CN112229514B - Three-dimensional data reconstruction method of liquid crystal hyperspectral calculation imaging system - Google Patents
Three-dimensional data reconstruction method of liquid crystal hyperspectral calculation imaging system Download PDFInfo
- Publication number
- CN112229514B CN112229514B CN202011054740.5A CN202011054740A CN112229514B CN 112229514 B CN112229514 B CN 112229514B CN 202011054740 A CN202011054740 A CN 202011054740A CN 112229514 B CN112229514 B CN 112229514B
- Authority
- CN
- China
- Prior art keywords
- liquid crystal
- data
- hyperspectral
- response
- imaging system
- 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
- 239000004973 liquid crystal related substance Substances 0.000 title claims abstract description 62
- 238000003384 imaging method Methods 0.000 title claims abstract description 47
- 238000004364 calculation method Methods 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 title claims abstract description 22
- 230000004044 response Effects 0.000 claims abstract description 73
- 230000003595 spectral effect Effects 0.000 claims abstract description 46
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 8
- 230000006835 compression Effects 0.000 claims description 17
- 238000007906 compression Methods 0.000 claims description 17
- 238000001228 spectrum Methods 0.000 claims description 9
- 238000005259 measurement Methods 0.000 claims description 4
- 230000003287 optical effect Effects 0.000 claims description 4
- 239000000835 fiber Substances 0.000 claims description 3
- 210000001747 pupil Anatomy 0.000 claims description 3
- 238000005215 recombination Methods 0.000 claims description 3
- 230000006798 recombination Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 2
- 238000012847 principal component analysis method Methods 0.000 claims description 2
- 238000000701 chemical imaging Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 abstract description 4
- 239000011159 matrix material Substances 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000008521 reorganization Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
- G01J3/02—Details
- G01J3/0205—Optical elements not provided otherwise, e.g. optical manifolds, diffusers, windows
- G01J3/0229—Optical elements not provided otherwise, e.g. optical manifolds, diffusers, windows using masks, aperture plates, spatial light modulators or spatial filters, e.g. reflective filters
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
- G01J3/28—Investigating the spectrum
- G01J3/2823—Imaging spectrometer
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Spectrometry And Color Measurement (AREA)
- Image Processing (AREA)
Abstract
The invention provides a three-dimensional data reconstruction method of a liquid crystal hyperspectral computed imaging system, which comprises the steps of constructing a convolutional neural network suitable for the liquid crystal hyperspectral computed imaging system, taking a compressed observation result and system response obtained by the computed imaging system as network input together, and finally outputting reconstructed high-resolution three-dimensional data through a plurality of hidden layers; wherein the system response comprises a spatial response and a spectral response of the system, respectively representing the spatial and spectral encoding effect of the system on the incident scene. The invention carries out the calculation and reconstruction of the compressed observation data under the framework of the convolutional neural network, simultaneously considers the compressed data and the system response, and under the condition of enough training data, the network can adapt to different coding templates and various types of calculation spectral imaging systems, thereby rapidly and accurately acquiring the reconstructed three-dimensional data.
Description
Technical Field
The invention relates to the field of three-dimensional data reconstruction, in particular to a three-dimensional data reconstruction method of a liquid crystal hyperspectral calculation imaging system.
Background
In the patent ZL201610920079.9, the invention provides a three-dimensional coded liquid crystal hyperspectral calculation imaging measurement device and a measurement method, the device can simultaneously realize the spatial and spectral dimension compression coding of hyperspectral data, reduce the data dimension in the data acquisition stage, avoid the data redundancy and improve the information utilization rate. The invention uses a front lens to converge the light of a target scene into a system, a liquid crystal tunable filter is used as a wave band selection and light splitting module of the system, the information of the selected wave band in the incident light is transmitted, the other wave bands are filtered, the emergent light passing through the liquid crystal tunable filter is subjected to space coding in a space coding module, and the coded information is subjected to aliasing imaging on an area array detector through a collimating lens. The role of the liquid crystal tunable filter in the device is to modulate and encode the spectral dimensions, including band selection and light splitting. The spatial encoding of the incident scene by the system is achieved by means of a spatial light modulator.
