Spectrum chip structure design method and device for specific application spectrum library

Abstract

The application provides a spectral chip structure design method and device for a specific application spectral library, and the method comprises the following steps: acquiring a specific application spectral library and determining spectral principal components based on the specific application spectral library; performing non-negative processing on the spectral principal components to obtain non-negative spectral principal components; and determining a transmission spectrum of a spectral chip structure based on the non-negative spectral principal components. The spectral chip structure design method and device for the specific application spectral library provided in the embodiment of the application summarize the spectral characteristics of the specific application spectral library through the spectral principal components, realize quantitative evaluation of the spectral library, design the spectral chip structure on this basis, realize targeted quantitative design for the specific application spectral library, and thus improve the spectral reconstruction accuracy of the spectral chip.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and apparatus for designing spectral chip structures for a specific application spectral library. Background Technology

[0002] The spectral chip of a miniature spectrometer uses a micro / nano structure array to modulate the incident light, and then uses algorithms to recover the spectral information of the incident spectrum from the response of the spectral chip. However, in the current structural design process, the spectral measurement recovery performance of a set of structures can only be roughly estimated by the complexity of the transmission spectrum of each micro / nano structure, and quantitative indicators cannot be given for evaluation. Consequently, the micro / nano structure array of the spectral chip is designed, resulting in low reconstruction accuracy of the designed spectral chip. Summary of the Invention

[0003] This invention provides a method for designing a spectral chip structure for a specific application spectral library, in order to solve the problem of low reconstruction accuracy of spectral chips designed in the prior art and improve the spectral reconstruction accuracy of spectral chips.

[0004] In a first aspect, the present invention provides a method for designing a spectral chip structure for a specific application spectral library, comprising:

[0005] Acquire a specific application spectral library and determine the principal spectral components based on the specific application spectral library;

[0006] The principal components of the spectrum are nonnegated to obtain nonnegative principal components.

[0007] The transmission spectrum of the spectral chip structure is determined based on the non-negative spectral principal components.

[0008] Optionally, determining the principal spectral components based on the specific application spectral library includes:

[0009] The specific application spectral library is subjected to zero-mean processing to obtain a preprocessed spectral library;

[0010] Principal component analysis is performed on the preprocessed spectral library to obtain at least one initial eigenvector and eigenvalues ​​corresponding to each initial eigenvector;

[0011] Based on the eigenvalues, principal component eigenvectors are determined, which are used to represent the spectral principal components.

[0012] Optionally, performing principal component analysis on the preprocessed spectral library to obtain at least one initial eigenvector includes:

[0013] Principal component analysis is performed on the preprocessed spectral library based on preset principal component analysis conditions to obtain at least one initial feature vector.

[0014] The preset principal component analysis conditions are as follows:

[0015]

[0016] A T A = I;

[0017] Obtaining at least one initial feature vector includes:

[0018] Obtain A = (a1, a2, ..., a n );

[0019] Where tr represents the sum of all elements on the main diagonal. This refers to the preprocessed spectral library. express The transpose of , where A represents the characteristic matrix, A T Let I denote the transpose of the characteristic matrix, and let a denote the identity matrix. n This represents the initial feature vector.

[0020] Optionally, determining the principal component feature vector based on the feature values ​​includes:

[0021] Obtain the contribution rate corresponding to each of the initial feature vectors, wherein the contribution rate is the ratio of the feature value corresponding to each initial feature vector to the sum of all the feature values;

[0022] The initial feature vectors that meet the preset contribution rate threshold are used as the principal component feature vectors.

[0023] Optionally, the method further includes:

[0024] The contribution rate threshold is optimized based on the preset structural measurement and reconstruction accuracy requirements.

[0025] Optionally, determining the principal component feature vector based on the feature values ​​includes:

[0026] The eigenvalues ​​are arranged from largest to smallest, and the initial eigenvectors corresponding to the first K eigenvalues ​​are determined as principal component eigenvectors, where K is a preset positive integer.

[0027] Optionally, the method further includes:

[0028] Based on the preset structural measurement and reconstruction accuracy requirements, the preset K value is optimized.

[0029] Optionally, the spectral chip includes at least one structural unit, and the method further includes:

[0030] Based on the eigenvalues, a quantitative weighting is performed to determine the number of structural units corresponding to each principal component eigenvector.

[0031] Optionally, the nonnegation processing of the spectral principal components to obtain nonnegative spectral principal components includes:

[0032] The non-negative spectral principal components are obtained by taking the absolute value of the principal components of the spectrum; or

[0033] The preset supplementary value is added to the spectral principal component to obtain the non-negative spectral principal component.

[0034] Optionally, the transmission spectrum for determining the spectral chip structure based on the non-negative spectral principal components includes:

[0035] Based on the non-negative spectral principal components, design the transmission spectrum of the spectral chip structure; or

[0036] Based on the non-negative spectral principal components, the transmission spectrum of the spectral chip structure is selected from the pre-acquired candidate transmission spectra.

[0037] Optionally, the method further includes:

[0038] The similarity between the transmission spectra of the non-negative spectral principal components and the spectral chip structure is evaluated, and the transmission spectra of the spectral chip structure are optimized based on the similarity evaluation results.

[0039] Optionally, the similarity assessment method includes any of the following:

[0040] Likelihood calculation;

[0041] Correlation calculation;

[0042] Find the difference point by point;

[0043] Calculation of root mean square difference.

[0044] Optionally, the method further includes:

[0045] Update the application-specific spectral library.

[0046] Secondly, the present invention also provides a spectral chip structure design apparatus for a specific application spectral library, comprising:

[0047] An acquisition unit is used to acquire a specific application spectral library and determine the spectral principal components based on the specific application spectral library;

[0048] The processing unit is used to perform nonnegation processing on the spectral principal components to obtain nonnegative spectral principal components;

[0049] The determining unit is used to determine the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components.

[0050] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the spectral chip structure design method for a specific application spectral library as described in the first aspect.

[0051] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the spectral chip structure design method for a specific application spectral library as described in the first aspect.

[0052] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the spectral chip structure design method for a specific application spectral library as described in the first aspect.

