U.S. patent application number 17/413729 was filed with the patent office on 2022-02-10 for x-ray single-pixel camera based on x-ray computational correlated imaging.
This patent application is currently assigned to INSTITUTE OF PHYSICS, CHINESE ACADEMY OF SCIENCES. The applicant listed for this patent is INSTITUTE OF PHYSICS, CHINESE ACADEMY OF SCIENCES. Invention is credited to Liming CHEN, Yuhang HE, Bingbing WANG, Lingan WU, Aixin ZHANG.
Application Number | 20220042928 17/413729 |
Document ID | / |
Family ID | 1000005969378 |
Filed Date | 2022-02-10 |
United States Patent
Application |
20220042928 |
Kind Code |
A1 |
ZHANG; Aixin ; et
al. |
February 10, 2022 |
X-RAY SINGLE-PIXEL CAMERA BASED ON X-RAY COMPUTATIONAL CORRELATED
IMAGING
Abstract
An X-ray single-pixel camera based on X-ray computational
correlated imaging, which belongs to the technical research fields
of X-ray computational correlated imaging and X-ray single-pixel
imaging. The X-ray single-pixel camera includes: an X-ray
modulation system (3), an X-ray modulation control system (4), an
X-ray single-pixel detector (5), a main control system unit (6), a
time synchronization system (7) and a computational imaging system
(8). The main control system unit (6) controls each module through
software; the time synchronization system (7) controls
synchronization of each module for automatic collection; and the
computational imaging system (8) is configured to perform a
second-order correlated computation or a compressed sensing
computation or a deep learning computation on the signals collected
by the X-ray single-pixel detector (5) and a preset modulation
matrix, so as to obtain an image of an object under test. The X-ray
single-pixel camera based on X-ray computational correlated
imaging, provided by the present invention, realizes single-pixel
imaging, greatly reduces the sampling number while ensuring the
imaging quality, and reduces the X-ray radiation dose in an imaging
process.
Inventors: |
ZHANG; Aixin; (Beijing,
CN) ; HE; Yuhang; (Beijing, CN) ; WU;
Lingan; (Beijing, CN) ; CHEN; Liming;
(Beijing, CN) ; WANG; Bingbing; (Beijing,
CN) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
INSTITUTE OF PHYSICS, CHINESE ACADEMY OF SCIENCES |
Beijing |
|
CN |
|
|
Assignee: |
INSTITUTE OF PHYSICS, CHINESE
ACADEMY OF SCIENCES
Beijing
CN
|
Family ID: |
1000005969378 |
Appl. No.: |
17/413729 |
Filed: |
December 10, 2019 |
PCT Filed: |
December 10, 2019 |
PCT NO: |
PCT/CN2019/124392 |
371 Date: |
June 14, 2021 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G01N 2223/42 20130101;
G01N 2223/306 20130101; G01N 23/043 20130101; G01N 2223/03
20130101; G01N 2223/1016 20130101; G01N 23/20 20130101; G01N
2223/401 20130101 |
International
Class: |
G01N 23/04 20060101
G01N023/04; G01N 23/20 20060101 G01N023/20 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 14, 2018 |
CN |
201811536901.7 |
Claims
1. An X-ray single-pixel camera based on X-ray computational
correlated imaging, characterized by comprising: an X-ray
modulation system, an X-ray modulation control system, an X-ray
single-pixel detector, a main control system unit, a time
synchronization system and a computational imaging system, wherein
the X-ray modulation system is configured to modulate X-rays; the
X-ray modulation control system is configured to control the X-ray
modulation system; the X-ray single-pixel detector is configured to
collect signals; the main control system unit controls each module
through software; the time synchronization system controls
synchronization of each module for automatic collection; and the
computational imaging system is configured to perform a
second-order correlated computation or a compressed sensing
computation or a deep learning computation on the signals collected
by the X-ray single-pixel detector and a preset modulation matrix,
so as to obtain an image of an object under test.
2. The X-ray single-pixel camera according to claim 1,
characterized by further comprising an X-ray source for emitting
X-rays, wherein an object under test is placed between the X-ray
source and the X-ray modulation system for exposure.
3. The X-ray single-pixel camera according to claim 1,
characterized by further comprising an X-ray source for emitting
X-rays, wherein an object under test is placed between the X-ray
modulation system and the X-ray single-pixel detector for
exposure.