In the invention patent ZL201810752547.5, a three-dimensional data reconstruction method based on a liquid crystal hyperspectral calculation imaging system is provided. The method carries out mathematical modeling on the functions of each module in the liquid crystal hyperspectral calculation imaging system, expresses the imaging process of the system by using a mathematical language, converts the data acquisition process of the liquid crystal hyperspectral calculation imaging system into a matrix form by using discretization representation of data, combines a compressive sensing principle, deduces how to reconstruct high-resolution three-dimensional data from a small amount of data acquired by the system, and provides a reconstruction algorithm. However, since the amount of the hyperspectral data is large, it takes a lot of time to reconstruct the three-dimensional data by using a general reconstruction method.
In recent years, a deep learning method is widely applied to the field of image processing, and some students apply the idea of deep learning to data reconstruction of a compressed spectrum imaging system. The method optimizes a coding template by utilizing a forward propagation mechanism of the neural network, takes compressed observation data as input of a reconstruction network, respectively designs a spectral reconstruction network and a spatial reconstruction network, and finally obtains reconstructed spectral data. The method has the advantages that the coding optimization and the compression reconstruction can be carried out simultaneously, and the defect is that the whole network needs to be retrained when the coding template is changed, so that the network has no mobility to different coding templates and different compression imaging systems.
Disclosure of Invention
In view of this, the present invention provides a three-dimensional data reconstruction method for a liquid crystal hyperspectral computed imaging system. A convolutional neural network suitable for a liquid crystal hyperspectral calculation imaging system is built, a compressed observation result and system response obtained by the calculation imaging system are jointly used as system input, and reconstructed high-resolution three-dimensional data are finally output through a plurality of hidden layers. Wherein the system response comprises a spatial response and a spectral response of the system, respectively representing a spatial and spectral encoding effect of the system on the incident scene. The invention carries out the calculation and reconstruction of the compressed observation data under the framework of the convolutional neural network, simultaneously considers the compressed data and the system response, and under the condition of enough training data, the network can adapt to different coding templates and various types of calculation spectral imaging systems, thereby rapidly and accurately acquiring the reconstructed three-dimensional data.
In order to solve the technical problem, the invention provides a three-dimensional data reconstruction method of a liquid crystal hyperspectral computed imaging system, which comprises the following steps:
step S1: acquiring a compression observation result of the liquid crystal hyperspectral calculation imaging system;
step S2: acquiring the spatial response and the spectral response of the liquid crystal hyperspectral calculation imaging system;
and step S3: generating input layer data for a convolutional neural network based on the compressed observations, the spatial response, and the spectral response;
and step S4: inputting the data of the input layer into the convolutional neural network, outputting the data of the input layer into high-resolution three-dimensional data corresponding to the compressed observation result, and training the convolutional neural network until the preset precision is reached;
step S5: and according to the steps S1-S3, acquiring data of an input layer to be reconstructed, inputting the data into the trained convolutional neural network, and outputting to obtain reconstructed high-resolution three-dimensional data.
Optionally, the step S1 includes:
the liquid crystal hyperspectral computation imaging system is used for carrying out compression coding on the incident scene in space and spectrum dimensions, and the compression observation result of the system is obtained on the detector.
Optionally, the number of the coding templates of the liquid crystal hyperspectral computed imaging system is marked as K, the number of the wave bands of the liquid crystal tunable filter is marked as L, and the number of the pixels on the detector is marked as M x M; the compressed observation result obtaining mode comprises the following steps:
s11, loading a coding template, selecting a wave band of the liquid crystal tunable filter, and acquiring compressed data with dimension M × M on a detector;
s12, keeping the encoding template unchanged, adjusting the liquid crystal tunable filter to the next wave band, and acquiring compressed data on the detector;
s13, repeating the step S12 until the compressed data acquisition of the liquid crystal tunable filter under all L wave bands is completed;
s14, loading the next coding template, and repeating the steps S11-S13 until the compressed data acquisition under all the K coding templates is completed; the data dimension from which the compressed observations were obtained was K M L.
Optionally, the step S2 of obtaining the spatial response of the liquid crystal hyperspectral computed imaging system includes:
a detector with the resolution equivalent to the coding aperture in the spatial coding module is placed at the position of the detector, system incident light is set into white light which is uniformly distributed, a coding template is loaded, and the response of the detector is divided by the light intensity of the incident light, namely the system spatial response under the coding template; and switching the coding templates, and repeating the steps until the measurement of the spatial responses of all the K coding templates is completed.