[0053] The spectral chip structure design method and apparatus for specific application spectral libraries provided in this invention summarize the spectral features of the specific application spectral library through principal components, reducing the dimensionality of the data space of the studied spectral library, facilitating the analysis of spectral library features, and ensuring that the principal components retain the features that contribute the most to the spectral library, thereby achieving an effective representation of the spectral library; therefore, this invention achieves quantitative evaluation of the spectral library by analyzing principal components, and designs a spectral chip structure based on this, realizing targeted quantitative design for specific application spectral libraries, thereby improving the accuracy of spectral chip-reconstructed spectra. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0055] Figure 1 This is one of the flowcharts illustrating the spectral chip structure design method for a specific application spectral library provided in this embodiment of the invention;

[0056] Figure 2 This is one of the structural schematic diagrams of the spectral chip provided in the embodiments of the present invention;

[0057] Figure 3 This is one of the structural schematic diagrams of the vertical layers of the spectral chip provided in the embodiments of the present invention;

[0058] Figure 4 This is the second flowchart illustrating the spectral chip structure design method for a specific application spectral library provided in this embodiment of the invention.

[0059] Figure 5 This is a schematic diagram of the modulation layer of the spectral chip designed according to an embodiment of the present invention;

[0060] Figure 6 This is a schematic diagram of the sensor partitioning corresponding to the modulation layer obtained from the design provided in the embodiment of the present invention;

[0061] Figure 7 This is a schematic diagram of the first principal component provided in an embodiment of the present invention;

[0062] Figure 8 This is a structural schematic diagram of structural unit one provided in an embodiment of the present invention;

[0063] Figure 9 A schematic diagram of the transmission spectrum of structural unit one provided in this embodiment of the invention;

[0064] Figure 10 This is a schematic diagram of the structure of structural unit two provided in an embodiment of the present invention;

[0065] Figure 11 A schematic diagram of the transmission spectrum of structural unit two provided in this embodiment of the invention;

[0066] Figure 12 This is one of the schematic diagrams of the modulation layer of the spectral chip provided in the embodiments of the present invention;

[0067] Figure 13 This is a schematic diagram of the structure of a spectral chip based on a multilayer modulation layer provided in an embodiment of the present invention;

[0068] Figure 14 This is a second schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention;

[0069] Figure 15 This is a second schematic diagram of the modulation layer of the spectral chip provided in this embodiment of the invention;

[0070] Figure 16 This is the third schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention;

[0071] Figure 17 This is the fourth schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention;

[0072] Figure 18 This is the fifth schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention;

[0073] Figure 19 This is the sixth schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention;

[0074] Figure 20It is a schematic structural diagram of a spectral chip structure design device for a specific application spectral library provided by an embodiment of the present invention;

[0075] Figure 21 It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. Specific embodiments

[0076] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0077] The following combines Figure 1 - Figure 19 to describe a method for designing a spectral chip structure for a specific application spectral library provided by an embodiment of the present invention.

[0078] Figure 1 It is one of the flow schematic diagrams of a method for designing a spectral chip structure for a specific application spectral library provided by an embodiment of the present invention. As Figure 1 shown, the method for designing a spectral chip structure for a specific application spectral library provided by an embodiment of the present invention includes:

[0079] Step 110, obtaining a specific application spectral library and determining spectral principal components based on the specific application spectral library;

[0080] The specific application spectral library refers to a spectral data containing multiple samples to be measured. The spectral data can be spectral information carrying sample component characteristics such as reflection, transmission, or fluorescence of the samples to be measured under a specific application. The number of wavelength sampling points of each spectrum is n. The specific application spectral library can be obtained through methods such as querying relevant sample information, actual testing, or numerical simulation calculation. It can be understood that the specific application spectral library is designed according to requirements, so as to design a more adaptable spectral chip structure based on the specific application spectral library.

[0081] The spectral principal components can be one or more vectors, and this vector can summarize the data distribution of the specific application spectral library in a lower dimension. Exemplarily, for n wavelength sampling points, that is, n-dimensional data, n vectors are required to represent its data distribution. After performing spectral principal component analysis, only m (m < n) spectral principal components, that is, m-dimensional data, are needed to retain the main characteristics of the data distribution of the corresponding spectral library.

[0082] Step 120, performing non-negativity processing on the spectral principal components to obtain non-negative spectral principal components;

[0083] Nonnegation of spectral principal components refers to processing the numerical values ​​of spectral principal components to be nonnegative while ensuring that the meaning of the spectral principal components (such as relative magnitude) remains unchanged.

[0084] Step 130: Determine the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components.

[0085] Specifically, Figure 2 This is one of the structural schematic diagrams of the spectral chip provided in the embodiments of the present invention, such as... Figure 2 As shown, the spectral chip provided in this embodiment of the invention includes a modulation layer 210 and an image sensor 220. The modulation layer 210 has a micro-nano structure array, which can modulate incident light of different wavelengths in different ways. The image sensor 220 can capture visible light and near-infrared image information, that is, the image sensor 220 can receive the incident light signal modulated by the modulation layer 210. The image sensor can be a CMOS or CCD, etc.

[0086] That is, in this invention, the modulation layer 210 is formed on the upper surface of the photosensitive area of ​​the image sensor 220. The modulation layer 210 is composed of multiple structural units, each corresponding to one or more pixels on the image sensor 220. Each structural unit is composed of at least one set of micro / nano structure arrays, which are implemented as one-dimensional or two-dimensional periodic structures (micro / nano structures). The micro / nano structure arrays of each structural unit are used to modulate the incident light, encoding the spectral information of the incident light onto the pixels of the image sensor 220 to obtain a spectral signal containing the spectral information of the incident light. The measurement process can be as follows: the incident light enters from above the chip, passes through the light modulation layer 210, and is subjected to broadband modulation by the micro / nano structure arrays of each structural unit. The intensity of the modulated light signal is detected by the image sensor 220 and converted into an electrical signal. Optionally, the spectral chip may further include a signal processing circuit 230.

[0087] Optionally, refer to Figure 3 , Figure 3 This is one of the structural schematic diagrams of the vertical layers of the spectral chip provided in the embodiments of the present invention. Figure 3 Can be with Figure 2 Corresponding to the structure, the spectral chip provided in this embodiment of the invention may include a modulation layer 210 and an image sensor 220. Optionally, the spectral chip provided in this embodiment of the invention may further include a signal processing circuit 230 and / or a light-transmitting transition layer 240. 211 and 212 correspond to different structural units in the modulation layer 210, and different structural units are generally composed of different micro-nano structure arrays. The light-transmitting transition layer 240 can be a flat structure, which is beneficial to the formation of the modulation layer.

