Int J Theor Phys

DOI 10.1007/s10773-014-2402-4

Quantum Image Morphology Processing Based

on Quantum Set Operation

Ri-Gui Zhou · Zhi-bo Chang · Ping Fan · Wei Li ·

Tian-tian Huan

Received: 15 July 2014 / Accepted: 29 October 2014

© Springer Science+Business Media New York 2014

Abstract Set operation is the essential operation of mathematical morphology, but it is

difficult to complete the set operation quickly on the electronic computer. Therefore, the
efficiency of traditional morphology processing is very low. In this paper, by adopting the
method of the combination of quantum computation and image processing, though multiple
quantum logical gates and combining the quantum image storage, quantum loading scheme
and Boyer search algorithm, a novel quantum image processing method is proposed, which
is the morphological image processing based on quantum set operation. The basic opera-
tions, such as erosion and dilation, are carried out for the images by using the quantum
erosion algorithm and quantum dilation algorithm. Because the parallel capability of quan-
tum computation can improve the speed of the set operation greatly,√ the image processing
gets higher efficiency. The runtime of our quantum algorithm is O( MN). As a result, this
method can produce better results.

Keywords Quantum morphology · Quantum erosion · Quantum dilation ·

Quantum loading scheme

1 Introduction

Along with the arrival of the era of big data, the huge amount of data to be processed by the
computer has brought a severe challenge to the hardware and the data algorithm [20]. It is
acknowledge that the failure of Moore law is only a matter of time, the method of manu-
facturing the computer has appeared to be inadequate in solving the difficulty of data scale.
When the electronic devices become smaller and smaller, their function is disturbed by the
quantum effects. Hence, the model of quantum computing theory is a possible way to solve
the Moore law failure. Compared with classical computer, it is quite superior that quantum

R.-G. Zhou · Z.-b. Chang () · P. Fan · W. Li · T.-t. Huan

School of Information Engineering, East China JiaoTong University,
Nanchang, Jiangxi, 330013, People’s Republic of China
e-mail: [email protected]
 Int J Theor Phys

computer deals with big data. Therefore, many researchers devoted to the study of quantum
computers to make the quantum computer to be the next generation of computer. In the past
few years there are a lot of discoveries from the perspective of theoretical research, which
has laid a solid foundation for the birth of quantum computer. For example, the quantum par-
allelism, coherence, entanglement and the superposition of quantum states, all of those are
the inherent characteristics in the quantum computing. So far, some representative quantum
image storage methods have been proposed, including FRQI [1], NAQSS [2] and quantum
gray-scale image expression [3]. In order to improve the efficiency of image processing,
some quantum algorithms, achieved good results, are proposed in the previous studies,
such as quantum image transformation [4], color transformation [5], quantum image com-
pression [6] and edge detection [7]. Even some more efficient quantum algorithms for the
image processing have brought great conveniences, for instance, quantum Fourier transform
[8–10], quantum wavelet transform [11–13], Grover search algorithm [14], Boyer search
algorithm [23], etc. However, the quantum algorithm about the mathematical morphology
has not yet been put forward. Only some methods about quantum-inspired mathemat-
ical morphology were presented [15–17]. This paper proposes an improved derivative
algorithm based on these researches, in order to obtain a real quantum morphology
algorithm.
The paper is organized as follows: In Section 2, we introduce the theories of the set
operation and discuss the quantum logic gates. In Section 3, we use the quantum loading
scheme (QLS) and Boyer algorithm to build the quantum erosion algorithm and quantum
dilation algorithm. In Section 4, we simulate the algorithm and analyze the performance of
algorithms. Summary and outlook are proposed in Section 5.

2 Quantum Logic Gates and Quantum Set Operations

2.1 Basic Quantum Logic Gates

Similar to the logic gate of the classical computer and circuit composed of the logic gate
line, the quantum computer is constituted by the quantum circuit, which is formed by the
rank of logic gate and used to quantum information processing. Quantum Gates are mainly
used for processing information and transforming information, making it a quantum a form
that can be handled by quantum computers. Quantum gates can be divided into single
quantum bit gates and multiple quantum bit gates in quantum computation. This section
mainly introduces the basis of multiple quantum bit—CNOT gate, the X gate and the swap
gate.
1 CNOT gate
Controlled not gate is the archetype of the multiple quantum logic gate. The gate has two
input qubits: control qubit and target qubit, as shown in Fig. 1:

Fig. 1 CNOT gate

Int J Theor Phys

In CNOT gate, if the control bit is 0, the target bit will flat. On the contrary, the target bit
will flip if control bit is 1, such as |00 → |00 , |01 → |01 , |10 → |11 , |11 → |10.
The matrix is expressed as:
⎛ ⎞
1 0 0 0
⎜0 1 0 0⎟
Cnot = ⎜
⎝0 0 0 1⎠
⎟ (1)
0 0 1 0
2 The basic logic gates
Any multiple qubit gates can be formed by different combinations of CNOT qubit gates
and single qubit gates. In addition, many other elementary logic gates will be used for the
computation. A master list includes the AND gate, the OR gate, the XOR gate, the NAND
gate, and the NOR gate [20]. These quantum gates have two qubits as input but only single
qubit as output. The actions of these gates are illustrated in Fig. 2.
3 X gate and swap gate
The role of X quantum gate is used to realize the conversion of single quantum bit, such as
1 to 0. Similarly, the role of swap gate is to implement the interchange between two qubits
(Fig. 3).
The combination of all the basic quantum gates can be used to realize the quantum image
translation [3].

2.2 Quantum Loading Scheme

It should be known how to load the set information into quantum state from electronic
memory, if we would like to realize quantum image processing. The conception of Quantum
Loading Scheme (QLS) was proposed by Pang in 2007 [18]. QLS is the unitary operation
that loads all information of data set into quantum state from electronic memory. Pang’s
study shows that, for a vector a → = {a1 , a2 , · · · , aN−1 }, there is a unitary operation UQLS
to implement the following process:
N−1
1
UQLS : |0register1|0register2 |ancillaregister3 → √ |iregister1|ai register2 |ancillaregister3
N i=0
(2)

a
a NOT a a AND b
b

a a
a OR b a XOR b
b b

a
a NAND b
b

a
a NOR b
b

Fig. 2 Elementary circuits performing the AND, OR, XOR, NAND and NOR gates
 Int J Theor Phys

Fig. 3 Swap gate a b

b a

The implementation of this U transformation method can be found in the [24].The stud-
ies of Seth Lloyd and Pangs’ groups both demonstrate that, using the path interference
of molecule, the speed of QLS is faster than classical one and the time complexity is
O log2 N . By contrast, the time complexity of classical via I/O interface data loading
scheme is O (N). It is an efficiency bottleneck that any classical algorithm is difficult to
overcome.

2.3 The Basic Concepts of Set Theory

Set operation is the most basic one in mathematics, like intersection, union, complement,
etc. These basic operations are also essential to the future quantum computer. In this section,
a collection of mathematical set operations will be introduced and discussed in the Hilbert
space.
22n −1
In the reference [3], the quantum image is stored by |Q = 1
2n |M ⊗ |j , here M
j =0
denotes the gray-scale. |j  = |h |v = |hn−1 , · · · , v − 1, v0  here |h represents the posi-
tion of the x direction information and the position of y direction information is presented
as |v. Supposing we have two vector sets A and B, they are both stored in this way.
If the vector |ω is an element of A, it will express |ω ∈∈ A. Reversely if the vector |ω
is not an element of A, it will express |ω ∈/ A. The set B of pixel coordinates that satisfying
a particular condition is written as B = {|ω |condition}. For example, the set of all pixel
coordinates that do not belong to set A, denoted as Ac , is given by Ac = {|ω | |ω ∈ / A}.
This set is called the complement of A. The union of two sets A and B, denoted by C =
A ∪ B is the set of all elements that belong to A, to B, or to both. Similarly, the intersection
of sets A and B, denoted by C = A ∩ B is the set of all elements that belong to both A and
B. The difference of sets A and B, denoted as A − B is the set of all elements that belong to
A but not to B: A − B = {|ω | |ω ∈ A, |ω ∈ / B}.
Figure 4 illustrates the set operations defined thus far. The result of each operation is
shown in gray.

3 Quantum Algorithm of Dilation and Erosion

This paper uses QLS to load the quantum image A and the structuring element B. It is
supposed that A is a 2n × 2n quantum image and B is a 2 × 2 structuring element. In this
paper, the size of A is denoted by N and the size of B is denoted by M.
Initially, the six registers are used to store the location and pixel values of A and B
images.

|Ck register1 |I register2 |J register3 |AI register4 |BJ register5 |fc (AI , BJ )register6 (3)
Int J Theor Phys

Fig. 4 Set operations

Ck is used to identify the origin location of structuring element. I, AI and J, BJ are

entangled state [19].

