CN112882057B - Photon counting non-view three-dimensional imaging super-resolution method based on interpolation - Google Patents
Photon counting non-view three-dimensional imaging super-resolution method based on interpolation Download PDFInfo
- Publication number
- CN112882057B CN112882057B CN202110068358.8A CN202110068358A CN112882057B CN 112882057 B CN112882057 B CN 112882057B CN 202110068358 A CN202110068358 A CN 202110068358A CN 112882057 B CN112882057 B CN 112882057B
- Authority
- CN
- China
- Prior art keywords
- interpolation
- distribution data
- size
- photon distribution
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003384 imaging method Methods 0.000 title claims abstract description 83
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 44
- 238000005070 sampling Methods 0.000 claims description 80
- 230000009466 transformation Effects 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims 3
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000009825 accumulation Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 7
- 230000000007 visual effect Effects 0.000 description 7
- 238000001514 detection method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 241000283973 Oryctolagus cuniculus Species 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 241000143236 Idaea efflorata Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 238000001161 time-correlated single photon counting Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/89—Lidar systems specially adapted for specific applications for mapping or imaging
- G01S17/894—3D imaging with simultaneous measurement of time-of-flight at a 2D array of receiver pixels, e.g. time-of-flight cameras or flash lidar
Abstract
The invention discloses a photon counting non-view three-dimensional imaging super-resolution method based on interpolation. The method comprises the following two steps: step one: selecting a proper interpolation method for the low-resolution acquired data to finish data interpolation and obtain updated acquired data; step two: performing inversion reconstruction on the updated acquired data according to a non-view reconstruction algorithm to obtain a three-dimensional image with higher resolution; the method can effectively solve the problem that a non-vision imaging system based on photon counting technology needs a large amount of accumulation time when performing high-resolution and high-frame-rate imaging.
Description
Technical Field
The invention relates to the technical field of photoelectric imaging, in particular to the field of non-visual field imaging, and specifically relates to a photon counting non-visual field three-dimensional imaging super-resolution method based on interpolation.
Background
With the rapid development of single photon detection technology, three-dimensional imaging of Non-line-of-sight (NLOS) targets using Time-dependent photon counting (TCSPC) technology has been rapidly developed in recent decades. Compared with the traditional photoelectric detection means, the photo-counting technology can detect photocurrent with lower intensity than the thermal noise level of the photoelectric detector at room temperature (10) -14 W), the technique is particularly suitable for very low light imaging techniques such as non-field of view imaging and single photon imaging. However, since the detection of a single echo photon by a single photon detector belongs to probability detection, under such a detection system, the number of echo photons of a single pulse is often less than 1, and thus a large accumulation time (pulse number) is often required to acquire sufficiently rich target echo information. For an imaging system using a single point detector, the algebraic relation between the accumulated pulse number of a single pixel and the imaging resolution and the imaging frame rate in the imaging process is as follows:
where K represents the number of accumulated pulses over a single pixel during scanning, PRF represents the repetition rate of the pulsed laser, S represents the imaging frame rate (in frames per second), and M represents the imaging resolution, i.e., the number of sampling points.
It follows that in order to ensure that a single pixel has sufficient echo photons (i.e. a sufficient number of accumulated pulses) and that a high imaging resolution is required, the imaging frame rate is greatly reduced (i.e. the imaging time of a single frame image is greatly increased). For example, the imaging system developed by Lindell D B, wetzstein G, O' tool M.wave-based non-line-of-sight imaging using fast f-k-means [ J ]. ACM Transactions On Graphics (TOG), 2019,38 (4): 1-13., lindell D B et al, shows real-time imaging (64X 64/2Hz; 32X 32/4 Hz) at lower resolution in an indoor dark environment, however, in order to three-dimensionally image a target in an outdoor scene, it takes 50 minutes to acquire data, because: 1. as an extremely weak light imaging, the signal-to-noise ratio is seriously reduced by extremely strong background noise in the outdoor environment, so that a large amount of accumulated pulses are required for each pixel to improve the signal-to-noise ratio of echo signals; 2. high resolution imaging means a dramatic increase in the number of sampling points, i.e., an increase in the total data acquisition time, and other similar typical cases such as document O' Toole M, lindell D B, wetzstein G.Confocal non-line-of-sight imaging based on the light-cone transform [ J ]. Nature,2018,555 (7696):338-341 and Liu X, guillen I, la Manna M, et al, non-line-of-sight imaging using phasor-field virtual wave optics [ J ]. Nature,2019,572 (7771):620-623.
