta, R.; Okubo, K.; Tagawa, N. Digital image interpolation method using higher-order Hermite interpolating
polynomials with compact finite-difference. In Proceedings of the 2009 APSIPA Annual Summit and
Conference, Sapporo, Japan, 4–7 October 2009.
21. Takamiya, K.; Okubo, K.; Tagawa, N. High accuracy image interpolation method combining higher-order
Hermite interpolation with pseudo spectral method. IEICE Trans. Commun. 2012, J95-Bo.2, 355–365. (In
Japanese)
22. Ito, I. Pseudo spectrum method based on symmetric extension. In Proceedings of the 7th European Workshop
on Visual Information Processing, Tampere, Finland, 26–28 Novober 2018; p. 29.
23. Oppenheim, A.V.; Schafer, R.W.; Buck, J.R. Discrete-Time Signal Processing; Prentice Hall: New Jersey, NJ,
USA, 1999.
24. Lele, S.K. Compact finite difference schemes with spectral-like resolution. J. Comput. Phys. 1992, 103, 16–42.
[CrossRef]
25. Standard Image Data BAse. Available online: http://www.ess.ic.kanagawa-it.ac.jp/app_images_j.html
(accessed on 27 March 2020).
26. Wang, Z. Fast algorithms for the discrete W transform and for the discrete Fourier transform. IEEE Trans.
Accoust. Speech Signal Process. 1984, 32, 803–816. [CrossRef]
27. Nussbaumer, H.J. Fast Fourier Transform and Convolution Algorithms; Springer: Berlin, Germany, 1982.
28. Nakayama, K. A new discrete Fourier transform algorithm using butterfly structure fast convolution.
IEEE Trans. Acoust. Speech Signal Process. 1985, ASSP-33, 1197–1208. [CrossRef]
29. Rossi, P.S.; Ciuonzo, D.; Ekman, T.; Dong, H. Energy detection for MIMO decision fusion in underwater
sensor networks. IEEE Sens. J. 2014, 15, 1630–1640. [CrossRef]
30. Devaney, A.J.; Marengo, E.A.; Gruber, F.K. Time-reversal-based imaging and inverse scattering of multiply
scattering point targets. J. Acoust. Soc. Am. 2005, 118, 3129–3138. [CrossRef]
31. Ciuonzo, D. On time-reversal imaging by statistical testing. IEEE Signal Process. Lett. 2017, 24, 1024–1028.
[CrossRef]

c 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).