[0088] The transmission spectrum of a structural unit can be understood as the transmittance of light at various wavelengths after it is incident from above, passes through the structural unit, and finally reaches the image sensor. Determining the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components refers to determining the overall transmission spectrum of the modulation layer 210 of the spectral chip based on the non-negative spectral principal components. As mentioned above, the modulation layer 210 includes a unit array composed of multiple structural units, each of which consists of at least one set of micro / nano structure arrays. By setting different micro / nano structure arrays, different transmittances for incident light can be achieved, resulting in different spectral signals received by the image sensor, thus enabling the reconstruction of spectral information. For example, based on the non-negative spectral principal components, a specific application spectral library is determined to have significant peaks in the wavelength ranges a, b, and c. Therefore, the modulation layer 210 can be designed to have higher transmittance in the wavelength ranges a, b, and c compared to other wavelength ranges.

[0089] The spectral chip structure design method for specific application spectral libraries provided in this invention summarizes the spectral features of the specific application spectral library through principal components, reducing the dimensionality of the data space of the studied spectral library, facilitating the analysis of spectral library features, and ensuring that the principal components retain the features that contribute the most to the spectral library, thereby achieving an effective representation of the spectral library; therefore, this invention achieves quantitative evaluation of the spectral library by analyzing principal components, and designs a spectral chip structure based on this, realizing targeted quantitative design for specific application spectral libraries, thereby improving the accuracy of spectral chip-reconstructed spectra.

[0090] The following is a further explanation of the possible implementation methods of the above steps in specific embodiments.

[0091] Step 110: Obtain a specific application spectral library and determine the principal spectral components based on the specific application spectral library.

[0092] Optionally, determining the principal spectral components based on the specific application spectral library includes:

[0093] Step 111: Perform zero-mean processing on the specific application spectral library to obtain a preprocessed spectral library;

[0094] Specifically, zero-mean subtraction refers to subtracting the mean of the set from each individual data point. Optionally, after obtaining the mean spectrum of the spectral samples (spectral information) corresponding to the sample to be tested in the specific application spectral library, this mean spectrum is subtracted from each spectral sample; thus obtaining the preprocessed spectral library.

[0095] Step 112: Perform principal component analysis on the preprocessed spectral library to obtain at least one initial eigenvector and eigenvalues ​​corresponding to each initial eigenvector;

[0096] Specifically, the initial eigenvectors refer to the unfiltered eigenvectors obtained through principal component analysis.

[0097] Optionally, performing principal component analysis on the preprocessed spectral library to obtain at least one initial eigenvector includes:

[0098] Principal component analysis is performed on the preprocessed spectral library based on preset principal component analysis conditions to obtain at least one initial feature vector.

[0099] The preset principal component analysis conditions are as follows:

[0100]

[0101] A T A = I;

[0102] Obtaining at least one initial feature vector includes:

[0103] A = (a1, a2, ..., a n );

[0104] That is, to solve for the condition that satisfies Maximum and A T Given A = I, the columns of A are: eigenvectors.

[0105] Where tr represents the sum of all elements on the main diagonal. This refers to the preprocessed spectral library. express The transpose of , where A represents the characteristic matrix, A T Let I denote the transpose of the characteristic matrix, and let a denote the identity matrix. n This represents the initial feature vector.

[0106] It should be understood that those skilled in the art can calculate eigenvalues ​​based on the initial eigenvector using existing techniques, and the calculation method for eigenvalues ​​will not be elaborated here.

[0107] Step 113: Based on the eigenvalues, determine the principal component eigenvectors, which are used to represent the spectral principal components.

[0108] Specifically, eigenvalues ​​can reflect the characteristics of the initial eigenvectors. By filtering eigenvalues ​​(such as through numerical comparison), the eigenvectors that have a greater impact on the spectral library (such as having a larger proportion or a greater contribution rate) in the initial eigenvectors are determined, and the eigenvectors with a greater impact are taken as the principal component eigenvectors.

[0109] Optionally, determining the principal component feature vector based on the feature values ​​includes:

[0110] Step 1131: Obtain the contribution rate corresponding to each of the initial feature vectors, wherein the contribution rate is the ratio of the feature value corresponding to each initial feature vector to the sum of all the feature values;

[0111] For example, the eigenvalue of initial feature vector a1 is 1, the eigenvalue of initial feature vector a2 is 2, and the eigenvalue of initial feature vector a3 is 3; then the contribution rate of initial feature vector a1 is... The contribution rate corresponding to the initial feature vector a2 is The contribution rate corresponding to the initial feature vector a3 is

[0112] Step 1132: Use the initial feature vector that meets the preset contribution rate threshold as the principal component feature vector.

[0113] For example, if the preset contribution rate threshold is 30%, and the condition is not less than 30%, then the initial feature vectors that satisfy the preset contribution rate threshold are initial feature vector a2 and initial feature vector a3, and the initial feature vector a2 and initial feature vector a3 are used as principal component feature vectors.

[0114] It should be understood that the above examples are for the purpose of facilitating understanding of the present invention and should not be construed as limiting the present invention in any way. The preset contribution rate threshold and the conditions for satisfying the preset contribution rate threshold (including but not limited to preset contribution rate thresholds that are greater than, less than, not greater than, or not less than) can be set according to actual applications.

[0115] Optionally, the method further includes:

[0116] Step 1133: Optimize the contribution rate threshold based on the preset structural measurement and reconstruction accuracy requirements.

[0117] Optionally, optimizing the contribution rate threshold can be achieved by increasing the contribution rate threshold. For example, after using initial feature vectors a2 and a3 as principal component feature vectors, the spectrum reconstructed by the designed spectral chip has a 40% similarity to the spectrum in the spectral library. Given a preset structural measurement and reconstruction accuracy requirement of 70%, the preset contribution rate threshold can be increased from 30% to 60%. It should be understood that the above are examples provided for ease of understanding the present invention and should not constitute any limitation on the present invention.

[0118] Optionally, determining the principal component feature vector based on the feature values ​​includes:

[0119] Step 1134: Arrange the feature values ​​from largest to smallest, and determine the initial feature vectors corresponding to the first K feature values ​​as principal component feature vectors, where K is a preset positive integer.

[0120] Specifically, the eigenvalues ​​are arranged from largest to smallest, and the k eigenvalues ​​with the largest values ​​are selected according to the requirements. The corresponding eigenvectors are the principal components. For example, the eigenvalues ​​of a1, a2, and a3 are arranged from largest to smallest, and the eigenvectors corresponding to the first two eigenvalues ​​are selected. The initial eigenvectors a2 and a3 are then used as the principal component eigenvectors.