A good solution to this problem is to use an area array single photon detector, i.e. by means of a detector array of size mxm, the same imaging resolution can be obtained by one shot without the need for a time-consuming and long scanning process. In the existing work, the documents Nam J H, brandt E, bauer S, et al real-time Non-line-of-Sight imaging of dynamic scenes [ J ]. ArXiv preprint arXiv:2010.12737,2020. Adopt a 16X 1 array detector, assisted by a special scanning mode, realize real-time Non-field imaging. However, limited by the manufacturing process and cost of large-size array detectors, there are significant difficulties with this approach to high resolution, high frame rate, non-field-of-view imaging techniques.
Disclosure of Invention
The invention aims to provide a photon counting non-view three-dimensional imaging super-resolution method, which aims to solve the problem that a non-view imaging system based on a single-point single photon detector needs a large amount of accumulated time when performing high-resolution and high-frame-rate imaging. The main operation process is that after the imaging system finishes data acquisition, one or more times of interpolation is carried out on the acquired three-dimensional data, namely, part of sampling process of the system is replaced by interpolation calculation, and then the interpolated data is analyzed by an inverse algorithm, so that high-resolution reconstruction of the target is completed.
The technical scheme of the invention is to provide a photon counting non-visual field three-dimensional imaging super-resolution method, which is characterized by comprising the following steps:
step one, interpolation is carried out on original echo photon distribution data with the size of MxMxt acquired by an imaging system, and echo photon distribution data with the size of [ (M-1) × (n+1) +1] × [ (M-1) × (n+1) +1] ×t is obtained; wherein M is the data sampling point, t is the time, and N is the interpolation frequency;
and step two, taking echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t as the input of a three-dimensional reconstruction algorithm to obtain a three-dimensional image with higher resolution.
Further, the first step may use a nearest neighbor interpolation algorithm, specifically:
step 1.1, selecting any one of four adjacent domain sampling points for original echo photon distribution data with the size of MxMxt acquired by an imaging system, and respectively marking the corresponding echo photon distribution data as:
step 1.2, carrying out interpolation on the inside of a sampling point in a four adjacent domains for N times, and obtaining new echo photon distribution data with the size of (2+N) x t after interpolation on echo photon distribution data with the size of 2 x t corresponding to the sampling point in the original four adjacent domains; according to the nearest interpolation principle, assigning values based on the sampling point nearest to the interpolation point; the interpolation data of the ith row and the jth column in the echo photon distribution data with the new size of (2+N) x t is recorded asThe expression is as follows:
and step 1.3, traversing all four adjacent domain sampling points in the original echo photon distribution data with the size of MxMxt, and finishing interpolation of the whole acquired data to obtain echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t.
Further, the first step may also adopt a bilinear interpolation algorithm, specifically:
step 1.1, selecting any one of four adjacent domain sampling points for original echo photon distribution data with the size of MxMxt acquired by an imaging system, and respectively marking the corresponding echo photon distribution data as:
step 1.2, carrying out interpolation on the inside of a sampling point in a four adjacent domains for N times, and obtaining new echo photon distribution data with the size of (2+N) x t after interpolation on echo photon distribution data with the size of 2 x t corresponding to the sampling point in the original four adjacent domains; according to the bilinear interpolation principle, different weights are given according to different distances from four neighborhood sampling points to interpolation points, and echo photon distribution data corresponding to the interpolation points are obtained by weighted average of echo photon distribution data corresponding to the four neighborhood sampling points; the interpolation data of the ith row and the jth column in the echo photon distribution data with the new size of (2+N) x t is recorded asThe expression is as follows:
and step 1.3, traversing all four adjacent domain sampling points in the original echo photon distribution data with the size of MxMxt, and finishing interpolation of the whole acquired data to obtain echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t.