Optionally, the step S2 of obtaining the spectral response of the liquid crystal hyperspectral computed imaging system includes:
obtaining a quantum response curve C (lambda) of the detector, measuring the spectrum curve S at the entrance pupil of the liquid crystal hyperspectral calculation imaging system by using the fiber spectrometer 0 (λ); selecting the ith (i is more than or equal to 1 and less than or equal to L) spectral band of the liquid crystal tunable filter, and measuring the spectral curve S filtered by all optical elements of the system on a detector plane i (λ), the spectral response of the system in the ith spectral band of the LC tunable filter is C (λ) × S i (λ)/S 0 (λ); and switching the liquid crystal tunable filter to the next wave band, and repeating the steps until the spectral responses of all the L wave bands are measured.
Optionally, the step S3 includes:
performing a reorganization of the compressed observations to generate three-dimensional data having a dimension of N x L (K x M is generally less than N x N, the dimension is reached by zero padding); performing recombination arrangement on the spatial responses to generate three-dimensional data with dimensions of N x K; stretching the dimensions of the spectral response to N L using principal component analysis; and connecting the processed compression observation result, the spatial response and the spectral response in series to be used as the data of the input layer of the convolutional neural network.
The invention has the beneficial effects that:
according to the three-dimensional data reconstruction method of the liquid crystal hyperspectral computed imaging system, a convolutional neural network applied to the liquid crystal hyperspectral computed imaging system is built, a compressed observation result and a system response obtained by the computed imaging system are jointly used as system input, and finally reconstructed high-resolution three-dimensional data are output through a plurality of hidden layers. Wherein the system response comprises a spatial response and a spectral response of the system, respectively representing the spatial and spectral encoding effect of the system on the incident scene. The invention carries out the calculation reconstruction of the compressed observation data under the framework of the convolutional neural network, and the convolutional neural network can adapt to various types of calculation spectral imaging systems under the condition of enough training data, thereby rapidly and accurately acquiring the reconstructed three-dimensional data.
Drawings
FIG. 1 is a network block diagram of three-dimensional data reconstruction of a liquid crystal hyperspectral computed imaging system according to the invention;
FIG. 2 is a schematic flow chart of a three-dimensional data reconstruction method of the liquid crystal hyperspectral computed imaging system according to the invention;
FIG. 3 is a diagram of the input layer data structure of the convolutional neural network of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail by the following embodiments with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
Referring to fig. 1, the invention constructs a convolutional neural network suitable for a liquid crystal hyperspectral computational imaging system, uses a compressed observation result and a system response obtained by the computational imaging system as system input together, and finally outputs reconstructed high-resolution three-dimensional data through a plurality of hidden layers. Wherein the system response comprises a spatial response and a spectral response of the system, respectively representing the spatial and spectral encoding effect of the system on the incident scene. The invention carries out the calculation reconstruction of the compressed observation data under the framework of the convolutional neural network, and under the condition of enough training data, the network can adapt to various types of calculation spectral imaging systems and quickly and accurately acquire the reconstructed three-dimensional data.
Referring to fig. 2, fig. 2 is a schematic flow chart of a three-dimensional data reconstruction method of a liquid crystal hyperspectral computed imaging system, which mainly includes the following steps:
step S1: and acquiring a compression observation result of the liquid crystal hyperspectral calculation imaging system.
The liquid crystal hyperspectral computation imaging system is used for carrying out compression coding on the incident scene in space and spectrum dimensions, and the compression observation result of the system is obtained on the detector. The number of the coding templates is recorded as K, the wave band number of the liquid crystal tunable filter is recorded as L, the pixel number on the detector is recorded as M, and the acquisition mode of the compressed observation result specifically comprises the following steps:
s11, loading a coding template, selecting a wave band of the liquid crystal tunable filter, and acquiring compressed data with dimension M × M on a detector;
s12, keeping the encoding template unchanged, adjusting the liquid crystal tunable filter to the next wave band, and acquiring compressed data on the detector;
s13, repeating the step S12 until the compressed data acquisition of the liquid crystal tunable filter under all L wave bands is completed;
s14, loading the next coding template, and repeating the steps S11-S13 until the compressed data acquisition under all the K coding templates is completed; the data dimension from which the compressed observations were obtained was K M L.