1 if AI = BJ
Thefc is a match function andfc (AI , BJ ) = (4)
0 otherwise

(1) Hadamard transform is applied to the initial state:

M−1 M−1
1
H : |0 |0 |0 |0 |0 |0 → |Ck  |I  |J  |0 |0 |0 (5)
M
I =0 J =0
 Int J Theor Phys

(2) Quantum loading scheme is constructed:

M−1 M−1 M−1 M−1

1 1
UL : |Ck  |I |J |0|0|0 → |Ck  |I  |J  |AI  |BJ  |0 (6)
M M
I =0 J =0 I =0 J =0
The function of UL is to load the value of A and B into the entangled state.
(3) Design black box of quantum computation Oc to compute the value of match function
fc and Oc which is a unitary operation.
M−1 M−1
1
UL : |Ck  |I  |J  |0 |0 |0
M
I =0 J =0
M−1 M−1
1
→ |Ck  |I  |J  |AI  |BJ  |fc (AI , BJ ) (7)
M
I =0 J =0
(4) Design black box of Grover algorithm Of which is used to search problem of the solu-
tion. When the searched target meets the conditions of the problem, Of will reverse
the phase position of the quantum state:
M−1 M−1
1
Of : |Ck  |I  |J  |AI  |BJ  |fc (AI , BJ )
M
I =0 J =0
M−1 M−1
1
→ |Ck  (−1)register6 |I  |J  |AI  |BJ  |fc (AI , BJ )register6 (8)
M
I =0 J =0
(5) Construct general Grover iteration:

Ggeneral = (2 |ξ  ξ | − I) (UL )† (Oc )† Of Oc UL (9)

The function is equal to Grover Iteration, but it more complicated to calculate.

3.1 Quantum Algorithm of Dilation

Due to the difference between quantum image and classical image, the structuring element
is certainly different from the classical one. Hence the quantum memory expression is
given in the reference [3] to store the structuring element. In fact, the structuring element is
generally far smaller than the image size. Therefore, structuring element usually only needs
3
1 or 2 qubits to meet the requirements. For example, |B = 1
2 |M ⊗ |j  is a structuring
j =0
element, and it is composed of the quantum bit. Beside we suppose the value of |M is 1.

3.1.1 Dilation

In digital image processing [21], it is known that the dilation is an operation that the object
of an image “grows” or “thickens”. The specific manner and extent of this thickening is
controlled by a shape, which is referred to as a structuring
 element. The dilation A by B,
∧
denoted as A ⊕ B, is presented as follows: A ⊕ B = z| B z ∩ A = ∅ .
Figure 5 shows a binary image, which is dilated by a structuring element.
Int J Theor Phys

000 001 010 011 100 101 110 111

000 0 0 0 0 0 0 0 0
001 0 0 0 0 0 0 0 0
010 0 0 0 0 0 0 0 0
011 0 0 1 1 1 1 0 0
100 0 0 1 1 1 1 0 0
101 0 0 1 1 1 1 0 0
110 0 0 0 0 0 0 0 0
111 0 0 0 0 0 0 0 0

(a)
Origin
location

0 1 0 1
0 1 1
0 1 1

Origin
1 1 1 location 1 1 1

Structuring Structuring
element B element
(b)

000 001 010 011 100 101 110 111

000 0 0 0 0 0 0 0 0
001 0 0 0 0 0 0 0 0
010 0 1 1 1 1 1 1 0
011 0 1 1 1 1 1 1 0
100 0 1 1 1 1 1 1 0
101 0 1 1 1 1 1 1 0
110 0 0 0 0 0 0 0 0
111 0 0 0 0 0 0 0 0
(c)

Fig. 5 The process of dilation

3.1.2 Quantum Dilation Algorithm

Algorithm thinking: The image A is regarded as a background image, the structuring ele-
ment B is regarded as the foreground image and the structuring element must indicate
the origin, just like Fig. 5b. In the process that the original point threads the region of
the image A, Ggeneral method is used to search problem of the solution and find where
the overlapping elements are. If there are elements that meet the conditions, the out-
put of image is setted as 1 at the corresponding location of the origin of the structuring
element.
 Int J Theor Phys

In the procedure of dilation, all-1 matrix of structuring element is used. Only in this
way can it be ensured that the intersection exists solution if n=2. The requirement for the
structuring element in dilation algorithm is also a limitation of the algorithm. Based on
analyzing the specialty of binary image information hiding, it is only needed to call this
algorithm once. When it has completed the program, the output image C can be assigned
values according to the values of n.