Further, the first step may also adopt a bicubic interpolation algorithm, specifically:
step 1.1, selecting a BiCubic basis function to describe weight coefficients of each sampled data in sixteen adjacent domains, wherein the BiCubic basis function is in the following form:
wherein a= -0.5;
step 1.2, selecting any one of four adjacent domain sampling points for original echo photon distribution data with the size of MxMxt acquired by an imaging system, and respectively marking the corresponding echo photon distribution data as:
step 1.3, carrying out interpolation on the inside of a sampling point in a four adjacent domains for N times, and obtaining new echo photon distribution data with the size of (2+N) x t after interpolation on echo photon distribution data with the size of 2 x t corresponding to the sampling point in the original four adjacent domains; according to the bicubic interpolation principle, according to the difference of distances from sixteen neighborhood sampling points (4 multiplied by 4) to interpolation points, different weights are given, and echo photon distribution data corresponding to sixteen neighborhood sampling points are weighted and averaged to obtain interpolation point data; the interpolation data of the ith row and the jth column in the echo photon distribution data with the new size of (2+N) x t is recorded asThe expression is as follows:
wherein m is an integer from I-1 to I+2, and n is an integer from J-1 to J+2;
sampling points in four adjacent domains on the boundary are subjected to tri-linear interpolation, the condition that the number of reference sampling points is insufficient can occur, and at the moment, zero is assigned to the missing sampling points in the 4 multiplied by 4 reference points;
and step 1.4, traversing all four adjacent domain sampling points in the original echo photon distribution data with the size of MxMxt, and finishing interpolation of the whole acquired data to obtain echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t.
Further, in the second step, the three-dimensional reconstruction algorithm may use a frequency-wavenumber (f-k) offset algorithm, a Light Cone Transformation (LCT) algorithm, a Filtered Back Projection (FBP) algorithm, and the like.
The beneficial effects of the invention are as follows:
1. the photon counting non-visual field three-dimensional imaging super-resolution method disclosed by the invention can effectively solve the problem that the imaging time and the imaging resolution in the non-visual field imaging process are difficult to balance, and has more remarkable effect on a high-resolution imaging scene;
2. in the occasion that the requirement tends to imaging instantaneity, the method disclosed by the invention can be used for interpolating the acquired data by reducing the sampling points (reducing the imaging time) of the imaging system, so that the imaging effect with higher resolution can be obtained on the premise of not obviously reducing the imaging quality;
3. in the occasion of the requirement trend of imaging resolution, the method disclosed by the invention is used for interpolating the acquired data, so that higher imaging resolution can be obtained; the same can also be helpful in improving resolution in imaging systems employing array detectors.
Drawings
FIG. 1 is a schematic diagram of pixel interpolation principles of photon counting non-field of view three-dimensional imaging techniques;
FIG. 2 is a schematic diagram of a nearest neighbor interpolation method according to the first embodiment; wherein (a) represents a sampling point of a four-neighborhood domain arbitrarily selected from the original acquired data MxMxt, and (b) represents interpolation of the four-neighborhood domain by taking the four sampling points as samples and adopting a nearest-neighbor interpolation principle, wherein large circle spots represent sampling points of a system and small circle spots represent interpolation points;
FIG. 3 is a schematic diagram of a bilinear interpolation method according to a second embodiment; wherein (a) represents a sampling point of a four-neighborhood domain arbitrarily selected from the original acquired data MxMxt, and (b) represents interpolation of the four sampling points serving as samples by adopting bilinear interpolation principle, wherein large circle spots represent sampling points of a system, and small circle spots represent interpolation points;
fig. 4 is a schematic diagram of a bicubic interpolation method according to the third embodiment; wherein (a) represents sixteen neighbor sampling points arbitrarily selected from original acquired data MxMxt, and (b) represents interpolation of the four neighbor areas by taking the sixteen sampling points as samples and adopting the principle of bicubic interpolation, wherein large circle spots represent system sampling points and small circle spots represent interpolation points;
FIG. 5 is a three-dimensional simulation schematic of the application effect of an embodiment of the present invention; wherein, (a) is the result of 32×32 resolution imaging of a three-dimensional rabbit, the imaging time is 60s; (b) Sequentially performing single interpolation (n=1) on the data of the step (a) by adopting nearest neighbor interpolation, bilinear interpolation and bicubic interpolation to obtain a 63×63 three-dimensional image; (e) The imaging system directly acquires 63×63 three-dimensional imaging, and the imaging time of the image is about 240s;
FIG. 6 is a quantized representation of the effect of an embodiment of the invention; (a) Each of (a) to (e) is a front view corresponding to each of (a) to (e) in fig. 5.
Detailed Description
For the purposes of promoting an understanding of the principles and advantages of the disclosure, reference will now be made to the embodiments illustrated in the drawings and specific language will be used to describe the same.
Compared with the traditional data acquisition-analysis reconstruction, the photon counting non-view three-dimensional imaging super-resolution method based on interpolation provided by the invention has the advantages that the interpolation process for acquired data is added after the data acquisition and before the analysis reconstruction, and the imaging resolution can be effectively increased on the premise of not obviously reducing the imaging quality as shown in fig. 1. The photon counting non-visual field three-dimensional imaging super-resolution method based on interpolation comprises the following steps:
step one: interpolating three-dimensional echo photon distribution data acquired by an imaging system;
step two: and taking the interpolated data as the input of a reconstruction algorithm to obtain a three-dimensional image with higher resolution.