4. The X-ray single-pixel camera according to claim 1,
characterized in that the X-ray modulation system comprises a
modulation matrix, and the modulation matrix comprises a plurality
of matrix units; any one of the matrix units is hollowed out with
different preset patterns on an X-ray absorption material; and the
X-ray modulation control system controls a movement of the X-ray
modulation system, so that X-rays irradiate one of the matrix units
to form an X-ray pattern with the same distribution as the preset
patterns.
5. The X-ray single-pixel camera according to claim 4,
characterized in that the X-ray modulation system comprises a
modulation matrix, and the modulation matrix comprises a plurality
of matrix units; any one of the matrix units is made of a material
performing a phase modulation on X-rays; and the X-ray modulation
control system controls a movement of the X-ray modulation system,
so that X-rays irradiate one of the matrix units to form an X-ray
pattern distributed corresponding to the matrix unit.
6. The X-ray single-pixel camera according to claim 4,
characterized in that the X-ray absorption material is a material
capable of absorbing X-rays, comprising iron and an elemental
simple substance with a high atomic number after iron in a periodic
table of elements or a compound thereof; the simple substance
comprises any one of iron, cobalt, nickel, copper, zinc,
molybdenum, silver, cadmium, tin, tantalum, tungsten, platinum,
gold and lead; and the compound comprises any one of iron oxide,
copper oxide, zinc oxide and silver iodide.
7. The X-ray single-pixel camera according to claim 4,
characterized in that the main control system unit triggers the
X-ray modulation control system through software control so that
X-rays irradiate different matrix units of the X-ray modulation
system, and the X-ray single-pixel detector is controlled through
software to perform a signal collection; and the time
synchronization system sets a time sequence to enable the software
to first trigger the X-ray modulation control system so that X-rays
irradiate different matrix units of the X-ray modulation system,
and then the software is set to trigger the X-ray single-pixel
detector to perform a signal collection.
8. The X-ray single-pixel camera according to claim 1,
characterized in that the image of the object under the
second-order correlated computation is obtained by the following
formula: Image .times. .times. ( .eta. , .xi. ) = I .function. (
.eta. , .xi. ) .times. S - I .function. ( .eta. , .xi. ) .times. S
= i = 1 N .times. .times. I i .function. ( .eta. , .xi. ) .times. S
i N - i = 1 N .times. I i .function. ( .eta. , .xi. ) N .times. i =
1 N .times. S i N , ##EQU00005## wherein I.sub.i(.eta., .xi.) is
each set modulation matrix; i is a positive integers less than a
total sampling number N; and the light intensity detected by a
bucket detector after each corresponding modulation matrix
irradiates the object is denoted by S.sub.i.
9. The X-ray single-pixel camera according to claim 1,
characterized in that a collection process of the compressed
sensing is a linear projection process as shown below: A = ( I 1 ,
1 I 1 , 2 I 1 , M I 2 , 1 I 2 , 2 I 2 , M I N , 1 I N , 2 I N , M )
##EQU00006## y = Ax .times. ( y 1 y 2 y N ) = ( I 1 , 1 I 1 , 2 I 1
, M I 2 , 1 I 2 , 2 I 2 , M I N , 1 I N , 2 I N , M ) .times. ( x 1
x 2 x M ) , ##EQU00006.2## wherein in N measurements, an M pixel
image that represents an object can be represented by a
one-dimensional vector x=(x.sub.1, x.sub.2, . . . , x.sub.M); A is
a two-dimensional matrix representing the modulation matrix
I.sub.i(.eta., .xi.); S.sub.i is the light intensity detected each
time; and S.sub.i is represented by a one-dimensional vector
y=(y.sub.1, y.sub.2, . . . , y.sub.N).
10. The X-ray single-pixel camera according to claim 1,
characterized in that the deep learning computation comprises the
following steps: inputting a series of functions as models to be
trained; evaluating a quality of each function using an error rate
as a standard; and comparing an output of each function with a
correct result to select an optimal matching function.
Description
TECHNICAL FIELD
[0001] The present invention relates to the technical research
fields of X-ray computational correlated imaging and X-ray
single-pixel imaging, and in particular, to an X-ray single-pixel
camera based on X-ray computational correlated imaging.
BACKGROUND ART
[0002] As a light source with high penetrability, X-rays can
quickly realize non-invasive imaging of samples. X-ray imaging, as
a powerful imaging diagnostic technology, has been widely used in
the fields of industry, medicine and basic scientific research.