3.2 Quantum Algorithm of Erosion

3.2.1 Erosion

The erosion can “shrink” or “thin” objects in a binary image. As in dilation, the manner and
extent of shrinking is controlledby a structuring
 element. The erosion of A by B, denoted
as AB, is defined as AB = z| (B)z ⊆ A .
Figure 6 illustrates the erosion process.

3.2.2 Quantum Erosion Algorithm

Algorithm thinking: Similarly, we use the structuring element and its covered image to
perform the “AND” operation. The Ggeneral method is used to search for the solution of
the intersection every time before moving and checking to see where the structuring ele-
ment B fit entirely within the background image A. If the structuring element B shares
every element with the background of A, the output image C is assigned 1-valued at the
corresponding location of the origin point of the structuring element.
Int J Theor Phys

4 Simulation and Algorithm Performance Analysis

In this part, some simulation experiments are carried out by using the dilation and erosion
algorithm which have been presented. What’s more, the properties of quantum algorithm
are explored and a quantitative analysis is given.
Figure 7a is the original image which is used to carry out the simulation. This is a bacteria
gray image. The first step, the image is processed to make it become a binary image in
Fig. 7b.
The Fig. 7b is a 256 × 256 image. This image needs a qubit to store two gray-scale and
three qubits to store eight position states. The structuring element is a 4 × 4 1-matric. The
mode of storing is same with the Fig. 7b. The Fig. 7c,d present the simulation result by
simulating the dilation and erosion algorithm in a classical computer.
1. The analysis of dilation algorithm
During the whole calculation of dilation program S|c| steps’ computation is needed to
include all solutions (S|c| denotes the step number that the origin of the structuring element
thread the background image). Now, the time complexity of each intersection is to be ana-
lyzed. Take the first intersection to explain the problem (C0 = A B0 ∩ B, here C0 denotes
the first value of C, AB0 denotes the region that is covered by B). AB0  = M, |B| = M, |•|
 Int J Theor Phys

Fig. 6 The process of erosion 000 001 010 011 100 101 110 111
000 0 0 0 0 0 0 0 0
001 0 0 0 0 0 0 0 0
010 0 0 0 0 0 0 0 0
011 0 0 1 1 1 1 0 0
100 0 0 1 1 1 1 0 0
101 0 0 1 1 1 1 0 0
110 0 0 0 0 0 0 0 0

00
00 1

Origin
01 1 location

10 1

Structuring Element B

000 001 010 011 100 101 110 111

000 0 0 0 0 0 0 0 0
001 0 0 0 0 0 0 0 0
010 0 0 0 0 0 0 0 0
011 0 0 0 0 0 0 0 0
100 0 0 1 1 1 1 0 0
101 0 0 0 0 0 0 0 0
110 0 0 0 0 0 0 0 0
111 0 0 0 0 0 0 0 0

⎛   ⎞
 
AB0 ×|B|
denotes the size of set. The time complexity of the dilation algorithm is O ⎝ T
⎠

(10) and the output of dilation algorithm is n = 1 or n = 2 is obtained after the

 operation
 is
executed once (T denotes the scale of the search scope). Then the values of AB0  , |B| and
√
T are plugged into the formula (10) to get the value of O( M). √
Similarly, the time of calculating C1 , C2 , · · · , CM−1 is also O( M). The whole pro-
cess of
√ the dilation costs N round-trips. Therefore, the complexity of dilation algorithm
is O( MN ). But the time complexity of classical algorithm is O (MN). Compared √
with the classical algorithm, the speed of this algorithm has been increased O( M)
times.
1. The analysis of erosion algorithm
The erosion algorithm is illustrated in Fig. 7d. Similar to the dilation algorithm, the erosion
algorithm also needs to cost S|C| steps’ computation. If the erosion operation is com-
pleted successfully, the output of each step will fulfill the requirements. That is to say,
Int J Theor Phys