The echo photon distribution data acquired by photon counting non-field of view imaging is a three-dimensional matrix shaped as m×m×t, m×m representing the data sampling point (i.e. imaging resolution), and t representing time. The echo photon distribution data for each data sample point represents the photon time of flight and the corresponding echo photon number. In the non-visual field imaging process, the echo photon distribution data of each sampling point carries most of three-dimensional geometric information of the target, so that the resolution can be effectively improved by carrying out data interpolation on the information of the sampling points in the field.
Example 1
In this embodiment, the pixel interpolation method adopts the nearest neighbor interpolation algorithm, and is described with reference to fig. 2, which includes the following specific steps:
step 1: as shown in fig. 2 (a), for the original echo photon distribution data mxmxmxt, any one of the four adjacent domain sampling points is selected, and the corresponding echo photon distribution data are respectively written as:
step 2: as shown in fig. 2 (b), N times of interpolation is performed on the inside of the sampling points in the four neighboring domains, and the echo photon distribution data with the size of 2×2×t corresponding to the sampling points in the four neighboring domains is obtained after interpolation, so that new echo photon distribution data with the size of (2+n) × (2+n) ×t is obtained. According to the nearest interpolation principle, assigning values based on the sampling point nearest to the interpolation point, and recording the echo photon distribution data of the ith row and the jth column in (2+N) x t asThen->The expression of (2) is:
step 3: and traversing all four adjacent domain sampling points in the collected data MxMxt to finish interpolation of the whole collected data, and obtaining new data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t by the original size of MxMxt.
And taking the interpolated data as input of a reconstruction algorithm to obtain a three-dimensional image with higher resolution, wherein the three-dimensional reconstruction algorithm can adopt a frequency-wave number (f-k) offset algorithm, a Light Cone Transformation (LCT) algorithm, a Filtered Back Projection (FBP) algorithm and the like.
Example two
In this embodiment, the pixel interpolation method adopts bilinear interpolation algorithm, and is described with reference to fig. 3, which includes the following specific steps:
step 1: as shown in fig. 3 (a), for the original echo photon distribution data mxmxmxt, any one of the four adjacent domain sampling points is selected, and the corresponding echo photon distribution data are respectively written as:
step 2: as shown in fig. 3 (b), N times of interpolation is performed on the inside of the sampling points in the four adjacent domains, that is, the echo photon distribution data with the size of 2×2×t corresponding to the sampling points in the original four adjacent domains is interpolated to obtain new echo photon distribution data with the size of (2+n) × (2+n) ×t. According to the bilinear interpolation principle, different weights are given according to different distances from four neighborhood sampling points to interpolation points, and echo photon distribution data corresponding to the interpolation points are obtained by weighted average of echo photon distribution data corresponding to the four neighborhood sampling points. The echo photon distribution data of the ith row and the jth column in (2+N) x t are recorded asThen->The expression of (2) is:
step 3: and traversing all four adjacent domains in the collected data MxMxt to finish interpolation of the whole collected data, and obtaining new data with the size [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t by the original size of the sampled data with the size of MxMxt.
And taking the interpolated data as input of a reconstruction algorithm to obtain a three-dimensional image with higher resolution, wherein the three-dimensional reconstruction algorithm can adopt a frequency-wave number (f-k) offset algorithm, a Light Cone Transformation (LCT) algorithm, a Filtered Back Projection (FBP) algorithm and the like.
Example III
In this embodiment, the pixel interpolation method adopts a bicubic interpolation algorithm, and is described with reference to fig. 4, which specifically includes the following steps:
step 1: the BiCubic basis function is defined to describe the weight coefficients of each sampling point in sixteen neighborhoods in the form:
wherein a= -0.5;
step 2: as shown in fig. 4 (a), for the original echo photon distribution data mxmxmxt, any one of the four adjacent domain sampling points is selected, and the echo photon distribution data corresponding to the sampling points are respectively recorded as:
step 3: as shown in fig. 4 (b), N times of interpolation is performed on the inside of the sampling points in the four adjacent domains, that is, the echo photon distribution data with the size of 2×2×t corresponding to the sampling points in the original four adjacent domains is interpolated to obtain new echo photon distribution data with the size of (2+n) × (2+n) ×t. According to the bicubic interpolation principle, according to the difference of distances from sixteen neighborhood sampling points to interpolation points, different weights are given, and echo photon distribution data corresponding to sixteen neighborhood sampling points are weighted and averaged to obtain interpolation point data. The echo photon distribution data of the ith row and the jth column in (2+N) x t are recorded asThen->The expression of (2) is:
wherein m is an integer from I-1 to I+2, and n is an integer from J-1 to J+2;
in particular, when the sampling points in the four adjacent domains located on the boundary are subjected to tri-linear interpolation, the situation that the number of reference sampling points is insufficient occurs, and at this time, zero is assigned to the missing sampling points in the 4×4 reference points.
