A multi-input spectrum reconstruction method and computational reconstruction on-chip spectrometer

Abstract

The application discloses a multi-input spectrum reconstruction method and a computing reconstruction type on-chip spectrometer. The on-chip spectrometer has multiple input / output ports, and the core area is an array waveguide containing a thermo-optic tunable component, so that multiple to-be-detected spectrums can be detected in parallel. Meanwhile, a reconstruction model is introduced, and a mapping relationship between the to-be-detected spectrums under the multiple-port input and measured patterns is synchronously established, so that the high-efficiency parallel detection of the spectrometer system of multiple to-be-detected spectrums is realized. The application can realize the high-resolution and large-window parallel detection of multiple to-be-detected spectrums by means of a single spectrometer, and greatly improves the detection efficiency.

Description

Technical Field

[0001] The present invention relates to a spectrum reconstruction method, in particular to a multi-input spectrum reconstruction method and a computational reconstruction-type on-chip spectrometer for realizing multi-spectrum parallel detection. Background Art

[0002] As an important scientific research tool, spectrometers are currently widely used in fields such as agricultural testing, medical analysis, astronomical research, and optical coherence tomography. Typically, in these applications, it is necessary to achieve parallel detection of multiple spectral data while maintaining high resolution and wide bandwidth.

[0003] For example, in aerospace, large-scale spectral surveys using large-field telescopes require dividing the large aperture into multiple sub-fields of view, which are then connected in parallel via optical fibers to hundreds of spectrometer systems. Increasing the number of spectrometers means higher detection efficiency. At the same time, high resolution must be maintained to resolve subtle features, revealing key information about the properties and composition of celestial objects. A larger operating window is also required to observe more celestial objects or more features of the same object.

[0004] However, traditional spectrometer systems face bottlenecks due to their large size and expensive components. Notably, multi-target detection techniques typically require array-based spectrometer systems, exacerbating this bottleneck. In recent years, although on-chip integrated spectrometers have gained great attention due to their compact size and cost-effectiveness, they still rely on the use of array stacking of functional units. This is often limited by large-scale optical systems, complex control circuits and high costs. In addition, achieving high resolution and large bandwidth in current spectral parallel detection has proven to be difficult. Therefore, it is of vital importance to propose a new multi-spectral parallel detection architecture to achieve high resolution, large window and multi-spectral parallel detection within a single spectrometer. Summary of the Invention

[0005] In response to the problems existing in the background technology, the purpose of the present invention is to provide a computationally reconstructed on-chip spectrometer that supports high-resolution, large-window, and multi-spectral parallel detection. The present invention proposes an on-chip spectrometer including a thermo-optical tunable component. The spectrometer has multiple input / output ports, and the core area realizes parallel detection of multiple spectra to be measured. At the same time, a reconstruction method is introduced to synchronously establish a mapping relationship between the spectra to be measured and the measured patterns under multi-port input, thereby realizing efficient parallel detection of multiple spectra to be measured. The present invention can achieve high-resolution, large-window parallel detection of multiple spectra to be measured with a single spectrometer, greatly improving the detection efficiency.

[0006] The technical solution adopted in the present invention is:

[0007] 1. A multi-input spectrum reconstruction method

[0008] Step 1: Laying a thermo-optical tunable component in the arrayed waveguide area of ​​the on-chip spectrometer. Specifically, a corresponding thermo-optical tunable component is laid on each waveguide in the arrayed waveguide area, so that the phase distribution of light on each waveguide is adjustable.

[0009] Step 2: Randomly activate the thermo-optical tunable components and then calibrate the calibration matrix of the current on-chip spectrometer;

[0010] Step 3: While maintaining the same random activation state as in step 2, multiple spectra to be measured are input to the current on-chip spectrometer at the same time, and the on-chip spectrometer outputs the detection spectrum;

[0011] Step 4: Based on the calibration matrix and detection spectrum of the current on-chip spectrometer, optimize and reconstruct the multiple spectra to be measured to obtain multiple reconstructed spectra.

[0012] The reconstruction method further comprises the following steps:

[0013] Step 5: If the accuracy of the current reconstructed spectrum does not meet the target accuracy, repeat steps 2 to 4 to reconstruct the spectrum again until the reconstructed spectrum with the optimal accuracy is obtained.

[0014] In step 2, the calibration matrix of the current on-chip spectrometer is calibrated, specifically:

[0015] A plurality of known spectra are input to the current on-chip spectrometer, and an output spectrum is measured; and a calibration matrix of the current on-chip spectrometer is calculated based on the plurality of known spectra and the corresponding output spectra.

[0016] In step 4, an optimization target for the kth spectrum to be measured is constructed, and the kth spectrum to be measured is optimized iteratively according to the constructed optimization target, so that the reconstructed spectrum corresponding to the kth spectrum to be measured is obtained after the optimization target is minimized. The formula of the optimization target for the kth spectrum to be measured is as follows:

[0017]

[0018] in, To reconstruct the spectrum, O N×1 To detect the spectrum, D() represents the difference operation, K is the number of spectra to be measured, α 1,k , α 2,k and α 3,k are the first weight - the third weight respectively; A N×M,k is the calibration matrix of the kth spectrum to be measured, S M×1,k is the kth spectrum to be measured, ||||2 is the L2 norm; ||||1 is the L1 norm.

[0019] The thermo-optical tunable component includes an electrode.

[0020] The accuracy of the reconstructed spectrum is specifically the relative reconstruction error ε r , the calculation formula is as follows:

[0021]

[0022] Among them, S M×1,k is the kth spectrum to be measured, is the reconstructed spectrum, and ||||2 is the L2 norm.

[0023] The number of input waveguides of the on-chip spectrometer is greater than or equal to the number of spectra to be measured.

[0024] 2. A computationally reconstructable on-chip spectrometer for multi-spectral parallel detection

[0025] The computational reconstruction on-chip spectrometer includes an on-chip spectrometer including a thermo-optical tunable component, a spectrometer calibration module, and a spectrum optimization and reconstruction module;

[0026] A spectrometer calibration module, used for randomly activating the thermo-optical tunable component and obtaining a calibration matrix of the spectrometer in the activated state;

[0027] An on-chip spectrometer including a thermo-optical tunable component for generating a plurality of detection spectra corresponding to the measured spectra;

[0028] The spectrum optimization and reconstruction module is used to reconstruct multiple spectra to be measured according to the calibration matrix and the detection spectrum to obtain reconstructed spectra.

[0029] The computational reconstruction on-chip spectrometer further includes a spectrum accuracy optimization module, which is used to calculate the accuracy of the reconstructed spectrum and optimize the reconstructed spectrum.

[0030] The present invention has the following beneficial effects:

[0031] This paper proposes, for the first time, a waveguide-based multi-spectral parallel detection spectrometer. For multi-port input spectra, the combination of an on-chip spectrometer containing thermo-optical tunable components and a reconstruction model allows for a larger operating window and improved spectral detection resolution. Furthermore, this invention can simultaneously reconstruct multiple spectra from multiple objects using a single spectrometer, significantly improving the efficiency of spectral detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Flow chart of the method of the present invention.

[0033] Figure 2 It is a schematic diagram of the architecture of the present invention.

[0034] Figure 3Schematic diagram of a single random spectral response generating unit according to an embodiment of the present invention.

[0035] Figure 4 3 is a schematic diagram of the spectrum of the multi-spectral parallel detection spectrometer in the embodiment to restore the four-channel input spectrum.

[0036] In the figure: 1-input port, 2-single random spectral response generating unit, 3-output port, 4-input waveguide, 5-output waveguide, 6-first slab area, 7-arrayed waveguide area with heating electrodes, 8-second slab area. DETAILED DESCRIPTION

[0037] The present invention will be further described below with reference to the accompanying drawings and examples.

[0038] like Figure 1 As shown, the multi-input spectrum reconstruction method proposed by the present invention includes the following steps:

[0039] Step 1: Laying electrodes in the arrayed waveguide area of ​​the on-chip spectrometer. Specifically, laying corresponding electrodes on each waveguide in the arrayed waveguide area makes the phase distribution of light on each waveguide adjustable.

[0040] Step 2: To increase the disorder of the spectrum, randomly activate the electrodes and then calibrate the calibration matrix of the current on-chip spectrometer;

[0041] Among them, the calibration matrix of the current on-chip spectrometer is calibrated as follows:

[0042] A plurality of known spectra are input to the current on-chip spectrometer, and an output spectrum is measured; and a calibration matrix of the current on-chip spectrometer is calculated based on the plurality of known spectra and the corresponding output spectra.

[0043] Step 3: Activate the electrodes in the same way as in step 2, input multiple spectra to be measured into the on-chip spectrometer through the multi-input waveguides of the on-chip spectrometer, and output the detection spectrum from the output waveguide of the on-chip spectrometer. Each output port of the on-chip spectrometer will have an output spectrum line (the horizontal axis is wavelength, and the vertical axis is energy). All output ports are finally synthesized into a matrix (the X-axis is the port number, the Y-axis is wavelength, and the Z-axis is energy), which is the detection spectrum.

[0044] Step 4: Based on the calibration matrix and the detection spectrum, multiple spectra to be measured are optimized and reconstructed to obtain multiple reconstructed spectra.

[0045] The optimization reconstruction in step 4 can be simply understood as establishing a connection between the measured pattern and the spectrum array to be measured using the mapping network in the spectrometer. The mapping network is a three-dimensional matrix A. Mathematically, the detected signal O N The spectral response matrix A of the architecture and the incident unknown spectrum array S are expressed ask The integral of (λ), that is:

[0046]

[0047] Here, λ start and λ stop are the starting wavelength and cut-off wavelength of the working band respectively. n,k (λ) is the calibration matrix. S k (λ) is the incident unknown multi-spectral array, K is the number of spectra to be measured, and k is the order of the incident ports. n,k is the signal detected by all output channels after a single spectrum is input from the kth channel. N represents the number of tuning states of the random module. n It is the sum of the signals detected by all output channels after multiple spectra are incident simultaneously, that is, the actual detected optical power. k Represents the spectrum mapping under a specific k-th port input. After the unknown spectrum is uniformly discretized into M columns, the above formula can be rewritten as:

[0048]

[0049] Among them, A N×M,K A is a three-dimensional matrix that represents the tuning state / wavelength / incident port order dependence of the architecture’s spectral response. k Each row vector in A corresponds to the spectral response of a specific sampling step, and k Each column vector in represents the temporal speckle of a single wavelength. N×M,K Matrix and O N×1 It can be measured experimentally, so S can be directly obtained M×1,K The solution.

[0050] However, in this case, the system is usually unconstrained, so the spectral reconstruction problem is transformed into an inverse matrix process, requiring optimization of regularization techniques to specify a unique solution. Here, it is assumed that the spectral values ​​are positive and exhibit continuous smooth characteristics, which is reasonable based on the properties of the spectrum.

[0051] In order to achieve spectral reconstruction of multiple spectra with narrowband, broadband, or even complex characteristics, a solver is needed to solve the current multispectral reconstruction problem. The present invention proposes an objective function for the current multispectral reconstruction problem. For the optimal reconstruction of the kth spectrum to be measured, the formula of the optimization objective is as follows. According to the constructed optimization objective, the kth spectrum to be measured is optimized and iterated, and the reconstructed spectrum corresponding to the kth spectrum to be measured is obtained after minimizing the optimization objective:

[0052]

[0053] in, To reconstruct the spectrum, O N×1 To detect the spectrum, min SM×1,k,SM×1,k≥0 (·) is the spectrum to be measured S M×1,k The global optimal value of k is the minimum output power and non-negative. The first term is used to minimize the detection spectrum O N×1 and the calculated α 1,k ·A N×M,k ·S M×1,k The second regularization term is used to promote the measured spectrum S M×1,k The sparsity of the third difference regularization term is used to smooth S M×1,k Here α 1,k , α 2,k and α 3,k is the weight of the corresponding item, and the standard k-fold cross-validation technique is used to determine the appropriate parameter α 1,k , α 2,k and α 3,k D() represents the difference operation, K is the number of spectra to be measured, α 1,k , α 2,k and α 3,k are the first weight - the third weight respectively; A N×M,k is the calibration matrix of the kth spectrum to be measured, S M×1,k is the kth spectrum to be measured, ||||2 is the L2 norm, which is the square root of the sum of the squares of the absolute values ​​of the vector elements; ||||1 is the L1 norm, which is the sum of the absolute values ​​of each element in the pointer quantity.

[0054] Step 5: If the current reconstructed spectrum does not meet the target accuracy, repeat steps 2-4 to reconstruct the spectrum until the optimal reconstructed spectrum is obtained. Experiments have shown that the more irregular the electrode activation, the higher the accuracy of the reconstructed spectrum.

[0055] The accuracy of the reconstructed spectrum is specifically the relative reconstruction error ε r , the calculation formula is as follows:

[0056]

[0057] Among them, S M×1,k is the kth spectrum to be measured, To reconstruct the spectrum, ||||2 is the L2 norm, which is the square root of the sum of the squares of the absolute values ​​of the vector elements.

[0058] The present invention also provides a computationally reconstructed on-chip spectrometer for multi-spectral parallel detection, which includes an on-chip spectrometer containing electrodes, a spectrometer calibration module, and a spectrum optimization and reconstruction module; wherein parallel detection of spectra refers to the technology of synchronously collecting multiple spectral information of the same light source or different light sources using a specially designed optical system and detector array during the process of spectral measurement or analysis.

[0059] A spectrometer calibration module, used to randomly activate electrodes and obtain a calibration matrix of the spectrometer in the activated state;

[0060] An on-chip spectrometer including electrodes for generating detection spectra corresponding to a plurality of spectra to be measured;

[0061] The spectrum optimization and reconstruction module is used to reconstruct multiple spectra to be measured according to the calibration matrix and the detection spectrum to obtain reconstructed spectra.

[0062] The computational reconstruction on-chip spectrometer further includes a spectrum accuracy optimization module, which is used to calculate the accuracy of the reconstructed spectrum and optimize the reconstructed spectrum.

[0063] like Figure 2 As shown, the on-chip spectrometer proposed by the present invention includes a plurality of input ports 1, a single random spectral response generating unit 2 and a plurality of output ports 3 connected in sequence.

[0064] Figure 3 The figure shows a single random spectral response generating unit of a specific embodiment. Its core device is a multi-port arrayed waveguide grating, which consists of multiple input waveguides 4, multiple output waveguides 5, a first flat plate area 6, a second flat plate area 8 and an arrayed waveguide area 7 with electrodes. The waveguide length of the arrayed waveguide area 7 gradually increases with a constant difference ΔL. Each waveguide in the arrayed waveguide area 7 is laid with a length of L. t The thermo-optical tunable part is the electrode. Specifically, the input light of multiple spectra to be measured is input from multiple input waveguides 4, and all the input light is diffracted when passing through the first flat plate area 6. Different input spectra do not interfere with each other during diffraction. The input light after diffraction is then coupled into the array waveguide area for transmission; due to the difference in physical length, the array waveguide area will change the phase distribution of light on each waveguide. In addition, by randomly activating a certain number of heating electrodes, the phase distribution can be further arbitrarily controlled. Then the output light emitted from each output position of the array waveguide will undergo multi-beam interference in the second flat plate area 8, which will result in the formation of a disordered pattern at the output waveguide along the flat plate area, that is, the detection spectrum. Finally, the reconstruction model is used to complete the restoration of multiple spectra to be measured.

[0065] Specific embodiments of the present invention are as follows:

[0066] A silicon nanowire optical waveguide based on silicon-on-insulator material was selected: its core layer was made of silicon (Si) with a thickness of 220 nm, and its shallow etch layer was 150 nm thick. Both the lower and upper cladding layers were made of silicon dioxide (SiO2), each with a thickness of 2 μm. The key parameters of the arrayed waveguide grating in this embodiment are shown in Table 1.

[0067] Table 1 shows the key parameters of arrayed waveguide gratings

[0068]

[0069] In this embodiment, 30 heating states are first determined. 30 heating electrodes are randomly selected for each state and a voltage of 3V is applied. One input port is fixed each time, and 32 output spectra under 30 heating states are recorded. There are 32 input ports in total. This is used as the A matrix of the random module. Then, 4 input ports are randomly selected to input the spectrum to be measured at the same time, and the output power of the 32 output ports is recorded. The spectrum is restored by the calculation reconstruction method to obtain 4 spectra to be measured. Figure 4 As shown, this embodiment can simultaneously restore four spectra to be measured, with a working bandwidth of 100 nm, a resolution of 0.02 nm, and a relative reconstruction error of <0.13.

[0070] The above embodiments are used to illustrate the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for reconstructing multiple input spectra, characterized in that: The following steps are involved: Step 1: Laying a thermo-optical tunable component in the arrayed waveguide area of ​​the on-chip spectrometer. Specifically, a corresponding thermo-optical tunable component is laid on each waveguide in the arrayed waveguide area, so that the phase distribution of light on each waveguide is adjustable. Step 2: Randomly activate the thermo-optical tunable components and then calibrate the calibration matrix of the current on-chip spectrometer; Step 3: While maintaining the same random activation state as in step 2, multiple spectra to be measured are input to the current on-chip spectrometer at the same time, and the on-chip spectrometer outputs the detection spectrum; Step 4: Based on the calibration matrix and detection spectrum of the current on-chip spectrometer, multiple spectra to be measured are optimized and reconstructed to obtain multiple reconstructed spectra; In step 4, an optimization target for the kth spectrum to be measured is constructed, and the kth spectrum to be measured is optimized and iteratively solved according to the constructed optimization target, so that the reconstructed spectrum corresponding to the kth spectrum to be measured is obtained after the optimization target is minimized. The formula of the optimization target for the kth spectrum to be measured is as follows: in, To reconstruct the spectrum, O N×1 To detect the spectrum, D() represents the difference operation, K is the number of spectra to be measured, α 1,k , α 2,k and α 3,k are the first weight - the third weight respectively; A N×M,k is the calibration matrix of the kth spectrum to be measured, S M×1,k is the kth spectrum to be measured, ||||2 is the L2 norm; ||||1 is the L1 norm.

2. The multi-input spectrum reconstruction method according to claim 1, characterized in that: The reconstruction method further comprises the following steps: Step 5: If the accuracy of the current reconstructed spectrum does not meet the target accuracy, repeat steps 2 to 4 to reconstruct the spectrum again until the reconstructed spectrum with the optimal accuracy is obtained.

3. The multi-input spectrum reconstruction method according to claim 1, characterized in that: In step 2, the calibration matrix of the current on-chip spectrometer is calibrated, specifically: A plurality of known spectra are input to the current on-chip spectrometer, and an output spectrum is measured; and a calibration matrix of the current on-chip spectrometer is calculated based on the plurality of known spectra and the corresponding output spectra.

4. The multi-input spectrum reconstruction method according to claim 1, characterized in that: The thermo-optical tunable component includes an electrode.

5. The multi-input spectrum reconstruction method according to claim 2, characterized in that: The accuracy of the reconstructed spectrum is specifically the relative reconstruction error ε r , the calculation formula is as follows: Among them, S M×1,k is the kth spectrum to be measured, is the reconstructed spectrum, and ||||2 is the L2 norm.

6. The multi-input spectrum reconstruction method according to claim 1, characterized in that: The number of input waveguides of the on-chip spectrometer is greater than or equal to the number of spectra to be measured.

7. A computationally reconstructable on-chip spectrometer for multi-spectral parallel detection, characterized in that: including an on-chip spectrometer including a thermo-optical tunable component, a spectrometer calibration module, and a spectrum optimization reconstruction module; A spectrometer calibration module, used for randomly activating the thermo-optical tunable component and obtaining a calibration matrix of the spectrometer in the activated state; An on-chip spectrometer including a thermo-optical tunable component for generating a plurality of detection spectra corresponding to the measured spectra; A spectrum optimization and reconstruction module is used to reconstruct multiple spectra to be measured according to the calibration matrix and the detection spectrum to obtain a reconstructed spectrum; The spectrum optimization and reconstruction module specifically includes: Construct an optimization target for the kth spectrum to be measured, and perform iterative optimization on the kth spectrum to be measured according to the constructed optimization target, so that the reconstructed spectrum corresponding to the kth spectrum to be measured is obtained after minimizing the optimization target. The formula for the optimization target of the kth spectrum to be measured is as follows: in, To reconstruct the spectrum, O N×1 To detect the spectrum, D() represents the difference operation, K is the number of spectra to be measured, α 1,k , α 2,k and α 3,k are the first weight - the third weight respectively; A N×M,k is the calibration matrix of the kth spectrum to be measured, S M×1,k is the kth spectrum to be measured, ||||2 is the L2 norm; ||||1 is the L1 norm.

8. The computationally reconstructable spectrometer on a chip for multi-spectral parallel detection according to claim 7, characterized in that: The computational reconstruction on-chip spectrometer further includes a spectrum accuracy optimization module, which is used to calculate the accuracy of the reconstructed spectrum and optimize the reconstructed spectrum.

Images