CN116242484B - Spectrum imaging chip and spectrum reconstruction algorithm collaborative design method - Google Patents
Spectrum imaging chip and spectrum reconstruction algorithm collaborative design method Download PDFInfo
- Publication number
- CN116242484B CN116242484B CN202310517537.4A CN202310517537A CN116242484B CN 116242484 B CN116242484 B CN 116242484B CN 202310517537 A CN202310517537 A CN 202310517537A CN 116242484 B CN116242484 B CN 116242484B
- Authority
- CN
- China
- Prior art keywords
- spectrum
- spectral
- imaging chip
- reconstruction
- random
- 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
- 238000001228 spectrum Methods 0.000 title claims abstract description 163
- 238000003384 imaging method Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000013461 design Methods 0.000 title claims abstract description 31
- 230000003595 spectral effect Effects 0.000 claims abstract description 47
- 230000003287 optical effect Effects 0.000 claims abstract description 44
- 238000002834 transmittance Methods 0.000 claims abstract description 22
- 238000005457 optimization Methods 0.000 claims abstract description 20
- 230000004044 response Effects 0.000 claims abstract description 8
- 238000004806 packaging method and process Methods 0.000 claims abstract description 6
- 238000000701 chemical imaging Methods 0.000 claims description 30
- 238000004364 calculation method Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 238000001259 photo etching Methods 0.000 claims description 3
- 238000002601 radiography Methods 0.000 claims 2
- 238000013135 deep learning Methods 0.000 abstract description 9
- 239000000758 substrate Substances 0.000 abstract description 3
- 238000005530 etching Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 3
- 230000000295 complement effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
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
- G01J3/28—Investigating the spectrum
- G01J3/2823—Imaging spectrometer
-
- 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
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G01J2003/283—Investigating the spectrum computer-interfaced
- G01J2003/284—Spectral construction
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Spectrometry And Color Measurement (AREA)
Abstract
A collaborative design method for a spectrum imaging chip and a spectrum reconstruction algorithm relates to the field of spectrum imaging, and comprises the following steps: calculating the transmittance of the random filter corresponding to the structural parameters of the random filter; the spectral response light intensity of the original spectral vector to the transmittance of the random filter is obtained through dot product operation, and the spectral response light intensity is input into a spectral reconstruction network to calculate a reconstruction spectral vector; optimizing and adjusting to enable the reconstructed spectrum vector to be continuously approximate to the original spectrum vector to realize optimization; the spectrum imaging chip basic unit comprises a plurality of narrow-band optical filters and random optical filters; and (3) after the optical filters are combined and arranged, etching the optical filters on a spectrum imaging chip substrate, and assembling the optical filters with a packaging shell to obtain the spectrum imaging chip. The invention improves the light energy utilization rate, the signal-to-noise ratio and the coverage spectrum range of the spectrum imaging chip, and solves the optimization problem caused by independent design of a physical structure and an algorithm based on compressed sensing and deep learning spectrum reconstruction algorithm.
Description
Technical Field
The invention relates to the technical field of spectrum imaging, in particular to a method for collaborative design of a spectrum imaging chip and a spectrum reconstruction algorithm.
Background
Compared with the traditional spectrum imaging system in a mode that a light splitting structure is separated from a detector, the spectrum imaging chip based on the pixel narrow-band filter and the spectrum imaging chip based on the pixel random filter have the advantages of simple structure, small volume, small spectrum image distortion and the like, so that the two spectrum imaging chips have wide application prospects. However, the spectrum imaging chip based on the pixel narrowband filter realizes multispectral spectrum splitting by using the narrowband filter, the technology can obtain spectrum image information through simple spectrum restoration, and meanwhile, the technology has the accuracy and precision of scientific metering level, but the chip has low signal-to-noise ratio and narrow coverage spectrum range due to low light energy utilization rate of the narrowband filter. The spectrum imaging chip based on the pixel random filter encodes an incident spectrum by using the broadband random filter, and the back-end spectrum reconstruction algorithm is used for realizing the multi-spectrum reconstruction. The two types of chips have obvious complementary advantages because of different light splitting modes.
The random optical filter spectrum encoding and decoding utilizes the sparsity of the spectrum, encodes the incident spectrum by using an optical filter with random transmittance distribution, compresses and represents the continuous spectrum into a low-dimensional light intensity vector, and then restores the low-dimensional light intensity vector into a spectrum vector which can be intuitively understood by human beings through a decoding algorithm. The smallest dimension in which a spectrum can be compressed in theory depends on the sparsity of the spectrum. According to different optimization ideas, two spectrum reconstruction algorithms of compressed sensing and deep learning exist at present, wherein the compressed sensing spectrum reconstruction algorithm realizes the solution of an underdetermined equation by means of sparse dimension reduction, and a spectrum vector is restored by using sparse solution; the deep learning spectrum reconstruction algorithm realizes spectrum reconstruction by means of automatic feature extraction and data generation capability of the incomplete self-encoder. Both algorithms have optimization problems due to the independent design of the physical structure and algorithm.
Disclosure of Invention
The invention aims to provide a collaborative design method for a spectrum imaging chip and a spectrum reconstruction algorithm, which aims to solve the problems of low light energy utilization rate, low signal to noise ratio, narrow coverage spectrum range and optimization caused by independent design of a physical structure and an algorithm in the existing spectrum imaging chip based on compressed sensing and deep learning spectrum reconstruction algorithm.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the invention discloses a collaborative design method of a spectrum imaging chip and a spectrum reconstruction algorithm, which comprises the following steps:
step one, calculating the transmittance of a random optical filter, optimizing and training parameters of a spectrum reconstruction network;
taking the structural parameters of the random optical filter as optimization parameters of a spectrum reconstruction network, and calculating the transmittance of the random optical filter corresponding to the structural parameters of the random optical filter; the spectral response light intensity of the original spectral vector to the transmittance of the random filter is obtained through dot product operation, the spectral response light intensity is input into a spectral reconstruction network, and the reconstructed spectral vector is calculated to finish forward propagation; the spectrum reconstruction network parameters are optimized and adjusted to enable the reconstruction spectrum vector to be continuously approximate to the original spectrum vector, so that optimization of the spectrum reconstruction network parameters is realized;
step two, designing and integrating a spectrum imaging chip;
the spectrum imaging chip basic unit comprises a plurality of narrow-band optical filters and a plurality of random optical filters; and the plurality of narrow-band optical filters and the plurality of random optical filters are combined and arranged, then etched on a spectrum imaging chip matrix and assembled with the packaging shell to obtain the spectrum imaging chip.
Further, in the first step, the following spectrum reconstruction network parameter optimization expression is adopted to perform spectrum reconstruction network parameter optimization;
In the method, in the process of the invention,reconstruction error term, representing reconstructed spectral vectorFitting degree with the original spectrum vector S; r (P) is a neural network parameter constraint term used for constraining the structural parameter range of the random optical filter and preventing the spectrum reconstruction network from being over fitted;respectively representing the sparsely represented signals and reconstructed network parameters; p is the random filter structure parameter vector,for spectral reconstruction network parameters, TF (P) is the transmittance of the random filter with structural parameters of P,REC is the light intensity of the incident light after being encoded by the random filter ω The network is reconstructed for the spectrum.
Further, in the first step, the spectrum reconstruction network parameters are optimized and adjusted to enable the reconstructed spectrum vector to continuously approach the original spectrum vector, so that the optimization of the spectrum reconstruction network parameters is realized, specifically: and calculating a loss value according to the loss function, updating the spectrum reconstruction network parameters through the gradient back propagation of the loss value, and repeating the process until the loss value converges to be close to zero, thereby completing the optimization of the spectrum reconstruction network parameters.
In the second step, the random filter and the narrow-band filter are designed into pixel sizes, and the narrow-band filter and the random filter with the pixel sizes are etched on the surface of the spectrum imaging chip substrate through a photoetching process according to the designed combined arrangement mode, and are assembled with the packaging shell to obtain the spectrum imaging chip.
In the second step, the spectral imaging chip has two reconstruction modes of calculation enhancement and calculation direct connection.
Furthermore, in the second step, the spectrum imaging chip can collect information of the high-precision narrow-band spectrum and information of the high-signal-to-noise-ratio broadband spectrum at the same time.
In the second step, the spectrum imaging chip basic unit is composed of 12 narrowband optical filters and 4 random optical filters, wherein the 12 narrowband optical filters are arranged on the periphery, and the 4 random optical filters are arranged in the middle.
The invention relates to a calculation enhanced pixel spectroscopic imaging chip, which is obtained by adopting the collaborative design method of the spectroscopic imaging chip and a spectrum reconstruction algorithm.
Furthermore, the calculation enhancement type pixel spectroscopic imaging chip has two reconstruction modes of calculation enhancement and calculation direct connection.
Furthermore, the calculation enhanced pixel spectroscopic imaging chip can collect information of a high-precision narrow-band spectrum and information of a high signal-to-noise ratio broadband spectrum at the same time.
The beneficial effects of the invention are as follows:
according to the spectrum imaging chip and spectrum reconstruction algorithm collaborative design method, the optimal matching of the random filter and the spectrum reconstruction network is realized by combining the calculation of the transmittance of the random filter and the training of the spectrum reconstruction network. And the fusion calculation of the deep neural network is utilized, so that the advantages of the random filter and the narrow-band filter on a single chip are complemented, and on-chip spectrum imaging with high signal-to-noise ratio, wide spectrum range and high spectrum precision is realized. In addition, the broadband transmittance of the random optical filter is restrained, so that the spectral energy utilization rate of the system is improved, the receiving efficiency of pixel light energy is balanced, the signal-to-noise ratio of spectral imaging is improved, and the dynamic range is enlarged.
Drawings
Fig. 1 is a schematic diagram of a method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to the present invention.
Fig. 2 is a schematic diagram of the structural composition of the basic unit of the spectral imaging chip.
Fig. 3 is a schematic diagram of spectral imaging chip assembly.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
In the existing random filter spectrum encoding and decoding framework based on compressed sensing and deep learning, the design of a random filter and a spectrum reconstruction algorithm is independently carried out. In order to reconstruct the original spectrum with high precision, the independent design mode often needs complex random transmittance distribution and more random filter types, so that the random filter design difficulty is high, the spectrum sampling efficiency is low, and the potential of a random filter spectrum coding and decoding architecture cannot be fully exerted. In the existing compressed sensing spectrum reconstruction algorithm, the original spectrum is compressed and encoded by the random optical filter set, the light intensity after being sensed and encoded by the light intensity detector, then the photocurrent of the light intensity detector is subjected to analog-to-digital conversion (ADC) to obtain digital quantity, the digital quantity is input into a Digital Signal Processor (DSP), and the compressed sensing spectrum reconstruction algorithm is executed to complete spectrum reconstruction. The compressed perceptual spectral reconstruction algorithm requires that the observation matrix is uncorrelated with the K-sparse complete orthonormal of the original spectrum in order to meet the finite equidistant property (Restricted Isometry Propery, RIP) condition. The observation matrix is usually a universal independent Gaussian random matrix with the same distribution, so that the design and processing difficulty of the random filter is increased. In order to reduce the difficulty of film processing, another method is to design a random filter set by randomly generating the film thickness and then design a spectrum reconstruction algorithm according to the random filter set, but the random characteristics of the filter obtained by the method are difficult to control, and the difficulty of designing a compressed sensing spectrum reconstruction algorithm is high.
Deep learning spectral coding is essentially an under-complete self-coding framework. The physical coding of the spectrum is realized by utilizing the random filter, the digital decoding of the spectrum is realized by the spectrum reconstruction network, the most obvious characteristic in the network learning spectrum data is reconstructed by forcedly setting the dimension of the coding output (namely the type of the random filter), so that the optimal representation of the spectrum data under the coding dimension is obtained, and meanwhile, the spectrum reconstruction network has certain denoising capability. But the spectral reconstruction accuracy is severely dependent on the random filter transmittance distribution (determined by the film material and thickness), resulting in difficulty in achieving optimal spectral codec performance at certain encoding dimensions.
Therefore, the invention provides a spectrum imaging chip and spectrum reconstruction algorithm collaborative design method based on a deep learning spectrum coding and decoding incomplete self-coding frame, which mainly comprises the steps of etching a pixel narrow-band filter and a pixel random filter on the surface of a detector, constructing a calculation enhanced pixel spectroscopic imaging chip, and utilizing fusion calculation of a depth neural network to enable the advantages of the two filters to be complementary on a single chip, so as to realize on-chip spectrum imaging with high signal-to-noise ratio, wide spectrum range and high spectrum precision.
The invention discloses a collaborative design method of a spectrum imaging chip and a spectrum reconstruction algorithm, which specifically comprises the following steps:
step one, calculating the transmittance of a random optical filter, optimizing and training parameters of a spectrum reconstruction network;
in the deep learning spectrum reconstruction algorithm, the main objective of the deep learning spectrum reconstruction algorithm is to train to obtain an optimal spectrum reconstruction networkAnd then outputting a reconstructed spectrum by using the optimal spectrum reconstruction network. As shown in fig. 1, the structural parameters P of the random filter [n] =[p n1 ,p n2 ……p ns ]N=1, 2 … … N; structural parameter P of random optical filter [n] As an optimization parameter of the spectrum reconstruction network, calculating a structural parameter P of the random filter [n] Corresponding random filter transmittance T [n] The method comprises the steps of carrying out a first treatment on the surface of the Then the transmittance T of the original spectrum vector S to the random filter is obtained through dot product operation [n] Spectral response intensity I of (2) n Discretizing the low-dimensional spectral response light intensity I n (I 1 、I 2 ……I N ) Input spectrum reconstruction network REC ω Calculating a high-dimensional reconstructed spectral vectorCompleting forward propagation; spectral reconstruction network REC by optimization ω Is such that the spectral vector is reconstructedThe original spectrum vector S is continuously approximated, namely, a loss value is calculated according to a loss function (mean square error), the spectrum reconstruction network parameters are updated through the back propagation of the loss value gradient, and the process is repeated until the loss value converges to be close to zero, so that the optimization of the spectrum reconstruction network parameters is realized.
The optimized expression of the adopted spectrum reconstruction network parameters is as follows:;
in the method, in the process of the invention,reconstruction error term, representing reconstructed spectral vectorFitting degree with the original spectrum vector S; r (P) is a neural network parameter constraint term used for constraining the structural parameter range of the random optical filter and preventing the spectrum reconstruction network from being over fitted, and generally adopts l 2 Regularizing constraints to causeGet II omega II 2 ≤C(0≤C<∞); Respectively representing the sparsely represented signals and reconstructed network parameters; p is the random filter structure parameter vector,for spectral reconstruction network parameters, TF (P) is the transmittance of the random filter with structural parameters of P,REC is the light intensity of the incident light after being encoded by the random filter ω The network is reconstructed for the spectrum.
Step two, designing and integrating a spectrum imaging chip;
the design of the basic unit of the spectrum imaging chip for imaging and spectrum acquisition can be realized through the first step, specifically, the design of the structural parameters and the transmittance of the random optical filter is finished, the basic unit of the spectrum imaging chip is equivalent to one pixel of an imaging spectrometer, and the basic units are required to be scanned or repeatedly arranged to traverse the whole field of view in order to realize the spectrum image acquisition of the whole field of view. The method comprises the following steps:
the spectral imaging chip basic unit is mainly composed of two light splitting structures of a narrow-band filter (FPx, x=1, 2 … … 10) and a random filter (RFx, x=1, 2, 3, 4) as shown in fig. 2, and the combined arrangement form of the two light splitting structures is not limited to the mode in fig. 2.
By utilizing a photoetching process, the random optical filter and the narrow-band optical filter are designed into pixel sizes, and the narrow-band optical filter and the random optical filter with the pixel sizes are etched on the surface of a substrate according to a designed combined arrangement mode and assembled with a packaging shell to obtain the spectrum imaging chip, as shown in figure 3. The obtained spectrum imaging chip has two reconstruction modes of calculation enhancement and calculation direct connection, and can collect information of a high-precision narrow-band spectrum and information of a high signal-to-noise ratio broadband spectrum at the same time.
In this embodiment, as a preferred combination arrangement mode, the combination arrangement mode of the narrowband filter and the random filter is as shown in fig. 2, a circle of narrowband filter is arranged at the periphery, and from the upper left corner, the narrowband filters FP 1-FP 12 are sequentially arranged right, downward, leftward and upward; four random filters are distributed in the middle, namely a random filter RF1 and a random filter RF4.
According to the spectrum imaging chip and spectrum reconstruction algorithm collaborative design method, the optimal matching of the random filter and the spectrum reconstruction network is realized by combining the calculation of the transmittance of the random filter and the training of the spectrum reconstruction network. In addition, the broadband transmittance of the random optical filter is restrained, so that the spectral energy utilization rate of the system is improved, the receiving efficiency of pixel light energy is balanced, the signal-to-noise ratio of spectral imaging is improved, and the dynamic range is enlarged.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (10)
1. The collaborative design method of the spectrum imaging chip and the spectrum reconstruction algorithm is characterized by comprising the following steps:
step one, calculating the transmittance of a random optical filter, optimizing and training parameters of a spectrum reconstruction network;
taking the structural parameters of the random optical filter as optimization parameters of a spectrum reconstruction network, and calculating the transmittance of the random optical filter corresponding to the structural parameters of the random optical filter; the spectral response light intensity of the original spectral vector to the transmittance of the random filter is obtained through dot product operation, the spectral response light intensity is input into a spectral reconstruction network, and the reconstructed spectral vector is calculated to finish forward propagation; the spectrum reconstruction network parameters are optimized and adjusted to enable the reconstruction spectrum vector to be continuously approximate to the original spectrum vector, so that optimization of the spectrum reconstruction network parameters is realized;
step two, designing and integrating a spectrum imaging chip;
the spectrum imaging chip basic unit comprises a plurality of narrow-band optical filters and a plurality of random optical filters; and the plurality of narrow-band optical filters and the plurality of random optical filters are combined and arranged, then etched on a spectrum imaging chip matrix and assembled with the packaging shell to obtain the spectrum imaging chip.
2. The method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to claim 1, wherein in the first step, the following spectral reconstruction network parameter optimization expression is adopted for optimization of the spectral reconstruction network parameter:
the method comprises the steps of carrying out a first treatment on the surface of the In the method, in the process of the invention,reconstruction error term, representing reconstructed spectral vector +.>Fitting degree with the original spectrum vector S; r (P) is a neural network parameter constraint term used for constraining the structural parameter range of the random optical filter and preventing the spectrum reconstruction network from being over fitted; />Respectively representing the sparsely represented signals and reconstructed network parameters; p is the random filter structure parameter vector,for spectral reconstruction network parameters, TF (P) is the random filter transmittance with structural parameters P, +.>REC is the light intensity of the incident light after being encoded by the random filter ω The network is reconstructed for the spectrum.
3. The method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to claim 1, wherein in the first step, the spectral reconstruction network parameters are optimized and adjusted to make the reconstructed spectral vector continuously approximate to the original spectral vector, so as to realize optimization of the spectral reconstruction network parameters, specifically: and calculating a loss value according to the loss function, updating the spectrum reconstruction network parameters through the gradient back propagation of the loss value, and repeating the process until the loss value converges to be close to zero, thereby completing the optimization of the spectrum reconstruction network parameters.
4. The method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to claim 1, wherein in the second step, a random filter and a narrow-band filter are designed into pixel sizes, and the narrow-band filter and the random filter with the pixel sizes are etched on the surface of a matrix of the spectral imaging chip according to a designed combined arrangement form through a photoetching process and assembled with a packaging shell to obtain the spectral imaging chip.
5. The method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to claim 1, wherein in the second step, the spectral imaging chip has two reconstruction modes of calculation enhancement and calculation pass-through.
6. The method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to claim 1, wherein in the second step, the spectral imaging chip can collect information of a high-precision narrow-band spectrum and information of a high-signal-to-noise-ratio broadband spectrum at the same time.
7. The method for collaborative design of a spectral imaging chip and a spectral reconstruction algorithm according to claim 1, wherein in the second step, the spectral imaging chip basic unit is composed of 12 narrowband optical filters and 4 random optical filters, the 12 narrowband optical filters are arranged at the periphery, and the 4 random optical filters are arranged in the middle.
8. A computational enhancement type pixel spectroscopic imaging chip, which is characterized in that the computational enhancement type pixel spectroscopic imaging chip is obtained by adopting the collaborative design method of the spectroscopic imaging chip and a spectrum reconstruction algorithm according to any one of claims 1 to 7.
9. The computed radiography spectral imaging chip of claim 8 wherein the computed radiography spectral imaging chip has two reconstruction modes, namely, compute enhancement and compute pass-through.
10. The computational enhancement pixel spectral imaging chip of claim 8, wherein the computational enhancement pixel spectral imaging chip is capable of simultaneously acquiring high-precision narrowband spectral information and high signal-to-noise ratio broadband spectral information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310517537.4A CN116242484B (en) | 2023-05-10 | 2023-05-10 | Spectrum imaging chip and spectrum reconstruction algorithm collaborative design method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310517537.4A CN116242484B (en) | 2023-05-10 | 2023-05-10 | Spectrum imaging chip and spectrum reconstruction algorithm collaborative design method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116242484A CN116242484A (en) | 2023-06-09 |
CN116242484B true CN116242484B (en) | 2023-08-01 |
Family
ID=86631688
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310517537.4A Active CN116242484B (en) | 2023-05-10 | 2023-05-10 | Spectrum imaging chip and spectrum reconstruction algorithm collaborative design method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116242484B (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115731191A (en) * | 2022-11-22 | 2023-03-03 | 国科大杭州高等研究院 | Narrow-band spectral imaging method based on neural network |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9300865B2 (en) * | 2014-01-24 | 2016-03-29 | Goodrich Corporation | Random imaging |
JP7089296B2 (en) * | 2016-09-06 | 2022-06-22 | ビー.ジー.ネゲブ テクノロジーズ アンド アプリケーションズ リミテッド, アット ベン‐グリオン ユニバーシティー | Recovery of hyperspectral data from images |
KR101986998B1 (en) * | 2017-06-19 | 2019-06-10 | 광주과학기술원 | Hyperspectral Imaging Device |
CN114912598A (en) * | 2021-02-08 | 2022-08-16 | 清华大学 | Optical artificial neural network intelligent chip and preparation method thereof |
CN114739511A (en) * | 2022-03-23 | 2022-07-12 | 中北大学 | Snapshot type hyperspectral imaging chip structure based on spectrum modulation array |
CN115855250A (en) * | 2022-12-21 | 2023-03-28 | 中国科学院长春光学精密机械与物理研究所 | Spectrum imaging system based on deep learning spectrum coding and decoding |
-
2023
- 2023-05-10 CN CN202310517537.4A patent/CN116242484B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115731191A (en) * | 2022-11-22 | 2023-03-03 | 国科大杭州高等研究院 | Narrow-band spectral imaging method based on neural network |
Also Published As
Publication number | Publication date |
---|---|
CN116242484A (en) | 2023-06-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yang et al. | Compressive hyperspectral imaging via sparse tensor and nonlinear compressed sensing | |
CN111968044A (en) | Low-illumination image enhancement method based on Retinex and deep learning | |
Ratnasingam | Deep camera: A fully convolutional neural network for image signal processing | |
CN107817641A (en) | Single pixel camera system based on multimode fibre | |
CN114119444A (en) | Multi-source remote sensing image fusion method based on deep neural network | |
WO2021110262A1 (en) | Noise reconstruction for image denoising | |
CN116228912B (en) | Image compressed sensing reconstruction method based on U-Net multi-scale neural network | |
CN111157114A (en) | Long-wave infrared multispectral imaging method and device based on wavelength conversion | |
EP3700197A1 (en) | Imaging device and method, and image processing device and method | |
CN115307733A (en) | Space dimension coding spectrum polarization integrated imaging system and design method thereof | |
Ma et al. | Extensions of compressed imaging: flying sensor, coded mask, and fast decoding | |
CN117571128B (en) | High-resolution polarized spectrum image imaging method and system | |
CN115855250A (en) | Spectrum imaging system based on deep learning spectrum coding and decoding | |
Su et al. | Fast hyperspectral image denoising and destriping method based on graph Laplacian regularization | |
CN115931123A (en) | Spectrum reconstruction calculation method and spectrum system based on self-adaptive optimization sparse dictionary | |
CN116242484B (en) | Spectrum imaging chip and spectrum reconstruction algorithm collaborative design method | |
CN112950505A (en) | Image processing method, system and medium based on generation countermeasure network | |
Dubrovkin | Data Compression in Spectroscopy | |
CN115855839B (en) | Improved spatial spectrum fusion hyperspectral calculation reconstruction method based on ADMM framework | |
CN101576412B (en) | Near-lossless real-time compression method for satellite-borne hyper-spectral interference fringe image | |
Ma et al. | Leca: In-sensor learned compressive acquisition for efficient machine vision on the edge | |
Tsagkatakis et al. | Non-negative matrix completion for the enhancement of snapshot mosaic multispectral imagery | |
CN114429424B (en) | Remote sensing image super-resolution reconstruction method suitable for uncertain degradation modes | |
CN116164841B (en) | Spectrum reconstruction method based on calculation enhanced pixel spectroscopic imaging chip | |
CN114359103B (en) | Hyperspectral image defogging method, hyperspectral image defogging device, computer product and storage medium |
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 |