Step 4: and traversing all four adjacent domains in the collected data MxMxt to finish interpolation of the whole collected data, and obtaining data of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t from the original sampled data MxMxt.
And taking the interpolated data as input of a reconstruction algorithm to obtain a three-dimensional image with higher resolution, wherein the three-dimensional reconstruction algorithm can adopt a frequency-wave number (f-k) offset algorithm, a Light Cone Transformation (LCT) algorithm, a Filtered Back Projection (FBP) algorithm and the like.
The effect of the embodiment is verified by using three-dimensional simulation, as shown in fig. 5, wherein (a) is the result of 32×32 resolution imaging on a three-dimensional rabbit, and the imaging time is 60s; (b) Sequentially performing single interpolation (n=1) on the data of the step (a) by adopting nearest neighbor interpolation, bilinear interpolation and bicubic interpolation to obtain a 63×63 three-dimensional image; (e) Three-dimensional imaging of 63 x 63 is acquired directly by the imaging system, with an imaging time of about 240s. Fig. 6 is a quantitative representation of the effect of the embodiment of the present invention, where (a) to (e) are front views corresponding to (a) to (e) in fig. 5, respectively, and in order to characterize the imaging effect after interpolation by the embodiment of the present invention, two image quality evaluation indexes, namely, structural similarity (Structural Similarity, SSIM) and Peak signal-to-noise Ratio (PSNR), are used, and the resolution of 63×63 (b) - (d) and actual 63×63 (e) obtained by interpolation are compared, and the results are shown in the following table:
as can be seen from fig. 5 and fig. 6, the method of the present invention is used to obtain a 63×63 (i.e. the images (b) - (d) in fig. 5) and an actual 63×63 (the image (e) in fig. 5) image by super-resolving the original 32×32 resolution image, where the SSIM is above 0.8 and the PSNR is above 25dB, but the acquisition time can be shortened from 240s to 60s.
Claims (3)
1. The photon counting non-view three-dimensional imaging super-resolution method based on interpolation is characterized by comprising the following steps of:
step one, interpolation is carried out on original echo photon distribution data with the size of MxMxt acquired by an imaging system, and echo photon distribution data with the size of [ (M-1) × (n+1) +1] × [ (M-1) × (n+1) +1] ×t is obtained; wherein M is the data sampling point, t is the time, and N is the interpolation frequency; the method comprises the following steps:
step 1.1, selecting any one of four adjacent domain sampling points for original echo photon distribution data with the size of MxMxt acquired by an imaging system, and respectively marking the corresponding echo photon distribution data as:
step 1.2, carrying out interpolation on the inside of a sampling point in a four adjacent domains for N times, and obtaining new echo photon distribution data with the size of (2+N) x t after interpolation on echo photon distribution data with the size of 2 x t corresponding to the sampling point in the original four adjacent domains; according to the nearest interpolation principle, assigning values based on the sampling point nearest to the interpolation point; the interpolation data of the ith row and the jth column in the echo photon distribution data with the new size of (2+N) x t is recorded asThe expression is as follows:
step 1.3, traversing all four adjacent domain sampling points in original echo photon distribution data with the size of MxMxt, and finishing interpolation of the whole acquired data to obtain echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t;
step two, taking echo photon distribution data with the size of [ (M-1) x (N+1) +1] × [ (M-1) x (N+1) +1] ×t as input of a three-dimensional reconstruction algorithm to obtain a higher-resolution three-dimensional image; the three-dimensional reconstruction algorithm adopts a frequency-wave number f-k offset algorithm, a light cone transformation LCT algorithm or a filtering back projection FBP algorithm.