However, as a type of electromagnetic wave of short wavelength and
high photon energy, X-rays can cause radiation damage when
obtaining the internal structure of a sample. For example, in the
field of medicine, the probability of cancerization of cells
receiving too much radiation will greatly increase. Therefore, how
to reduce the radiation dose of X-rays while ensuring the image
quality is a problem that people are concerned about. In addition,
a large-area array high-pixel X-ray detector has a complex
structure, and its high cost can bring certain economic pressure to
the fields of industry, medicine and basic scientific research.
[0003] X-ray correlated imaging combined with correlated imaging
technology can well solve the problem of high-energy radiation of
X-rays, and thus reduce the requirements for an X-ray detector.
Intensity correlated imaging, as an indirect imaging manner, has
developed rapidly because of its unique properties since it was
first implemented in the laboratory using quantum light sources in
1995. This non-localized imaging manner subverts the perception of
people for traditional imaging: an image of an object can be
retrieved by beam splitting determination or pre-setting of the
distribution of a light field irradiating the object, and then
performing statistical correlated computation on the distribution
of the light field and the total transmitted (or reflected) light
intensity passing through the object. In this way, the light energy
does not have to be distributed on each pixel of an area array
detector, so as to improve the intensity of the light signal of the
object, thus reducing the influence of shot noise, and improving
the signal-to-noise ratio. The difference in imaging principle
makes the ghost imaging method capable of realizing not only
ultra-high resolution imaging but also imaging under extremely low
light compared with traditional imaging. However, the wavelength of
X-rays is too short, and there is no suitable spatial light
modulation device to perform known and controllable modulation on
the light field, and therefore, the currently reported experiments
on X-ray ghost imaging all use randomly modulated pseudothermal
light; not only is a large-area array X-ray detector required to
pre-measure the modulated light field, but also a large number of
exposure frames are required to restore the image of the object,
and the imaging quality is poor.
SUMMARY OF THE INVENTION
[0004] One objective of the present invention is to provide an
X-ray single-pixel camera based on X-ray computational correlated
imaging aimed at the above defects existing in the prior art, which
can perform a controllable and specific modulation on X-rays and
realize X-ray computational correlated imaging.
[0005] Another objective of the present invention is to provide an
X-ray single-pixel camera based on X-ray computational correlated
imaging, so as to realize single-pixel imaging, greatly reduce the
number of exposure frames while ensuring the imaging quality, and
reduce the X-ray radiation dose in an imaging process.
[0006] In particular, the present invention provides an X-ray
single-pixel camera based on X-ray computational correlated
imaging, including: an X-ray modulation system, an X-ray modulation
control system, an X-ray single-pixel detector, a main control
system unit, a time synchronization system and a computational
imaging system.
[0007] The X-ray modulation system is configured to modulate
X-rays; the X-ray modulation control system is configured to
control the X-ray modulation system; the X-ray single-pixel
detector is configured to collect signals; the main control system
unit controls each module through software; the time
synchronization system controls synchronization of each module for
automatic collection; and the computational imaging system is
configured to perform a second-order correlated computation or a
compressed sensing computation or a deep learning computation on
the signals collected by the X-ray single-pixel detector and a
preset modulation matrix, so as to obtain an image of an object
under test.
[0008] Optionally, the X-ray single-pixel camera further includes
an X-ray source for emitting X-rays.
[0009] Optionally, an object under test is placed between the X-ray
source and the X-ray modulation system for exposure.
[0010] Optionally, an object under test is placed between the X-ray
modulation system and the X-ray single-pixel detector for
exposure.
[0011] Optionally, the X-ray modulation system includes a
modulation matrix, and the modulation matrix includes a plurality
of matrix units; any one of the matrix units is hollowed out with
different preset patterns on an X-ray absorption material; and the
X-ray modulation control system controls a movement of the X-ray
modulation system, so that X-rays irradiate one of the matrix units
to form an X-ray pattern with the same distribution as the preset
patterns.
[0012] Optionally, the X-ray modulation system includes a
modulation matrix, and the modulation matrix includes a plurality
of matrix units; any one of the matrix units is made of a material
performing a phase modulation on X-rays; and the X-ray modulation
control system controls a movement of the X-ray modulation system,
so that X-rays irradiate one of the matrix units to form an X-ray
pattern distributed corresponding to the preset patterns.
[0013] Optionally, the X-ray absorption material is a material
capable of absorbing X-rays, including iron and an elemental simple
substance with a high atomic number after iron in a periodic table
of elements or a compound thereof; the simple substance includes
any one of iron, cobalt, nickel, copper, zinc, molybdenum, silver,
cadmium, tin, tantalum, tungsten, platinum, gold, and lead; and the
compound includes any one of iron oxide, copper oxide, zinc oxide
and silver iodide.
[0014] Optionally, the main control system unit triggers the X-ray
modulation control system through software control so that X-rays
irradiate different matrix units of the X-ray modulation system,
and the X-ray single-pixel detector is controlled through software
to perform a signal collection.
[0015] The time synchronization system sets a time sequence to
enable the software to first trigger the X-ray modulation control
system so that X-rays irradiate different matrix units of the X-ray
modulation system, and then, the software is set to trigger the
X-ray single-pixel detector to perform a signal collection.
[0016] Optionally, the image of the object under the second-order
correlated computation is obtained by the following formula:
Image .times. .times. ( .eta. , .xi. ) = I .function. ( .eta. ,
.xi. ) .times. S - I .function. ( .eta. , .xi. ) .times. S = i = 1
N .times. .times. I i .function. ( .eta. , .xi. ) .times. S i N - i
= 1 N .times. I i .function. ( .eta. , .xi. ) N .times. i = 1 N
.times. S i N , ##EQU00001##
[0017] wherein I.sub.i(.eta., .xi.) is each set modulation matrix;
i is a positive integer less than a total sampling number N; and
the light intensity detected by a bucket detector after each
corresponding modulation matrix irradiates the object is denoted by
S.sub.i.
[0018] Optionally, a collection process of the compressed sensing
computation is a linear projection process as shown below:
A = ( I 1 , 1 I 1 , 2 I 1 , M I 2 , 1 I 2 , 2 I 2 , M I N , 1 I N ,
2 I N , M ) ##EQU00002## y = Ax .times. ( y 1 y 2 y N ) = ( I 1 , 1
I 1 , 2 I 1 , M I 2 , 1 I 2 , 2 I 2 , M I N , 1 I N , 2 I N , M )
.times. ( x 1 x 2 x M ) , ##EQU00002.2##
[0019] wherein in N measurements, an M pixel image that represents
an object can be represented by a one-dimensional vector
x=(x.sub.1, x.sub.2, . . . , x.sub.M); A is a two-dimensional
matrix representing the modulation matrix I.sub.i(.eta., .xi.);
S.sub.i is the light intensity detected each time; and S.sub.i is
represented by a one-dimensional vector y=(y.sub.1, y.sub.2, . . .
, y.sub.N). The problem of compressed sensing is to solve an
underdetermined system of equations y=A x based on the known
measurement value y and measurement matrix A, so as to obtain the
original signal M pixel image x.
[0020] Optionally, the deep learning computation includes the
following steps:
[0021] inputting a series of functions as models to be trained;
[0022] evaluating a quality of each function using an error rate as
a standard; and
[0023] comparing an output of each function with a correct result
to select an optimal matching function.
[0024] The X-ray single-pixel camera based on X-ray computational
correlated imaging, provided by the present invention, uses a
special measurement matrix to perform a controllable and known
modulation on an original X-ray image under test, or projects the
speckles modulated by the special measurement matrix on an object;
the total light intensity received by the X-ray single-pixel
detector together with the measurement matrix are subjected to an
intensity correlation algorithm for image restoration, so that the
requirements for imaging detectors can be greatly reduced in the
case of obtaining the same resolution, which is of great
significance for reducing the cost of X-ray imaging devices.
[0025] In addition, compared with a random measurement matrix, the
X-ray single-pixel camera based on X-ray computational correlated
imaging, provided by the present invention, has the advantages that
a special matrix can obtain an image with a higher
contrast-to-noise ratio while the number of measurements is less,
and can greatly reduce the number of measurements for an image with
the same contrast-to-noise ratio, thereby reducing the radiation
dose received by a sample, which is of great significance in the
medical field.
[0026] According to the following detailed descriptions of specific
embodiments of the present invention in conjunction with the
drawings, those skilled in the art will more clearly understand the
above and other objectives, advantages and features of the present
invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Some specific embodiments of the present invention are
described in detail below with reference to the drawings by way of
example and not limitation. The same reference numerals in the
drawings indicate the same or similar components or parts. Those
skilled in the art should understand that these drawings are not
necessarily drawn to scale. In the drawings:
[0028] FIG. 1 is a schematic composition diagram of an X-ray
single-pixel camera based on X-ray computational correlated imaging
according to an embodiment of the present invention.
[0029] FIG. 2 is a schematic composition diagram of an X-ray
single-pixel camera based on X-ray computational correlated imaging
according to another embodiment of the present invention.
[0030] FIG. 3 is a schematic structural diagram of a modulation
matrix of an X-ray modulation control system according to an
embodiment of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0031] The present invention will be further described below with
reference to the figures and implementation manners. However, the
protection scope of the present invention is not limited to the
following examples, and should include all of the contents in the
claims.
[0032] The inventors of the present invention found in research
that: in the prior art, experiments on X-ray ghost imaging all use
randomly modulated pseudothermal light; not only is a large-area
array X-ray detector required to pre-measure a modulated light
field, but also a large number of exposure frames are required to
reconstruct an image of an object, and the imaging effect is poor.
The reason is that the wavelength of X-rays is too short, and there
is no suitable spatial light modulation device to perform a known
and controllable modulation on the light field. Therefore, on the
basis of an in-depth research on how to perform a controllable and
specific modulation on X-rays, the inventors proposed the design of
the present invention by implementing a method of X-ray
computational correlated imaging. The design of the present
invention not only can really realize single-pixel imaging, but
also can greatly reduce the sampling number while ensuring the
imaging quality, and reduce the X-ray radiation dose in an imaging
process.
[0033] FIG. 1 is a schematic composition diagram of an X-ray
single-pixel camera based on X-ray computational correlated imaging
according to an embodiment of the present invention. FIG. 2 is a
schematic composition diagram of an X-ray single-pixel camera based
on X-ray computational correlated imaging according to another
embodiment of the present invention. FIG. 3 is a schematic
structural diagram of a modulation matrix of an X-ray modulation
control system according to an embodiment of the present invention.
The present invention will be described in detail below with
reference to FIG. 1 to FIG. 3. As shown in FIG. 1 and FIG. 2, an
X-ray single-pixel camera based on X-ray computational correlated
imaging, provided by the present invention, may generally include:
an X-ray modulation system 3, an X-ray modulation control system 4,
an X-ray single-pixel detector 5, a main control system unit 6, a
time synchronization system 7 and a computational imaging system 8.
The X-ray modulation system 3 is configured to modulate X-rays. The
X-ray modulation control system 4 is configured to control the
X-ray modulation system 3. The X-ray single-pixel detector 5 is
configured to collect signals. The main control system unit 6
controls each module through software. The time synchronization
system 7 controls synchronization of each module for automatic
collection. The computational imaging system 8 is configured to
perform a second-order correlated computation or a compressed
sensing computation or a deep learning computation on the signals
collected by the X-ray single-pixel detector 5 and a preset
modulation matrix, so as to obtain an image of an object 2 under
test.
[0034] Specifically, the X-ray single-pixel camera further includes
an X-ray source 1 for emitting X-rays. The X-ray source 1 emits
X-rays which irradiate the object 2 under test and then irradiate
the X-ray modulation system 3, so that the X-ray modulation system
3 modulates the X-rays in space or in phase to generate known and
controllable X-ray patterns. In an optional implementation manner,
the object 2 under test is placed between the X-ray source 1 and
the X-ray modulation system 3 for exposure. In another optional
implementation manner, the object 2 under test is placed between
the X-ray modulation system 3 and the X-ray single-pixel detector 4
for exposure.
[0035] The X-ray modulation system 3 includes a modulation matrix.
The modulation matrix includes a plurality of matrix units, and
every matrix unit is hollowed out with different preset patterns on
an X-ray absorption material. FIG. 3 shows a schematic structural
diagram of a modulation matrix of an X-ray modulation control
system. FIG. 3 includes a total of 4*4 matrix units, and each
matrix unit has a different preset pattern. The black part
represents the modulation material, including but not limited to
materials based on X-ray absorption. If the black part is a
material capable of absorbing X-rays, the material may be a metal
plate. The white part represents the hollowed part through which
X-rays can pass through. The X-ray modulation system 3 allows some
X-rays from the X-ray source to pass through each modulation
matrix, which aims to enable uniform X-rays to be subjected to
spatial modulation, so as to generate known and controllable X-ray
patterns to irradiate the object. Therefore, in order to generate
known and controllable X-ray patterns, the material of the X-ray
modulation system 3 needs to be capable of absorbing X-rays (such
as metals, for example copper, iron or gold), or has certain phase
modulation on X-rays. Here, a metal plate based on X-ray absorption
is taken as an example. First, the modulation matrixes need to be
hollowed out of the metal plate one by one as the X-ray modulation
system 3. Then, uniform X-rays irradiate a certain modulation
matrix of the X-ray modulation system 3, so that an X-ray pattern
with the same distribution can be formed behind the X-ray
modulation system 3. The X-ray modulation system 3 is moved through
the X-ray modulation control system 4 to enable X-rays to irradiate
different modulation matrixes of the X-ray modulation system 3, so
that different X-ray patterns with known distribution can be
formed.
[0036] The X-ray modulation control system 4 controls the movement
of the X-ray modulation system 3, so that X-rays irradiate one of
the matrix units to form an X-ray pattern which is the same as the
preset pattern or distributed corresponding to the matrix unit. The
X-ray absorption material is a material capable of absorbing
X-rays, including iron and an elemental simple substance with a
high atomic number after iron in the periodic table of elements or
a compound thereof. The simple substance includes, but is not
limited to, any one of iron, cobalt, nickel, copper, zinc,
molybdenum, silver, cadmium, tin, tantalum, tungsten, platinum,
gold and lead. The compound includes, but is not limited to, any
one of iron oxide, copper oxide, zinc oxide and silver iodide.
[0037] In another specific implementation solution, the modulation
matrix includes a plurality of matrix units, and each of the matrix
units is made of a material performing a phase modulation on
X-rays. The X-ray modulation control system 4 controls the movement
of the X-ray modulation system 3, so that X-rays irradiate one
matrix unit to form an X-ray pattern distributed corresponding to
the matrix unit. The matrix unit performs a predictable modulation
on the phase of X-rays, so that the obtained X-ray pattern is also
predictable. The material capable of performing a phase modulation
on X-rays includes glass and the like.
[0038] The main control system unit 6 controls each module through
software. That is, the X-ray modulation control system 4 is
triggered through software control so that X-rays irradiate
different matrix units of the X-ray modulation system 3, and the
X-ray single-pixel detector 5 is controlled through the software to
perform a signal collection. The time synchronization system 7 sets
a time sequence to enable the software to first trigger the X-ray
modulation control system 4 so that X-rays irradiate different
matrix units of the X-ray modulation system 3, and then, the
software is set to trigger the X-ray single-pixel detector 5 to
perform a signal collection.
[0039] In the X-ray single-pixel camera based on X-ray
computational correlated imaging, provided by the present
invention, each set modulation matrix is marked as I.sub.i(.eta.,
.xi.); i is a positive integer less than the total sampling number
N; and the light intensity detected by a bucket detector after each
corresponding modulation matrix irradiates the object is marked as
S.sub.i. The image of the object under the second-order correlated
computation can be obtained by the following formula:
Image .times. .times. ( .eta. , .xi. ) = I .function. ( .eta. ,
.xi. ) .times. S - I .function. ( .eta. , .xi. ) .times. S = i = 1
N .times. .times. I i .function. ( .eta. , .xi. ) .times. S i N - i
= 1 N .times. I i .function. ( .eta. , .xi. ) N .times. i = 1 N
.times. S i N . ##EQU00003##
[0040] The compressed sensing computation comprises: the collection
process, which is a linear projection process as shown below:
A = ( I 1 , 1 I 1 , 2 I 1 , M I 2 , 1 I 2 , 2 I 2 , M I N , 1 I N ,
2 I N , M ) ##EQU00004## y = Ax .times. ( y 1 y 2 y N ) = ( I 1 , 1
I 1 , 2 I 1 , M I 2 , 1 I 2 , 2 I 2 , M I N , 1 I N , 2 I N , M )
.times. ( x 1 x 2 x M ) , ##EQU00004.2##
[0041] wherein in N measurements, an M pixel image that represents
an object can be represented by a one-dimensional vector
x=(x.sub.1, x.sub.2, . . . , x.sub.M); A is a two-dimensional
matrix representing the modulation matrix I.sub.i(.eta., .xi.);
S.sub.i is the light intensity detected each time; and S.sub.i is
represented by a one-dimensional vector y=(y.sub.1, y.sub.2, . . .
, y.sub.N). The problem of compressed sensing is to solve an
underdetermined system of equations y=A x based on the known
measurement value y and measurement matrix A, so as to obtain the
original signal M pixel image x.
[0042] Deep learning is a method based on representative learning
of data in machine learning. The observed value (referring to the
image of the object in the present invention) can be expressed in
many ways, such as a vector of each pixel intensity value, or more
abstractly expressed as a series of edges, regions of specific
shapes, and the like. It is easier to learn tasks from examples
using certain specific representation methods, such as acquiring
object information from noise.
[0043] The process of deep learning is a process of high-level
abstraction of data using a plurality of processing layers to
obtain multiple non-linear transformation functions. For now, deep
learning is mainly combined with artificial neural networks, so the
deep learning algorithm framework here can also become a deep
neural network algorithm framework.
[0044] Deep learning simulates the working principle of the human
brain by constructing a deep neural network. The deep neural
network consists of an input layer, a plurality of hidden layers
and an output layer. Each layer has a plurality of neurons, and
there are connection weights between the neurons. Each neuron
simulates a human nerve cell, and the connection between nodes
simulates the connection between nerve cells. The deep learning
computation includes the following steps:
[0045] inputting a series of functions as models to be trained;
[0046] evaluating a quality of each function using an error rate as
a standard; and
[0047] comparing an output of each function with a correct result
to select an optimal matching function.
[0048] The principle of the working process of the X-ray
single-pixel camera based on X-ray computational correlated
imaging, provided by the present invention, is as follows: after
the X-ray source 1 irradiates the object 2 under test, an image
under test is projected onto the X-ray modulation system 3. The
X-ray modulation system 3 comprises a modulation material on which
all preset modulation matrices are engraved, wherein the modulation
material can absorb X-rays or perform a certain phase modulation on
X-rays. For example, the modulation material is a metal plate. The
smallest modulation unit in the modulation matrix determines the
resolution of the X-ray single-pixel camera. The X-ray modulation
control system 4 controls the image of the object under test to be
projected onto each modulation matrix on the X-ray modulation
system 3 in sequence, or the X-ray modulation control system 4
controls each modulation matrix on the X-ray modulation system 3 to
be projected onto the object under test, thereby forming a known
and controllable modulation on the image under test. The
single-pixel detector 5 is configured to collect the total light
intensity after the X-ray modulation system 3 modulates the image
under test. All modules have corresponding control software
integrated into the main control system unit 6, and the time
synchronization system 7 is designed through the sequence of
experimental logic to realize an automatic collection. After the
collection is completed, the computational imaging system 8
performs a second-order correlation computation or a compressed
sensing computation or a deep learning computation on the light
intensity sequence collected by the single-pixel detector 5 and the
preset modulation matrix (that is, the modulation matrix engraved
on the X-ray modulation system 3), and finally, the image of the
object under test is obtained.
[0049] The X-ray single-pixel camera based on X-ray computational
correlated imaging, provided by the present invention, uses a
special measurement matrix to perform a controllable and known
modulation on an original X-ray image under test, or projects the
speckles modulated by the special measurement matrix on an object;
the total light intensity received by the X-ray single-pixel
detector together with the measurement matrix are subjected to an
intensity correlation algorithm for image restoration, so that the
requirements for imaging detectors can be greatly reduced in the
case of obtaining the same resolution, which is of great
significance for reducing the cost of X-ray imaging devices.
[0050] In addition, compared with a random measurement matrix, the
X-ray single-pixel camera based on X-ray computational correlated
imaging, provided by the present invention, has the advantages that
a special matrix can obtain an image with a higher
contrast-to-noise ratio while the number of measurements is less,
or can greatly reduce the number of measurements for an image with
the same contrast-to-noise ratio, thereby reducing the radiation
dose received by a sample, which is of great significance in the
medical field.
[0051] So far, those skilled in the art should recognize that
although various exemplary embodiments of the present invention
have been shown and described in detail herein, many other
variations or modifications consistent with the principles of the
present invention still can be directly determined or derived
according to the disclosed contents of the present invention
without departing from the spirit and scope of the present
invention. Therefore, the scope of the present invention should be
understood and deemed to cover all these other variations or
modifications.
* * * * *