[0121] Optionally, the method further includes:

[0122] Step 1135: Optimize the preset K value based on the preset structural measurement and reconstruction accuracy requirements.

[0123] Specifically, optimizing the preset K value can involve reducing the K value. For example, after using initial feature vectors a2 and a3 as principal component feature vectors, the spectrum reconstructed by the designed spectral chip has a 40% similarity to the spectrum in the spectral library. Given a preset structural measurement and reconstruction accuracy requirement of 70%, the K value can be reduced from 2 to 1. It should be understood that the above are examples provided for ease of understanding of the invention and should not constitute any limitation on the invention.

[0124] Optionally, the spectral chip includes at least one structural unit, and the method further includes:

[0125] Based on the eigenvalues, a quantitative weighting is performed to determine the number of structural units corresponding to each principal component eigenvector.

[0126] In one embodiment, the weighting weight is set to 1, and the principal component eigenvectors correspond one-to-one with the structural units, that is, the transmission spectra of all spectral principal components correspond one-to-one with the structural units.

[0127] In one embodiment, different weights are assigned to the number of structural units corresponding to each principal component feature vector based on the contribution rate: that is, the spectral principal component with a larger contribution rate corresponds to a larger number of structural units. Optionally, the types of structural units can be determined based on similarity. For an introduction to similarity, please refer to step 130. The structural units are arranged in descending order of contribution rate: first principal component, second principal component, third principal component, ..., Kth principal component. The first principal component corresponds to the 6 structural units with the highest similarity, the second principal component corresponds to the 5 structural units with the highest similarity, the third principal component corresponds to the 4 structural units with the highest similarity, the fourth principal component corresponds to the 3 structural units with the highest similarity, the fifth principal component corresponds to the 2 structural units with the highest similarity, and the remaining K-5 principal components each correspond to 1 structural unit. The total number of structural units should be less than the number of wavelength sampling points n. The above number of structural units is only an example and can be adjusted according to whether the spectral reconstruction effect meets the preset target.

[0128] Step 120: Nonnegatively process the spectral principal components to obtain nonnegative spectral principal components.

[0129] Optionally, the nonnegation processing of the spectral principal components to obtain nonnegative spectral principal components includes:

[0130] The non-negative spectral principal components are obtained by taking the absolute value of the principal components of the spectrum; or

[0131] The preset supplementary value is added to the spectral principal component to obtain the non-negative spectral principal component.

[0132] Specifically, the nonnegation processing of the spectral principal components includes, but is not limited to, directly taking the absolute value of the spectral principal components or simultaneously adding a positive value. Optionally, the mean spectrum of the spectral library is added to all spectral principal components.

[0133] Step 130: Determine the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components.

[0134] Optionally, the transmission spectrum for determining the spectral chip structure based on the non-negative spectral principal components includes:

[0135] Step 131: Design the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components; or

[0136] Step 132: Based on the non-negative spectral principal components, select the transmission spectrum of the spectral chip structure from the pre-acquired candidate transmission spectra.

[0137] Specifically, using the non-negative spectral principal components as a reference, the structural transmission spectrum (i.e., the transmission spectrum of the spectral chip structure) is designed, or the transmission spectra of existing structures are selected. The higher the similarity between the structural transmission spectrum and the spectral principal components, the better the quality of the spectral chip structure design, and the better the spectral measurement and reconstruction effect. Here, good spectral chip structure design quality means that, for the same number of units, the likelihood, center wavelength accuracy, resolution, and other indicators of spectral measurement and reconstruction are better, or the fewer structural units are required to achieve the same spectral measurement and reconstruction effect. The structural transmission spectrum here includes, but is not limited to, methods obtained through numerical simulation or actual structural testing.

[0138] Optionally, the method further includes:

[0139] The similarity between the transmission spectra of the non-negative spectral principal components and the spectral chip structure is evaluated, and the transmission spectra of the spectral chip structure are optimized based on the similarity evaluation results.

[0140] Specifically, using the non-negative spectral principal component as a reference, the transmission spectrum of the spectral chip structure is designed, or the transmission spectrum of an existing structure is selected to construct the spectral chip structure. The transmission spectrum of the spectral chip structure has a high similarity to the spectral principal component, thereby improving the spectral measurement and reconstruction effect. For example, if there are six candidate spectral chip structures with transmission spectra, the two with the highest similarity can be selected based on the similarity between the transmission spectrum and the non-negative spectral principal component; or if the transmission spectrum of a candidate spectral chip structure has a similarity of 60% to the non-negative spectral principal component, it can be optimized based on the transmission spectrum of that candidate spectral chip structure to improve the similarity. It should be understood that the quantities in the embodiments of this invention are examples for ease of understanding and should not constitute any limitation on the embodiments of this invention.

[0141] The similarity between the structural transmission spectrum and the principal components can be measured by methods including, but not limited to, likelihood, correlation, pointwise difference, and root mean square error. The concept of correlation is defined as follows: if the transmission spectrum of a certain structure is X∈R... n×1 A principal component is Y∈R n×1 ;

[0142] Likelihood

[0143] Correlation is

[0144] The spectral chip structure design method for a specific application spectral library provided in this invention can analyze the main characteristics of the spectrum of a specific application based on the spectral library of samples under the existing application. The design or screening is carried out around the broadband modulation characteristics of the modulation layer based on the characteristics, so as to achieve high-precision measurement of the incident light spectrum and effectively improve the accuracy of spectral recovery.

[0145] Optionally, the method further includes:

[0146] Update the application-specific spectral library.

[0147] This invention provides an embodiment of the invention that obtains features or spectral samples that are not included in the specific application spectral library, repeats the spectral chip structure design, and further analyzes, optimizes and enriches the spectral chip structure design.

[0148] Figure 4 This is the second flowchart illustrating the spectral chip structure design method for a specific application spectral library provided in this embodiment of the invention. Figure 4 As shown, the spectral chip structure design method for a specific application spectral library provided in this embodiment of the invention includes:

[0149] Step 1: Obtain the spectral principal components of a specific application object (scene), specifically including:

[0150] 1) Obtain a spectral library for a specific application. A spectral library refers to the spectral information of a sample under test for a specific application, including its reflection, transmission, or fluorescence, which carries the characteristics of the sample's composition. Each spectrum has n wavelength sampling points. The spectral library can be obtained through methods such as querying relevant sample information, actual testing, or numerical simulation calculations.