2. The photon counting non-view three-dimensional imaging super-resolution method based on interpolation is characterized by comprising the following steps of:
step one, interpolation is carried out on original echo photon distribution data with the size of MxMxt acquired by an imaging system, and echo photon distribution data with the size of [ (M-1) × (n+1) +1] × [ (M-1) × (n+1) +1] ×t is obtained; wherein M is the data sampling point, t is the time, and N is the interpolation frequency; the method comprises the following steps:
step 1.1, selecting any one of four adjacent domain sampling points for original echo photon distribution data with the size of MxMxt acquired by an imaging system, and respectively marking the corresponding echo photon distribution data as:
step 1.2, carrying out interpolation on the inside of a sampling point in a four adjacent domains for N times, and obtaining new echo photon distribution data with the size of (2+N) x t after interpolation on echo photon distribution data with the size of 2 x t corresponding to the sampling point in the original four adjacent domains; according to the bilinear interpolation principle, different weights are given according to different distances from four neighborhood sampling points to interpolation points, and echo light corresponding to the four neighborhood sampling points is givenThe sub-distribution data are weighted and averaged to obtain echo photon distribution data corresponding to interpolation points; the interpolation data of the ith row and the jth column in the echo photon distribution data with the new size of (2+N) x t is recorded asThe expression is as follows:
step 1.3, traversing all four adjacent domain sampling points in original echo photon distribution data with the size of MxMxt, and finishing interpolation of the whole acquired data to obtain echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t;
step two, taking echo photon distribution data with the size of [ (M-1) x (N+1) +1] × [ (M-1) x (N+1) +1] ×t as input of a three-dimensional reconstruction algorithm to obtain a higher-resolution three-dimensional image; the three-dimensional reconstruction algorithm adopts a frequency-wave number f-k offset algorithm, a light cone transformation LCT algorithm or a filtering back projection FBP algorithm.
3. The photon counting non-view three-dimensional imaging super-resolution method based on interpolation is characterized by comprising the following steps of:
step one, interpolation is carried out on original echo photon distribution data with the size of MxMxt acquired by an imaging system, and echo photon distribution data with the size of [ (M-1) × (n+1) +1] × [ (M-1) × (n+1) +1] ×t is obtained; wherein M is the data sampling point, t is the time, and N is the interpolation frequency; the method comprises the following steps:
step 1.1, selecting a BiCubic basis function to describe weight coefficients of each sampled data in sixteen adjacent domains, wherein the BiCubic basis function is in the following form:
wherein a= -0.5;
step 1.2, an imaging systemThe method comprises the steps of selecting any one of four adjacent domain sampling points from collected original echo photon distribution data with the size of MxMxt, and respectively marking the corresponding echo photon distribution data as:
step 1.3, carrying out interpolation on the inside of a sampling point in a four adjacent domains for N times, and obtaining new echo photon distribution data with the size of (2+N) x t after interpolation on echo photon distribution data with the size of 2 x t corresponding to the sampling point in the original four adjacent domains; according to the bicubic interpolation principle, according to the difference of distances from sixteen neighborhood sampling points (4 multiplied by 4) to interpolation points, different weights are given, and echo photon distribution data corresponding to sixteen neighborhood sampling points are weighted and averaged to obtain interpolation point data; the interpolation data of the ith row and the jth column in the echo photon distribution data with the new size of (2+N) x t is recorded asThe expression is as follows:
wherein m is an integer from I-1 to I+2, and n is an integer from J-1 to J+2;
sampling points in four adjacent domains on the boundary are subjected to tri-linear interpolation, the condition that the number of reference sampling points is insufficient can occur, and at the moment, zero is assigned to the missing sampling points in the 4 multiplied by 4 reference points;
step 1.4, traversing all four adjacent domain sampling points in original echo photon distribution data with the size of MxMxt, and finishing interpolation of the whole acquired data to obtain echo photon distribution data with the size of [ (M-1) × (N+1) +1] × [ (M-1) × (N+1) +1] ×t;