Fig. 7 The simulation results of the two algorithms

AI0 = BJ0 , AI1 = BJ1 , · · · , AI

M = BJM and the final result of n is 2M. Take the first

intersection, there are actually c MM |B| calculations (c is a constant) after invoking algo-
rithm 2. Subsequently, the output has a solution and two possibilities. One is AIi = BJi (i =
0, · · · , M − 1) and the other is AIi = BJj (i = j ). Suppose that the probability of the first
case is bigger than 1/2. In this case, the second one must be less than 1/2. That is to say,
there is a small value ε which is between the two cases, namely case1 = 12 + ε, case2 =
2 − ε.
1

According to the time complexity of each intersection, the recursion equation is shown
as follows:

⎧     
⎪
⎪ I = 1
− ε c MM
+ 1
+ ε IM−1
⎪
⎪
M
 2
  |B| 2
 
⎪
⎪
⎪
⎪ I = 1
− ε c (M−1)(M−1)
+ 1
+ ε IM−2
⎨ M−1 2 |B−1| 2
.. (10)
⎪
⎪ . 
⎪
⎪
⎪
⎪ = 1
−
⎪
⎪
I 1 2 ε c
⎩
1 ≤ M ≤ |B| , 0 < ε < 12
 Int J Theor Phys

where IM denotes the first computation and there are M numbers of unknown solution. The
rest can be denoted in the same manner. Next, IM−1 · · · I1 are plugged into the equation IM
and that is shown as follows:
⎡ ⎡  √ ⎤⎤
       1
1 √ 1 1 √ 1 −ε c M −2+
+ε ⎣ −ε c M −1+ +ε ⎣     ⎦⎦
2
IM = −ε c M +
2 2 2 1
2 +ε [· · · ] · · · 1
2 −ε c
      2  
√ 1 √ 1 1 √ 1 1
= c M −ε +c M −1 +ε −ε +c M −2 +ε −ε
2 2 2 2 2
 M−1  
√ 1 1
+ · · · + +c 1 +ε −ε (11)
2 2
This result is an asymptotic expression
 of time complexity. So it only need to get the max-
√
imum value, namely c M 2 − ε . Therefore, the program of execution speed is about
1
√
λ M in each intersection (λ is a constant and λ > 0).
That is,
√
I|Ci | = O( M) (i = 1 · · · N) (12)

where I|Ci | denotes the pixel position of C.

According to√the analysis mentioned above, the time complexity of the whole program
is S|C| = O(N M), while O (N × M) steps are needed when√it comes to classical com-
putation. That is to say, the speed of the erosion algorithm is M times as fast as that of
classical algorithm.

5 Conclusions

In this paper, a quantum method of dilation and erosion has been described. The quantum
dilation or quantum
√ erosion proposed in our algorithm both have achieved the time com-
plexity with O( MN). The time complexity is much smaller than the classical method
for the quantum computing parallelism. What’s more, our algorithm is more efficient than
some other improved algorithms. For example, the improved algorithm of the erosion √ and
the dilation which is put forward by the authors [22]. Its time complexity is O(2 MN).
Based on those properties, we have reason to believe that our algorithms have more obvious
superiority than classical algorithm. Furthermore, we can also combine the quantum dila-
tion and quantum erosion to realize the operation of morphological opening and closing.
This paper only talks about the binary image. The target of the next research will focus on
the gray-scale image. It will bring us more broad application prospects.

Acknowledgments This work is supported by the National Natural Science Foundation of China under
Grant No.61463016,61340029, Program for New Century Excellent Talents in University under Grant
No.NCET-13-0795, Landing project of science and technique of colleges and universities of Jiangxi Province
under Grant No.KJLD14037, Humanities and Social Sciences planning project of Ministry of Education
under Grant No.12YJAZH050, Project of International Cooperation and Exchanges of Jiangxi Province under
Grant No.20141BDH80007,Project of the science and technique funds of Nanchang City Grant No.2012-
KJZC-GY-CXYHZKF-001, “Control Science and Engineering” high-level discipline of Jiangxi Province
and Key Laboratory of Advanced Control & Optimization of Jiangxi Province, the National Natural Science
Foundation of China under Grant No.61462026, the item of science and technology awarded by Education
Bureau of Jiangxi Province under Grant No.GJJ14395.
Int J Theor Phys

References

1. Le, P.Q., Iliyasu, A.M., Dong, F., et al.: A Flexible Representation and Invertible Transformations
for Images on Quantum Computers[M]//New Advances in Intelligent Signal Processing, pp. 179–202.
Springer, Berlin Heidelberg (2011)
2. Li, H.S., Zhu, Q., Zhou, R.G., et al.: Multi-dimensional color image storage and retrieval for a normal
arbitrary quantum superposition state[J]. Quantum Inf. Process. 13(4), 991–1011 (2014)
3. Zhou, R., Wu, Q., Zhang, M.-Q., Shen, C.-Y.: Application of quantum gray-scale image based on pixel
mapping and pointer storing[J]. J. Comput. Inf. Syst. 8(19), 8143–8150 (2012)
4. Jones, J.A., Mosca, M., Hansen, R.H.: Implementation of a quantum search algorithm on a quantum
computer[J]. Nature 393(6683), 344–346 (1998)
5. Pang, C.-Y.: Quantum image compression, Postdoctoral report, Key Laboratory of Quantum Information,
University of Science and Technology of China (CAS) Hefei, China (2006)
6. Li, H.S., Qingxin, Z., Lan, S., et al.: Image storage, retrieval, compression and segmentation in a quantum
system[J]. Quantum Inf. Process. 12(4), 2269–2290 (2013)
7. Tseng, C.C., Hwang, T.M.: Quantum digital image processing algorithms[C]. In: 16th IPPR Conference
on Computer Vision, Graphics and Image Processing., pp. 827–834 (2003)
8. Rosenberg, B.: Quantum Fourier Transforms[J] (2003)
9. Hales, L., Hallgren, S.: An improved quantum Fourier transform algorithm and applications[C]. In:
Proceedings. 41st Annual Symposium on Foundations of Computer Science, pp. 515–525. IEEE (2000)
10. Weinstein, Y.S., Pravia, M.A., Fortunato, E.M., et al.: Implementation of the quantum Fourier trans-
form[J]. Phys. Rev. Lett. 86(7), 1889 (2001)
11. Fijany, A., Williams, C.P.: Quantum wavelet transforms[R] (1998)
12. Fijany, A., Williams, C.P.: Quantum wavelet transforms: Fast algorithms and complete circuits[M].
Springer, Berlin Heidelberg (1999)
13. Ruskai, M.B., Beylkin, G., Coifman, R.: Wavelets and their applications[J]. In: Ruskai, M.B., Beylkin,
G., Coifman, R. (eds.) Jones and Bartlett Books in Mathematics, vol. 1. Jones and Bartlett, Boston (1992)
14. Grover, L.K.: Quantum mechanics helps in searching for a needle in a haystack[J]. Phys. Rev. Lett.
79(2), 325 (1997)
15. Ke-fu, X., An, L.: Research on quantum-inspired mathematical morphology[J]. ACTA Electron. Sin.
33(2), 284–287 (2005)
16. Xie, K., Xu, W.: Quantum-inspired image decomposition and edge detection[J]. Jisuanji Yingyong/ J.
Comput. Appl. 33(4), 1089–1091 (2013)
17. Eldar, Y.C., Oppenheim, A.V.: Quantum signal processing[J]. Signal Process. Mag. IEEE 19(4), 12–32
(2002)
18. Pang, C.Y.: Loading N-dimensional vector into quantum registers from classical memory
withO(logN)Steps. arXiv:quant-ph/0612061 (2006)
19. Rungta, P.: The quadratic speedup in Grover’s search algorithm from the entanglement perspective. Phys.
Lett. A 373(31), 2652–2659 (2009)
20. Nielsen, M.A., Chuang, I.L.: Quantum computation and quantum information[M]. Cambridge University
Press, Cambridge (2010)
21. Gonzalez, R.C., Woods, R.E., Eddins, S.L.: Digital image processing using MATLAB[J]. Pearson
Prentice Hall, Upper Saddle River (2004)
22. Ragnemalm, I.: Fast erosion and dilation by contour processing and thresholding of distance maps[J].
Pattern Recog. Lett. 13(2), 161–166 (1992)
23. Boyer, M., Brassard, G., Høyer, P., et al.: Tight bounds on quantum searching[J]. arXiv:quan-
t-ph/9605034 (1996)
24. Hu, B.Q., Huang, X.D., Zhou, R.G., et al.: A theoretical framework for quantum image representation
and data loading scheme[J]. Sci. China Inf. Sci. 57(2), 1–11 (2014)