Step S2: and acquiring the spatial response and the spectral response of the liquid crystal hyperspectral computed imaging system.
The dimensions of the original spectral scene are denoted N x N λ. The spatial response of the system can be regarded as K-times mapping from N x N spatial positions of the incident scene to M x M detector pixels; the spectral response of the system can be regarded as that the liquid crystal tunable filter performs L-time spectral filtering on the N lambda dimensional spectral information of the incident scene. The spatial magnification of the system, R = N/M, R usually being an integer for ease of calculation.
The system space response measuring method comprises the following steps: a detector with the resolution equivalent to the coding aperture in the spatial coding module is placed at the position of the detector, system incident light is set into white light which is uniformly distributed, a coding template is loaded, and the response of the detector is divided by the light intensity of the incident light, namely the system spatial response under the coding template; and switching the coding templates, and repeating the steps until the measurement of the spatial responses of all the K coding templates is completed. The system space response matrix is formed by M dimensions, M dimensions and K dimensions R 2 Is used to form the matrix.
The system spectral response determination method comprises the following steps: obtaining a quantum response curve C (lambda) of the detector from a detector manufacturer, measuring the spectrum curve S at the entrance pupil of the liquid crystal hyperspectral calculation imaging system by using a fiber spectrometer 0 (λ); selecting the ith (i is more than or equal to 1 and less than or equal to L) spectral band of the liquid crystal tunable filter, and measuring the spectral curve S filtered by all optical elements of the system at the detector plane i (λ), the spectral response of the system in the ith spectral band of the LC tunable filter is C (λ) × S i (λ)/S 0 (λ); and switching the liquid crystal tunable filter to the next wave band, and repeating the steps until the spectral responses of all the L wave bands are measured.
And step S3: and generating input layer data of the convolutional neural network based on the compressed observation result, the spatial response and the spectral response.
Referring to fig. 3, fig. 3 is a diagram of a data structure of the input layer of the convolutional neural network of the present invention. The input layer data of the convolutional neural network consists of a compressed observation result of a system and a system response, wherein the system response comprises a system space response and a system spectrum response. Performing reorganization arrangement on the compressed observation results of the system to generate three-dimensional data with the dimension of N × L (K × M is usually smaller than N × N, and the dimension is achieved by zero filling); recombining and arranging the system space response matrix to generate a three-dimensional data cube with dimension N x K; stretching the dimensionality of the system spectral response matrix to N L by using a principal component analysis method; the compression observation result of the system and the system response are connected in series and can be used as the input layer data of the convolutional neural network.
And step S4: and inputting the data of the input layer into the convolutional neural network, outputting the data into high-resolution three-dimensional data corresponding to the compressed observation result, and training the convolutional neural network until the preset precision is reached.
The convolution layer, normalization layer and activation layer are used as basic units, and the structure is repeated for multiple times to form a convolution neural network. And (4) inputting the data of the input layer generated in the step (S3) into a designed convolutional neural network, outputting high-resolution three-dimensional data corresponding to the compressed observation result, and training the network until the preset precision is reached, wherein the data dimension is NxNxNlambda. N λ represents the spectral information dimension of the incident scene.
Step S5: and according to the steps S1-S3, acquiring data of an input layer to be reconstructed, inputting the data into the trained convolutional neural network, and outputting to obtain reconstructed high-resolution three-dimensional data.
In the input part of the convolutional neural network, compressed observation data and system response are considered at the same time, and compared with a compressed sensing reconstruction neural network only taking compressed observation as system input, the accuracy of the obtained reconstruction spectrum data is expected to be higher.
The network has strong robustness to the compression and reconstruction of different systems and different coding templates, and can respond to the trained system by using different systems, thereby being suitable for various compression spectral imaging systems. The same idea can be applied to other systems of the compressed sensing principle.
It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented in a general purpose computing device, they may be centralized in a single computing device or distributed across a network of multiple computing devices, and alternatively, they may be implemented in program code executable by a computing device, such that they may be stored on a computer storage medium (ROM/RAM, magnetic disk, optical disk) for execution by a computing device, and in some cases, the steps shown or described may be performed in an order different from that described herein, or they may be separately fabricated as individual integrated circuit modules, or multiple ones of them may be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.
The foregoing is a more detailed description of the present invention that is presented in conjunction with specific embodiments, and the practice of the invention is not to be considered limited to those descriptions. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (2)
1. A three-dimensional data reconstruction method of a liquid crystal hyperspectral computed imaging system is characterized by comprising the following steps:
step S1: acquiring a compression observation result of the liquid crystal hyperspectral calculation imaging system;
step S2: acquiring the spatial response and the spectral response of the liquid crystal hyperspectral calculation imaging system;
and step S3: generating input layer data for a convolutional neural network based on the compressed observations, the spatial response, and the spectral response;
and step S4: inputting the data of the input layer into the convolutional neural network, outputting the data of the input layer into high-resolution three-dimensional data corresponding to the compressed observation result, and training the convolutional neural network until the preset precision is reached;
step S5: according to the steps S1-S3, acquiring data of an input layer to be reconstructed, inputting the data into a trained convolutional neural network, and outputting to obtain reconstructed high-resolution three-dimensional data;
the method comprises the following steps of calculating the number of encoding templates of an imaging system by liquid crystal hyperspectral computation, wherein the number of encoding templates of the imaging system by liquid crystal hyperspectral computation is recorded as K, the number of wave bands of a liquid crystal tunable filter is recorded as L, the number of pixels on a detector is recorded as M, and the acquisition mode of a compressed observation result comprises the following steps:
s11, loading an encoding template, selecting a wave band of the liquid crystal tunable filter, and acquiring compressed data with the dimension of M x M on a detector;
s12, keeping the encoding template unchanged, adjusting the liquid crystal tunable filter to the next wave band, and acquiring compressed data on the detector;
s13, repeating the step S12 until the compressed data acquisition of the liquid crystal tunable filter under all L wave bands is completed;
s14, loading the next coding template, and repeating the steps S11-S13 until the compressed data acquisition under all the K coding templates is completed; obtaining a data dimension of the compressed observations as K x M x L;
the step S2 of obtaining the spatial response of the liquid crystal hyperspectral computed imaging system comprises the following steps:
a detector with the resolution equivalent to the coding aperture in the spatial coding module is placed at the position of the detector, system incident light is set into white light which is uniformly distributed, a coding template is loaded, and the response of the detector is divided by the light intensity of the incident light, namely the system spatial response under the coding template; switching the coding templates, and repeating the steps until the measurement of the spatial responses of all the K coding templates is completed;
the step S2 of obtaining the spectral response of the liquid crystal hyperspectral computed imaging system comprises the following steps:
obtaining a quantum response curve C (lambda) of the detector, and measuring the spectrum curve S0 (lambda) at the entrance pupil of the liquid crystal hyperspectral calculation imaging system by using the fiber spectrometer; selecting the ith (i is more than or equal to 1 and less than or equal to L) spectral band of the liquid crystal tunable filter, measuring a spectral curve Si (lambda) after filtering by all optical elements of the system on a detector plane, wherein the spectral response of the system at the ith spectral band of the liquid crystal tunable filter is C (lambda) × Si (lambda)/S0 (lambda); switching the liquid crystal tunable filter to the next wave band, and repeating the steps until the spectral responses of all the L wave bands are measured;
wherein the step S3 comprises:
performing recombination arrangement on the compressed observation results to generate three-dimensional data with dimensions of N x L; performing recombination arrangement on the spatial responses to generate three-dimensional data with dimensions of N x K; stretching the dimensions of the spectral response to NxL using a principal component analysis method; and connecting the processed compression observation result, the spatial response and the spectral response in series to be used as the input layer data of the convolutional neural network.
2. The three-dimensional data reconstruction method of the liquid crystal hyperspectral computed imaging system according to claim 1, wherein the step S1 comprises:
the liquid crystal hyperspectral computation imaging system is used for carrying out space and spectrum dimensional compression coding on an incident scene, and a compression observation result of the system is obtained on a detector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011054740.5A CN112229514B (en) | 2020-09-27 | 2020-09-27 | Three-dimensional data reconstruction method of liquid crystal hyperspectral calculation imaging system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011054740.5A CN112229514B (en) | 2020-09-27 | 2020-09-27 | Three-dimensional data reconstruction method of liquid crystal hyperspectral calculation imaging system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112229514A CN112229514A (en) | 2021-01-15 |
| CN112229514B true CN112229514B (en) | 2023-04-18 |
Family
ID=74119666
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011054740.5A Active CN112229514B (en) | 2020-09-27 | 2020-09-27 | Three-dimensional data reconstruction method of liquid crystal hyperspectral calculation imaging system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112229514B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102661794A (en) * | 2012-03-20 | 2012-09-12 | 清华大学 | Multispectral calculation reconstruction method and system |
| CN108955882A (en) * | 2018-07-10 | 2018-12-07 | 北京理工大学 | The three-dimensional data reconstructing method of imaging system is calculated based on liquid crystal EO-1 hyperion |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200094058A (en) * | 2019-01-29 | 2020-08-06 | 한국과학기술원 | Lensless Hyperspectral Imaging Method and Apparatus Therefore |
-
2020
- 2020-09-27 CN CN202011054740.5A patent/CN112229514B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102661794A (en) * | 2012-03-20 | 2012-09-12 | 清华大学 | Multispectral calculation reconstruction method and system |
| CN108955882A (en) * | 2018-07-10 | 2018-12-07 | 北京理工大学 | The three-dimensional data reconstructing method of imaging system is calculated based on liquid crystal EO-1 hyperion |
Non-Patent Citations (1)
| Title |
|---|
| 《Compressive spectral imaging system for soil classification with three-dimensional convolutional neural network》;Yue Yu等;《Optics EXPRESS》;第27卷(第16期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112229514A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110175971B (en) | Deep learning image reconstruction method for multispectral single-pixel imaging | |
| CN105492878B (en) | Apparatus and method for snapshot light spectrum image-forming | |
| Arce et al. | Compressive coded aperture spectral imaging: An introduction | |
| CN111047681B (en) | Single-pixel three-dimensional end-to-end reconstruction method and device based on deep learning | |
| CN108955882B (en) | Three-dimensional data reconstruction method based on liquid crystal hyperspectral calculation imaging system | |
| CN110650340B (en) | Space-time multiplexing compressed video imaging method | |
| CN109741407A (en) | A kind of high quality reconstructing method of the spectrum imaging system based on convolutional neural networks | |
| CN109087262B (en) | Multi-view spectral image reconstruction method and storage medium | |
| CN111652059A (en) | Target identification model construction and identification method and device based on computational ghost imaging | |
| CN110378975A (en) | A kind of compressed encoding aperture imaging method and system based on deep neural network | |
| CN115700727A (en) | Spectral super-resolution reconstruction method and system based on self-attention mechanism | |
| CN114419392A (en) | Hyperspectral snapshot image recovery method, device, equipment and medium | |
| Jacquemont et al. | Indexed operations for non-rectangular lattices applied to convolutional neural networks | |
| CN111598962A (en) | Single-pixel imaging method and device based on matrix sketch analysis | |
| Degraux et al. | Multispectral compressive imaging strategies using Fabry–Pérot filtered sensors | |
| He et al. | Single-shot hyperspectral imaging based on dual attention neural network with multi-modal learning | |
| CN112229514B (en) | Three-dimensional data reconstruction method of liquid crystal hyperspectral calculation imaging system | |
| Chen et al. | Prior image guided snapshot compressive spectral imaging | |
| CN113256733A (en) | Camera spectral sensitivity reconstruction method based on confidence voting convolutional neural network | |
| CN116579959B (en) | Fusion imaging method and device for hyperspectral image | |
| CN108801457B (en) | Three-dimensional map acquisition and reconstruction method based on coding sampling plate design and secondary energy constraint correction | |
| CN107895063A (en) | One kind compression EO-1 hyperion mask optimization method | |
| CN115342916A (en) | Spectral imaging method based on high-order mathematical modeling and fitting calibration | |
| Jiao et al. | Image recognition based on compressive imaging and optimal feature selection | |
| Kim et al. | Computational hyperspectral imaging with diffractive optics and deep residual network |
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 |