step two, taking echo photon distribution data with the size of [ (M-1) x (N+1) +1] × [ (M-1) x (N+1) +1] ×t as input of a three-dimensional reconstruction algorithm to obtain a higher-resolution three-dimensional image; the three-dimensional reconstruction algorithm adopts a frequency-wave number f-k offset algorithm, a light cone transformation LCT algorithm or a filtering back projection FBP algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110068358.8A CN112882057B (en) | 2021-01-19 | 2021-01-19 | Photon counting non-view three-dimensional imaging super-resolution method based on interpolation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110068358.8A CN112882057B (en) | 2021-01-19 | 2021-01-19 | Photon counting non-view three-dimensional imaging super-resolution method based on interpolation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112882057A CN112882057A (en) | 2021-06-01 |
| CN112882057B true CN112882057B (en) | 2023-12-08 |
Family
ID=76049666
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110068358.8A Active CN112882057B (en) | 2021-01-19 | 2021-01-19 | Photon counting non-view three-dimensional imaging super-resolution method based on interpolation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112882057B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112946990B (en) * | 2021-05-13 | 2021-08-31 | 清华大学 | Non-vision field dynamic imaging system based on confocal mode |
| CN113820726B (en) * | 2021-09-30 | 2023-06-13 | 中国科学院光电技术研究所 | Noise suppression method based on multidimensional filtering in non-visual field target detection |
| CN115993611B (en) * | 2023-03-22 | 2023-06-20 | 清华大学 | Non-visual field imaging method and device based on transient signal super-resolution network |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008151470A1 (en) * | 2007-06-15 | 2008-12-18 | Tsinghua University | A robust human face detecting method in complicated background image |
| CN104887266A (en) * | 2015-05-29 | 2015-09-09 | 西安交通大学 | Method for small-area three-dimensional passive cavitation imaging and three-dimensional composite imaging based on area array |
| CN106651938A (en) * | 2017-01-17 | 2017-05-10 | 湖南优象科技有限公司 | Depth map enhancement method blending high-resolution color image |
| CN106772428A (en) * | 2016-12-15 | 2017-05-31 | 哈尔滨工业大学 | A kind of non-ken three-dimensional image forming apparatus of no-raster formula photon counting and method |
| CN107798290A (en) * | 2017-09-14 | 2018-03-13 | 中国科学院西安光学精密机械研究所 | 3-D view SNR estimation and compensation and mixing regularization reconstructing method based on photon counting |
| CN110764088A (en) * | 2019-10-25 | 2020-02-07 | 哈尔滨工程大学 | Super-resolution stationary point scanning real-time imaging algorithm |
| WO2020124992A1 (en) * | 2018-12-19 | 2020-06-25 | 南京理工大学 | Aperture coding imaging system based on transmission-type dual slits, and super-resolution method therefor |
| CN111679290A (en) * | 2020-06-04 | 2020-09-18 | 上海禾赛光电科技有限公司 | Photon counting correction method, laser radar, and computer-readable medium |
| CN111694014A (en) * | 2020-06-16 | 2020-09-22 | 中国科学院西安光学精密机械研究所 | Laser non-visual field three-dimensional imaging scene modeling method based on point cloud model |
| CN112168196A (en) * | 2019-07-02 | 2021-01-05 | 通用电气精准医疗有限责任公司 | System and method for high resolution spectral computed tomography imaging |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7103235B2 (en) * | 2001-04-25 | 2006-09-05 | Lockheed Martin Corporation | Extended range image processing for electro-optical systems |
| JP6766045B2 (en) * | 2014-11-20 | 2020-10-07 | コーニンクレッカ フィリップス エヌ ヴェＫｏｎｉｎｋｌｉｊｋｅ Ｐｈｉｌｉｐｓ Ｎ．Ｖ． | How to generate a synthetic mammogram from tomosynthesis data |
| WO2019178311A1 (en) * | 2018-03-15 | 2019-09-19 | Redzone Robotics, Inc. | Image processing techniques for multi-sensor inspection of pipe interiors |
| US11609328B2 (en) * | 2018-05-23 | 2023-03-21 | Massachusetts Institute Of Technology | Methods and apparatus for improved imaging through scattering media |
-
2021
- 2021-01-19 CN CN202110068358.8A patent/CN112882057B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008151470A1 (en) * | 2007-06-15 | 2008-12-18 | Tsinghua University | A robust human face detecting method in complicated background image |
| CN104887266A (en) * | 2015-05-29 | 2015-09-09 | 西安交通大学 | Method for small-area three-dimensional passive cavitation imaging and three-dimensional composite imaging based on area array |
| CN106772428A (en) * | 2016-12-15 | 2017-05-31 | 哈尔滨工业大学 | A kind of non-ken three-dimensional image forming apparatus of no-raster formula photon counting and method |