[0151] 2) For zero-mean processing of the spectral library, after obtaining the mean spectrum of all spectral samples, the mean is subtracted from each sample; thus obtaining the preprocessed spectral library.

[0152] 3) Then, perform principal component analysis based on the preprocessed spectral library, i.e., solve for:

[0153] Where A T A = I, and we obtain a set of eigenvalues ​​and corresponding eigenvectors (columns of A).

[0154] That is, the spectral library for the target application (application-specific spectral library) is zero-mean normalized, and then principal component analysis is performed to obtain a set of eigenvalues ​​λ = (λ1, λ2, ..., λ3). n ) T and the corresponding eigenvectors A = (a1, a2, ..., a... n ).

[0155] 4) Arrange the feature values ​​from largest to smallest. Define the ratio of each feature value to the sum of all feature values ​​as the contribution rate of that feature. Then, set a contribution rate threshold and select the top k features whose sum of contribution rates satisfies the threshold. The feature vectors corresponding to these k feature values ​​are the principal components. Alternatively, this step can be done directly: select the k features with the largest feature values ​​according to the requirements; their corresponding feature vectors are the principal components.

[0156] Step 2: Further nonnegation processing is performed on the principal components of the spectrum, including but not limited to taking the absolute value directly or adding a positive value at the same time. Optionally, the mean spectrum of the spectral library is added to all principal components of the spectrum.

[0157] Step 3: Using the nonnegative principal components as a reference, design the transmission spectrum of the spectral chip structure or select from the transmission spectra of existing structures. The higher the similarity between the transmission spectrum of the spectral chip structure and the principal components, the better the design quality of the spectral chip structure, and the better the spectral measurement and reconstruction effect. Here, a good spectral chip structure design quality means that for the same number of units, the better the likelihood, center wavelength accuracy, resolution, and other indicators of spectral measurement and reconstruction, or the fewer units are required to achieve the same spectral measurement and reconstruction effect. The transmission spectrum of the structure here includes, but is not limited to, obtaining it through numerical simulation or testing actual structures. The similarity between the transmission spectrum of the structure and the principal components can be measured by methods including, but not limited to, likelihood, correlation, pointwise difference, and root mean square error. Among them, the relevant concept is defined as follows: If the transmission spectrum of a certain structure is X∈R n×1 A principal component is Y∈R n×1 ;

[0158] Likelihood

[0159] Correlation is

[0160] Step four: Optionally, based on the similarity between the principal spectral components and the structural transmission spectrum, the principal spectral components can be matched one-to-one with the structural units; alternatively, multiple structural units can be matched for a single principal spectral component to ensure sufficient preservation of its characteristics. In principle, the spectral chip structure should have high transmittance in the band corresponding to the principal spectral component.

[0161] Step 5: Optionally, based on the requirements for the accuracy of spectral chip structure measurement and reconstruction, the aforementioned contribution rate threshold can be further optimized to adjust the selection of feature values.

[0162] Step 6: Optionally, if features or spectral samples not previously included are encountered in subsequent research or testing, the aforementioned spectral library can be added to the spectral library, and the above steps can be repeated to further analyze, optimize, and enrich the spectral chip structure design.

[0163] The following describes the spectral chip structure design method for a specific application spectral library provided by the present invention, with reference to embodiments:

[0164] Example 1:

[0165] The spectral library for the target application is zero-mean normalized, and then principal component analysis is performed to obtain a set of eigenvalues ​​λ = (λ1, λ2, ..., λ3). n ) T and the corresponding eigenvectors A = (a1, a2, ..., a... n ), each feature vector a iThat is, a principal component of the spectrum, which is a feature in the spectral library, and its contribution rate is... The higher the contribution rate, the more important the feature is to this spectral library, and the higher its probability of occurrence. Sort the features by contribution rate from largest to smallest, set a contribution rate threshold Γ, and select the top k features, such that... These eigenvectors are the principal components of the spectrum. They are non-negatively processed, for example, by adding the mean spectrum of the spectral library and the absolute value of the smallest negative number among all eigenvectors. For each non-negative principal component, the structural unit with the most similar transmission spectrum is selected, for example, the one with the highest correlation coefficient. A series of structural units selected according to the method provided in this embodiment are used to prepare the structural units of the modulation layer for testing or numerical simulation to calculate the effect on the measurement reconstruction of samples in the spectral library. If the accuracy requirements are met, it is used as the final structural unit; if not, the threshold Γ is increased, and the above process is repeated. For example, if k=9 is finally obtained, the structural units corresponding to the transmission spectra of the first 9 features are selected to form a spectral pixel. A spectral pixel refers to a set of structural units used to realize the complete spectral measurement reconstruction function, including structural units of at least two types of modulated incident light. (Reference) Figure 5 and Figure 6 , Figure 5 This is a schematic diagram of the modulation layer of the spectral chip designed according to an embodiment of the present invention. Figure 6 This is a schematic diagram of the sensor partitioning corresponding to the modulation layer obtained from the design provided in the embodiment of the present invention.

[0166] Example 2:

[0167] The spectral library for the target application is zero-mean normalized, followed by principal component analysis to obtain a series of eigenvectors whose contribution rates satisfy the threshold Γ, which are then non-negatively normalized. Using theories such as structural inverse design, this set of non-negative principal components is used as the transmission spectrum to inversely design the micro / nanostructure array of the modulation layer. For transmission spectra that cannot be realized, structural units with similar transmission spectra can be selected as substitutes, or the spectrum can be divided into *a* parts and implemented separately, resulting in the design of *a* corresponding micro / nanostructure arrays. These *a* micro / nanostructure arrays constitute at least one structural unit, which is then added to the final designed micro / nanostructure array.

[0168] Example 3:

[0169] The spectral library A for the target application is zero-mean normalized, followed by principal component analysis. Finally, a series of eigenvectors whose contribution rates satisfy the threshold Γ are obtained and non-negatively normalized. The first p principal components are selected and retained for further structural transmission spectrum matching. After calculating the likelihood of each structural transmission spectrum with the first principal component, the two structures with the highest likelihood are structural unit one and structural unit two. (Refer to...) Figure 7 , Figure 8 , Figure 9, Figure 10 and Figure 11 , Figure 7 This is a schematic diagram of the first principal component provided in an embodiment of the present invention. Figure 8 This is a structural schematic diagram of structural unit one provided in an embodiment of the present invention. Figure 9 A schematic diagram of the transmission spectrum of structural unit one provided in this embodiment of the invention. Figure 10 This is a structural schematic diagram of structural unit two provided in an embodiment of the present invention. Figure 11 A schematic diagram of the transmission spectrum of structural unit two provided in this embodiment of the invention. The likelihood of structural unit one with the first principal component is 0.965; the likelihood of structural unit two with the first principal component is 0.953. The first principal component can be understood as the component with the highest contribution rate in the spectral library.

[0170] The two structural units have high likelihoods with the first principal component and highlight the features of different parts of the principal component; therefore, these two structures are retained for the first principal component. For other principal components, the structural unit with the highest likelihood is retained, resulting in p+1 structures in the final spectral pixels.

[0171] The spectral chip structure design method for a specific application spectral library provided in this invention extracts features from the spectral library for a specific application based on principal component analysis. It designs the structural units of the modulation layer of the spectral chip according to the main features of the target application spectrum, or can be understood as designing the micro-nano structure array corresponding to the structural units. This clarifies the structural design objective, improves the measurement recovery accuracy of the spectral chip, and helps to further reduce the number of structural units required.

[0172] The spectral chip designed according to an embodiment of the present invention will be further described below:

[0173] The structural transmission spectrum (i.e., the transmission spectrum of a spectral chip structure) refers to the transmittance of light at various wavelengths when incident from above through the structural unit. Figure 12 This is one of the schematic diagrams of the modulation layer of the spectral chip provided in the embodiments of the present invention, such as... Figure 3 and Figure 12 As shown, 211 and 212 are different structural units with different transmission spectra. Their structures may be one-dimensional gratings, two-dimensional periodic aperture structures, etc. The basic shape of the aperture can be a regular shape such as a circle, polygon, ellipse, or cross, or it can be an irregular shape. The period, duty cycle, and other parameters of the grating and periodic aperture structure can be adjusted. The lattice of the periodic aperture structure can be square or other polygonal structures. That is, based on the structural transmission spectra, the structural units of the spectral chip can be designed and fabricated to a certain extent, making the spectral chip more suitable for the corresponding specific application scenarios.

[0174] The modulation layer material can be selected from silicon, silicon compounds, germanium, germanium compounds, germanium-silicon materials, III-V group materials, etc. Compared with materials such as metals, it has less optical loss. Furthermore, it can be further optionally achieved by making the overall structure pattern C4 symmetrical (it can coincide after rotating 90°, 180°, or 270°) to achieve the characteristic of insensitivity to incident light polarization.

[0175] A light-transmitting transition layer 240 can be provided between the modulation layer 210 and the image sensor layer 220, consisting of polymers such as PDMS and SU-8, or compounds such as silicon oxide and sapphire, to facilitate processing or to achieve filtering and other functions after further processing. Figure 3 As shown, this layer may or may not be necessary.

[0176] The modulation layer can consist of one or more layers (including two layers). When multiple layers are stacked as the modulation layer, the final transmission spectrum is determined by the combined effect of the multiple layers. When the optical path between two structural units is long, it can be considered as the superposition (multiplication) of the transmission spectra of each layer. In this embodiment of the invention, if the multiple layers are bonded together, it should be understood that the multiple layers interact with each other, which can be understood as a whole, corresponding to the generation of a new structural transmission spectrum T. The structural transmission spectrum T is not simply the superposition of the structural transmission spectra of each layer, but rather it is understood that each structural unit in each layer influences each other to generate the structural transmission spectrum. In the multi-layer structure scheme, preferably, it can be understood that each layer corresponds to different structural units, for example, different transmission spectra, different structures, or different sizes of the same structure; this is not limited here.

[0177] Furthermore, the advantages of this embodiment are described in conjunction with the technical solution of the present invention. The present invention obtains the principal components of the spectral library and then designs the corresponding structural transmission spectrum based on the principal components. However, in certain cases, the required principal components need to be modulated by complex structural transmission spectra. However, the complexity of the structural transmission spectrum is related to the structural complexity to a certain extent. That is, a complex structure can make the structural transmission spectrum complex. However, the design of complex structures is difficult, and the processing is even more difficult, especially in ensuring accuracy. In this embodiment, the multilayer structure (two layers) can reduce the complexity of each layer structure, and then obtain a complex structural transmission spectrum by modulating the two together.

[0178] Furthermore, the present invention provides specific embodiments to illustrate the multilayer structure. Figure 13 This is a schematic diagram of the structure of a spectral chip based on a multi-layer modulation layer provided in an embodiment of the present invention, as shown below. Figure 13As shown, taking two layers as an example, the transmission spectrum of the structure of the first modulation layer 210-1 and the corresponding structure of the second modulation layer 210-2 are different. The so-called corresponding region can be understood as follows: after a certain incident light enters a certain modulation unit A of the first modulation layer, it enters the modulation unit B of the second modulation layer and is then received by the sensing unit. Thus, the modulation unit A of the first modulation layer and the modulation unit B of the second modulation layer can be called corresponding.

[0179] Figure 14 This is a second schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention, as shown below. Figure 14 As shown, the structure of each layer can be divided, and the structural units of each layer correspond to each other.

[0180] It is possible to design such that at least one unit area of ​​a certain layer is either completely etched through or completely unetched. Figure 15 This is the second schematic diagram of the modulation layer of the spectral chip provided in this embodiment of the invention. Figure 16 This is the third schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention, as shown below. Figure 15 and Figure 16 As shown, the modulation layer includes multiple structural units: structural unit 210-11, structural unit 210-22, structural unit 210-33, structural unit 210-44, structural unit 210-55, and structural unit 210-66; structural unit 210-11 includes micro-nano structure arrays 11-0, 11-1, 11-2, and 11-3; structural unit 210-44 includes micro-nano structure arrays 44-0, 44-1, 44-2, and 44-3, wherein micro-nano structure array 11-0 represents completely transparent, and micro-nano structure array 44-0 represents completely opaque. Figure 16 In the first modulation layer, the left side can represent complete etching, and the blank area on the right side of the first modulation layer can represent no etching at all.

[0181] Figure 17 This is the fourth schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention, as shown below. Figure 17 As shown, the structural units between layers may not correspond one-to-one, and some layers may even be only a periodic structure.

[0182] Figure 18 This is the fifth schematic diagram of the structure of the spectral chip provided in the embodiment of the present invention, as shown below. Figure 18As shown, transition layers 240, as described above, can also be added between the sub-layers of the modulation layer to facilitate processing or enhance filtering properties. Preferably, the transition layer 4 has a low refractive index. Furthermore, in this embodiment, the transition layer 240 and the connecting layer 240-2 can also be understood as modulation layers. That is, multiple modulation layers work together to increase the overall complexity, making the transmission spectrum of the structure corresponding to the modulation layer conform to the spectral library of the specific application. In other words, the multiple layers in this invention can be two or more layers, the essence of which is to obtain a transmission spectrum of a structure that is closer to the requirements.

[0183] Figure 19 This is the sixth schematic diagram of the spectral chip provided in the embodiment of the present invention, as shown below. Figure 19 As shown, in other embodiments, the first modulation layer and the second modulation layer may have the same shape, but at least one layer contains filler material, thus the corresponding regions of the first modulation layer and the second modulation layer can be considered different; if both layers have filler material, the filler material properties can also be different to achieve the desired effect. Furthermore, the same first modulation layer and the same second modulation layer may have deviations, such as offset, rotation, or deflection. Furthermore, a protective layer may be provided, formed above the first modulation layer.

[0184] It should be understood that the materials of each modulation layer can be the same or different materials.

[0185] The spectral chip structure provided in this invention, through a multi-layer stacking design, can realize a modulation layer composed of multiple structures on a single chip, enriching the types of transmission spectra in the structural design.

[0186] The following describes the spectral chip structure design apparatus for a specific application spectral library provided by the present invention. The spectral chip structure design apparatus for a specific application spectral library described below and the spectral chip structure design method for a specific application spectral library described above can be referred to in correspondence with each other.

[0187] Figure 20 This is a schematic diagram of the spectral chip structure design device for a specific application spectral library provided in an embodiment of the present invention, as shown below. Figure 20 As shown, the spectral chip structure design apparatus for a specific application spectral library provided in this embodiment of the invention includes:

[0188] The acquisition unit 2010 is used to acquire a specific application spectral library and determine the spectral principal components based on the specific application spectral library;

[0189] Processing unit 2020 is used to perform nonnegation processing on the spectral principal components to obtain nonnegative spectral principal components;

[0190] The determining unit 2030 is used to determine the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components.

[0191] Optionally, the acquisition unit 2010 is configured to determine the principal spectral components based on the specific application spectral library, including:

[0192] The acquisition unit 2010 is used to perform zero-mean processing on the specific application spectral library to obtain a preprocessed spectral library.

[0193] The acquisition unit 2010 is used to perform principal component analysis on the preprocessed spectral library to obtain at least one initial feature vector and feature values ​​corresponding to each initial feature vector.

[0194] The acquisition unit 2010 is used to determine the principal component feature vector based on the feature value, and the principal component feature vector is used to represent the spectral principal component.

[0195] Optionally, the acquisition unit 2010 is used to perform principal component analysis on the preprocessed spectral library to obtain at least one initial feature vector, including:

[0196] The acquisition unit 2010 is used to perform principal component analysis on the preprocessed spectral library based on preset principal component analysis conditions to obtain at least one initial feature vector.

[0197] The preset principal component analysis conditions are as follows:

[0198]

[0199] A T A = I;

[0200] Obtaining at least one initial feature vector includes:

[0201] Obtain A = (a1, a2, ..., a n );

[0202] Where tr represents the sum of all elements on the main diagonal. This refers to the preprocessed spectral library. express The transpose of , where A represents the characteristic matrix, A T Let I denote the transpose of the characteristic matrix, and let a denote the identity matrix. n This represents the initial feature vector.

[0203] Optionally, the acquisition unit 2010 is used to determine the principal component feature vector based on the feature values, including:

[0204] The acquisition unit 2010 is used to acquire the contribution rate corresponding to each initial feature vector, wherein the contribution rate is the ratio of the feature value corresponding to each initial feature vector to the sum of all feature values;

[0205] The acquisition unit 2010 is used to take the initial feature vector that meets the preset contribution rate threshold as the principal component feature vector.

[0206] Optionally, the acquisition unit 2010 is further configured to optimize the contribution rate threshold based on preset structural measurement and reconstruction accuracy requirements.

[0207] Optionally, the acquisition unit 2010 is used to determine the principal component feature vector based on the feature values, including:

[0208] The acquisition unit 2010 is used to arrange the feature values ​​from largest to smallest and determine the initial feature vectors corresponding to the first K feature values ​​as principal component feature vectors, where K is a preset positive integer.

[0209] Optionally, the acquisition unit 2010 is further configured to optimize the preset K value based on the preset structural measurement and reconstruction accuracy requirements.

[0210] Optionally, the spectral chip includes at least one structural unit, and the device further includes a weighting unit, which is used to assign quantitative weights based on the feature values ​​to determine the number of structural units corresponding to each principal component feature vector.

[0211] Optionally, the processing unit 2020 is configured to perform nonnegation processing on the spectral principal components to obtain nonnegative spectral principal components, including:

[0212] The processing unit 2020 is configured to take the absolute value of the spectral principal components to obtain the non-negative spectral principal components; or

[0213] The processing unit 2020 is used to add the preset supplementary value to the spectral principal component to obtain the non-negative spectral principal component.

[0214] Optionally, the apparatus further includes an evaluation unit, which is used to evaluate the similarity between the transmission spectra of the non-negative spectral principal components and the spectral chip structure, and to optimize the transmission spectra of the spectral chip structure based on the similarity evaluation results.

[0215] Optionally, the similarity assessment method includes any of the following:

[0216] Likelihood calculation;

[0217] Correlation calculation;

[0218] Find the difference point by point;

[0219] Calculation of root mean square difference.

[0220] Optionally, the apparatus further includes an update unit for updating the application-specific spectral library.

[0221] It should be noted that the apparatus provided in this embodiment of the invention can implement all the method steps implemented in the above method embodiment and can achieve the same technical effect. Therefore, the parts and beneficial effects that are the same as those in the method embodiment will not be described in detail here.

[0222] Figure 21 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 21 As shown, the electronic device may include a processor 2110, a communications interface 2120, a memory 2130, and a communication bus 2140, wherein the processor 2110, the communications interface 2120, and the memory 2130 communicate with each other via the communication bus 2140. The processor 2110 can call logical instructions in the memory 2130 to execute a spectral chip structure design method for a specific application spectral library. This method includes: acquiring the specific application spectral library and determining the spectral principal components based on the specific application spectral library; performing nonnegative processing on the spectral principal components to obtain nonnegative spectral principal components; and determining the transmission spectrum of the spectral chip structure based on the nonnegative spectral principal components.

[0223] Furthermore, the logical instructions in the aforementioned memory 2130 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0224] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute a spectral chip structure design method for a specific application spectral library provided by the methods described above. The method includes: acquiring a specific application spectral library and determining spectral principal components based on the specific application spectral library; performing nonnegative processing on the spectral principal components to obtain nonnegative spectral principal components; and determining the transmission spectrum of the spectral chip structure based on the nonnegative spectral principal components.

[0225] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for designing a spectral chip structure for a specific application spectral library provided by the methods described above. This method includes: acquiring the specific application spectral library and determining spectral principal components based on the specific application spectral library; performing nonnegative processing on the spectral principal components to obtain nonnegative spectral principal components; and determining the transmission spectrum of the spectral chip structure based on the nonnegative spectral principal components.

[0226] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0227] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0228] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for designing a spectral chip structure for a specific application spectral library, characterized in that, The spectral chip includes an image sensor and a modulation layer located above the image sensor, the modulation layer including at least one structural unit, and the method includes: Acquire a specific application spectral library and determine the principal spectral components based on the specific application spectral library; The principal components of the spectrum are nonnegated to obtain nonnegative principal components. The transmission spectrum of the spectral chip structure is determined based on the non-negative spectral principal components. The transmission spectrum used to determine the spectral chip structure based on the non-negative spectral principal components includes: Based on the non-negative spectral principal components, design the transmission spectrum of the spectral chip structure; or Based on the non-negative spectral principal components, the transmission spectrum of the spectral chip structure is selected from the pre-acquired candidate transmission spectra; The similarity between the transmission spectra of the non-negative spectral principal component and the spectral chip structure is evaluated. Based on the similarity evaluation results, the transmission spectrum of the spectral chip structure is optimized. Specifically, the transmission spectrum of the spectral chip structure is designed with the non-negative spectral principal component as a reference, or the transmission spectra of existing structures are selected to form the spectral chip structure, wherein the transmission spectrum of the spectral chip structure has high similarity with the non-negative spectral principal component.

2. The spectral chip structure design method for a specific application spectral library according to claim 1, characterized in that, The determination of spectral principal components based on the specific application spectral library includes: The specific application spectral library is subjected to zero-mean processing to obtain a preprocessed spectral library; Principal component analysis is performed on the preprocessed spectral library to obtain at least one initial eigenvector and eigenvalues ​​corresponding to each initial eigenvector; Based on the eigenvalues, principal component eigenvectors are determined, which are used to represent the spectral principal components.

3. The spectral chip structure design method for a specific application spectral library according to claim 2, characterized in that, The step of performing principal component analysis on the preprocessed spectral library to obtain at least one initial eigenvector includes: Principal component analysis is performed on the preprocessed spectral library based on preset principal component analysis conditions to obtain at least one initial feature vector. The preset principal component analysis conditions are as follows: ； ； Obtaining at least one initial feature vector includes: get ; in, This represents the sum of all elements on the main diagonal. This refers to the preprocessed spectral library. express transpose, Represents the characteristic matrix, This represents the transpose of the characteristic matrix. Represents the identity matrix. This represents the initial feature vector.

4. The spectral chip structure design method for a specific application spectral library according to claim 2, characterized in that, Based on the aforementioned eigenvalues, determining the principal component eigenvectors includes: Obtain the contribution rate corresponding to each of the initial feature vectors, wherein the contribution rate is the ratio of the feature value corresponding to each initial feature vector to the sum of all the feature values; The initial feature vectors that meet the preset contribution rate threshold are used as the principal component feature vectors.

5. The spectral chip structure design method for a specific application spectral library according to claim 4, characterized in that, The method further includes: The contribution rate threshold is optimized based on the preset structural measurement and reconstruction accuracy requirements.

6. The spectral chip structure design method for a specific application spectral library according to claim 2, characterized in that, Based on the aforementioned eigenvalues, determining the principal component eigenvectors includes: The eigenvalues ​​are arranged from largest to smallest, and the initial eigenvectors corresponding to the first K eigenvalues ​​are determined as principal component eigenvectors, where K is a preset positive integer.

7. The spectral chip structure design method for a specific application spectral library according to claim 6, characterized in that, The method further includes: Based on the preset structural measurement and reconstruction accuracy requirements, the preset K value is optimized.

8. The method for designing a spectral chip structure for a specific application spectral library according to any one of claims 4-7, characterized in that, The spectral chip includes at least one structural unit, and the method further includes: Based on the eigenvalues, a quantitative weighting is performed to determine the number of structural units corresponding to each principal component eigenvector.

9. The spectral chip structure design method for a specific application spectral library according to claim 1, characterized in that, The nonnegation process of the spectral principal components to obtain nonnegative spectral principal components includes: The non-negative spectral principal components are obtained by taking the absolute value of the principal components of the spectrum; or The preset supplementary value is added to the spectral principal component to obtain the non-negative spectral principal component.

10. The spectral chip structure design method for a specific application spectral library according to claim 9, characterized in that, The similarity assessment method includes any of the following: Likelihood calculation; Correlation calculation; Find the difference point by point; Calculation of root mean square difference.

11. The method for designing a spectral chip structure for a specific application spectral library according to any one of claims 1-7, or the method for designing a spectral chip structure for a specific application spectral library according to claim 9, or the method for designing a spectral chip structure for a specific application spectral library according to claim 10, characterized in that, The method further includes: Update the application-specific spectral library.

12. A spectral chip structure design apparatus for a specific application spectral library, applied to the spectral chip structure design method for a specific application spectral library as described in any one of claims 1 to 11, characterized in that, include: An acquisition unit is used to acquire a specific application spectral library and determine the spectral principal components based on the specific application spectral library; The processing unit is used to perform nonnegation processing on the spectral principal components to obtain nonnegative spectral principal components; The determining unit is used to determine the transmission spectrum of the spectral chip structure based on the non-negative spectral principal components.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the spectral chip structure design method for a specific application spectral library as described in any one of claims 1 to 11.

14. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the spectral chip structure design method for a specific application spectral library as described in any one of claims 1 to 11.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the spectral chip structure design method for a specific application spectral library as described in any one of claims 1 to 11.

Images