| CN106651938A (en) * | 2017-01-17 | 2017-05-10 | 湖南优象科技有限公司 | Depth map enhancement method blending high-resolution color image |
| CN107798290A (en) * | 2017-09-14 | 2018-03-13 | 中国科学院西安光学精密机械研究所 | 3-D view SNR estimation and compensation and mixing regularization reconstructing method based on photon counting |
| WO2020124992A1 (en) * | 2018-12-19 | 2020-06-25 | 南京理工大学 | Aperture coding imaging system based on transmission-type dual slits, and super-resolution method therefor |
| CN112168196A (en) * | 2019-07-02 | 2021-01-05 | 通用电气精准医疗有限责任公司 | System and method for high resolution spectral computed tomography imaging |
| CN110764088A (en) * | 2019-10-25 | 2020-02-07 | 哈尔滨工程大学 | Super-resolution stationary point scanning real-time imaging algorithm |
| CN111679290A (en) * | 2020-06-04 | 2020-09-18 | 上海禾赛光电科技有限公司 | Photon counting correction method, laser radar, and computer-readable medium |
| CN111694014A (en) * | 2020-06-16 | 2020-09-22 | 中国科学院西安光学精密机械研究所 | Laser non-visual field three-dimensional imaging scene modeling method based on point cloud model |
Non-Patent Citations (7)
| Title |
|---|
| A High-Linearity, 17 ps Precision Time-to-Digital Converter Based on a Single-Stage Vernier Delay Loop Fine Interpolation;Bojan Markovic;IEEE Transactions on Circuits and Systems;557-569 * |
| Single Space Object Image Super Resolution Reconstructing Using Convolutional Networks in Wavelet Transform Domain;Xubin Feng.ETL;2020 IEEE 3rd international conference on Electronics technology（ICET);862-866 * |
| Super-Resolution Reconstruction Method for Single Space Object Image based on Optimized Convolution Neural Network;Xubin Feng.etl;2019 IEEE internationak conference on Unmanned Systems and Artificial Intelligence(ICUSAI));222-226 * |
| 光子脉冲外差探测系统的测距精度;罗韩君;元秀华;;中国激光(12);全文 * |
| 基于学习的图像超分辨率技术;黄丽;中国优秀硕士论文信息科技辑(第3期);6-12 * |
| 基于非降采样Contourlet的单帧图像超分辨率算法;张威;陈微;;微电子学与计算机;第30卷(第12期);73-76 * |
| 改进的峰值检测反距离加权插值算法;李超;陈钱;顾国华;钱惟贤;;红外与激光工程(02);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112882057A (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112882057B (en) | Photon counting non-view three-dimensional imaging super-resolution method based on interpolation | |
| CN108742532B (en) | Wide-field chromatographic ultra-spectral microscopic imaging method and device based on space-time focusing | |
| US11169367B2 (en) | Three-dimensional microscopic imaging method and system | |
| CN102466521B (en) | TDI (Time Delay Integration) scan imaging method for Snapshot type area array infrared detector | |
| CN105069748B (en) | A kind of method that high-definition picture is obtained based on microsatellite object space scanning technique | |
| RU2734447C2 (en) | System for forming a synthesized two-dimensional image of a biological sample with high depth of field | |
| CN1246929A (en) | Analog circuit for autofocus microscope system | |
| CN111343376B (en) | Transmission-type double-slit-based aperture coding imaging system and super-resolution method thereof | |
| CN107621463A (en) | Image rebuilding method, device and microscopic imaging device | |
| CN105911692A (en) | Optimal system parameter selection method for Fourier laminated micro imaging system | |
| CN114660060A (en) | Macroscopic Fourier laminated super-resolution imaging method based on matrix scanning | |
| CN116342452A (en) | Image generation method and fusion imaging system | |
| CN114241072B (en) | Laminated imaging reconstruction method and system | |
| CN112946990B (en) | Non-vision field dynamic imaging system based on confocal mode | |
| Dong et al. | Two-stage convolutional neural network for joint removal of sensor noise and background interference in lock-in thermography | |
| CN104143179B (en) | Method for enhancing moving target through multi-linear-array time difference scanning expansion sampling | |
| Guo et al. | Ghost imaging of the low or high frequency based on the corresponding spatial-frequency of the reference pattern | |
| CN108132266A (en) | A kind of x-ray light path cascades micro imaging system | |
| Wang et al. | Infrared image super-resolution enhancement based on convolutional neural network | |
| CN114460805A (en) | Shielding scattering imaging system based on high-pass filtering | |
| KR20140045625A (en) | Structured illumination imaging method using laser bean scanning | |
| CN117333574B (en) | High-speed image acquisition processing method and application thereof in scanning imaging device | |
| Nickerson et al. | Scan conversion of radar images | |
| WO2022075954A1 (en) | A calibration method having a super-resolution for magnetic particle imaging | |
| CN111601096B (en) | Image synthesis method with single photon avalanche diode |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |