Terahertz spectrum information detection method and device
By forming a controlled spatial distribution of conductivity on the pumped material and reconstructing the terahertz spectrum using machine learning algorithms, the problems of complex, large size and high cost in the terahertz spectrum detection system in the prior art are solved, and efficient and accurate spectral detection and real-time measurement capabilities are achieved.
Patent Information
- Application Number
- CN202411742051.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-03-07
AI Technical Summary
The existing terahertz spectral detection technology has problems such as complex system structure, large size, high manufacturing costs and difficulty in meeting real-time measurement requirements.
Through optical pump modulation combined with machine learning algorithms, the controllable conductivity spatial distribution on the pumped material is used as the diffraction grating to generate a diffraction pattern containing spectral information, and the terahertz spectrum is reconstructed from the diffraction pattern using machine learning algorithm.
It realizes efficient and accurate detection of terahertz spectroscopy, avoiding the dependence on mechanical scanning and complex optical systems in traditional technology. The system is simple in structure, small in size and low in cost, and is suitable for real-time measurement needs.
Smart Images
Figure CN119574496A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of spectral detection, and in particular to a method and device for detecting terahertz spectral information. Background Art
[0002] The terahertz band is an important frequency band in the electromagnetic spectrum. Terahertz spectroscopy detection technology has broad application prospects in non-destructive testing, biomedical imaging, safety testing and other fields.
[0003] Currently, mainstream terahertz spectroscopy technologies include terahertz time-domain spectroscopy systems and Fourier transform spectrometers. These traditional technologies suffer from the following drawbacks: complex system structures, requiring numerous optical components and high-precision control systems, resulting in bulky and costly manufacturing. In specific application scenarios, such as nondestructive testing, which requires the ability to change the measurement position in real time, these traditional technologies clearly fail to meet practical requirements, hindering their widespread adoption and application.
[0004] Therefore, a terahertz spectrum detection technology with simple structure, small size, low cost and convenient real-time measurement is needed. Summary of the Invention
[0005] This paper proposes a terahertz spectral information detection solution, which realizes spectral information detection by combining optical pump modulation with machine learning algorithms. It has the characteristics of simple system structure, small size, and easy real-time detection.
[0006] According to one embodiment of the present disclosure, a terahertz spectroscopy information detection method is proposed, comprising:
[0007] Allowing the terahertz light emitted by the terahertz light source to be measured to irradiate the pumped material;
[0008] generating a plurality of pump lights with different optical pump patterns and irradiating the pumped material to form a conductivity spatial distribution corresponding to the optical pump patterns in the pumped material;
[0009] The terahertz light passes through the area of the pumped material illuminated by the pump light, generating a diffraction pattern;
[0010] using a terahertz image sensor to collect a light intensity information sequence of a diffraction pattern corresponding to the optical pump pattern;
[0011] Using a trained machine learning algorithm, the terahertz spectrum emitted by the terahertz light source to be tested is reconstructed based on the light intensity information sequence of multiple diffraction patterns corresponding to multiple optical pump patterns.
[0012] In some possible implementations, generating a plurality of pump lights having different optical pump patterns includes:
[0013] A spatial light modulator is used to generate multiple pump lights with different optical pump patterns. The optical pump pattern includes multiple independently controllable areas, and each area is independently controlled to be in an optical pump state or a non-optical pump state.
[0014] In some possible embodiments, the optical pump pattern includes N×N square areas, where N is an integer greater than or equal to 2, and / or the terahertz image sensor uses an M×M array detector to collect a two-dimensional light intensity information sequence of the diffraction pattern, where M is an integer greater than or equal to 2.
[0015] In some possible implementations, the irradiation area of the pump light on the pumped material completely covers the irradiation area of the terahertz light on the pumped material.
[0016] In some possible implementations, the pumped material is composed of a material with high terahertz transmittance and a conductivity change before and after optical pumping that is greater than a set condition.
[0017] In some possible implementations, using a trained machine learning algorithm to reconstruct a terahertz spectrum emitted by a terahertz light source to be measured based on a sequence of light intensity information of a plurality of diffraction patterns corresponding to a plurality of optical pump patterns includes:
[0018] Arranging a light intensity information sequence of a plurality of diffraction patterns corresponding to a plurality of optical pump patterns into one-dimensional information P;
[0019] The terahertz spectrum emitted by the terahertz light source is reconstructed based on the one-dimensional information P using a trained machine learning algorithm.
[0020] In some possible implementations, the method further includes training the machine learning algorithm by:
[0021] Construct a training set containing L samples, where the samples are numbered k, k = 1, 2, 3, ..., L, where L is an integer greater than 1;
[0022] The transmission spectrum f of each sample in the training set is measured using a terahertz time-domain spectroscopy system. k ;
[0023] A broadband terahertz light source is used to illuminate each sample. According to the emission spectrum S0 and transmission spectrum f k The terahertz spectrum S of each sample was calculated k =S0*f k ;
[0024] Collect the light intensity information sequence P of the diffraction pattern formed by each sample under the plurality of pump lights with different optical pump patterns mk , m is the number of the optical pump pattern;
[0025] Using the light intensity information sequence P mk and terahertz spectroscopy S k The machine learning algorithm is trained.
[0026] According to one embodiment of the present disclosure, a terahertz spectrum information detection device is also proposed, comprising:
[0027] The pumped material is arranged on the transmission path of the terahertz light to be measured;
[0028] A spatial light modulator, configured to generate a plurality of pump lights with different optical pump patterns, wherein the pump lights are irradiated on the pumped material to form a conductivity spatial distribution corresponding to the optical pump patterns;
[0029] A terahertz image sensor for collecting a sequence of light intensity information of a diffraction pattern, the diffraction pattern being generated after the terahertz light passes through the area of the pumped material illuminated by the pump light;
[0030] A data processing module is used to reconstruct the spectrum of the terahertz light emitted by the terahertz light source based on the light intensity information sequence of the multiple diffraction patterns corresponding to the multiple optical pump patterns using a trained machine learning algorithm.
[0031] In some possible implementations, the optical pump pattern generated by the spatial light modulator includes a plurality of independently controllable regions, and each region is independently controlled to be in an optical pumping state or a non-optical pumping state.
[0032] In some possible embodiments, the light pump pattern includes N×N square areas, where N is an integer greater than or equal to 2, and / or the terahertz image sensor uses an M×M array detector to collect a two-dimensional light intensity information sequence of the diffraction pattern, where M is an integer greater than or equal to 2.
[0033] In some possible implementations, the irradiation area of the pumped material by the pump light completely covers the irradiation area of the terahertz light on the pumped material.
[0034] In some possible implementations, the pumped material is composed of a material with high terahertz transmittance and a conductivity change before and after optical pumping that is greater than a set condition.
[0035] The proposed terahertz spectral information detection scheme uses a controllable conductivity spatial distribution formed on the pumped material as a diffraction grating, causing the terahertz light to produce a diffraction pattern containing spectral information. This spectral information is then reconstructed from the diffraction pattern using a machine learning algorithm. This enables efficient and accurate detection in the terahertz band, while avoiding the reliance on mechanical scanning and complex optical systems in traditional terahertz spectral detection technologies. Because the system consists of only a few components, such as a spatial light modulator, a pumped material, and a terahertz image sensor, the overall structure is simple and compact, significantly reducing manufacturing costs while facilitating miniaturization and portability, meeting practical application requirements such as real-time measurement.
[0036] The technical solution proposed in this paper, based on computational spectroscopy detection technology, utilizes machine learning algorithms to efficiently and accurately extract spectral information from diffraction patterns. This not only simplifies hardware requirements and reduces system size, but also significantly improves flexibility and adaptability thanks to the rapid learning and optimization characteristics of artificial intelligence algorithms. By combining artificial intelligence algorithms with advanced optical components such as spatial light modulators, the system provides a more feasible and reliable solution for terahertz applications and is expected to demonstrate excellent performance in a variety of terahertz application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the specification and, together with the description, serve to explain the principles of the specification.
[0038] Figure 1 A flow chart of a terahertz spectroscopy information detection method according to an embodiment of the present disclosure is shown.
[0039] Figure 2 A plurality of optical pump patterns and corresponding diffraction patterns according to an exemplary embodiment of the present disclosure are shown.
[0040] Figure 3 A schematic diagram of the workflow of terahertz spectral information detection according to an exemplary embodiment of the present disclosure is shown.
[0041] Figure 4 An exemplary schematic diagram of terahertz spectrum detection using a terahertz spectrum information detection device according to an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION
[0042] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
[0043] Figure 1 A flowchart of a terahertz spectrum information detection method according to an embodiment of the present disclosure is shown. As shown in the figure, the method includes steps 1 to 5.
[0044] Step 1: irradiate the terahertz light emitted by the terahertz light source to be measured onto the pumped material.
[0045] In some embodiments, the pumped material is composed of a material with high terahertz transmittance and a conductivity change before and after optical pumping that is greater than a set condition. The pumped material selected in this embodiment, on the one hand, has a high terahertz transmittance that ensures the measured terahertz light can effectively pass through the material, maintaining signal strength; on the other hand, its sensitive response to optical pumping ensures that the optical pumping can form a spatial distribution of conductivity on the material surface, allowing the material to function as a modulated diffraction grating. Those skilled in the art can select appropriate terahertz transmittance and set conditions for measuring the rate of conductivity change.
[0046] Furthermore, in some embodiments, the pumped material is made of high-resistance silicon.
[0047] As a preferred embodiment, high-resistivity silicon can be used as the pumped material. High-resistivity silicon not only meets the characteristics of high terahertz transmittance and significant conductivity change before and after optical pumping, but can also be prepared through micro-nano processing technology, facilitating the realization of the required material properties.
[0048] Step 2: Generate a plurality of pump lights with different optical pump patterns and irradiate the pumped material to form a conductivity spatial distribution corresponding to the optical pump patterns in the pumped material.
[0049] Since pump light with different optical pump patterns will form different conductivity distributions on the irradiated pumped material, it is equivalent to forming a diffraction grating corresponding to the optical pump pattern on the pumped material, so that the terahertz light can produce a diffraction pattern under the optical pump pattern after passing through the pumped material.
[0050] In some embodiments, a spatial light modulator (SLM) can be used to generate multiple pump lights with different optical pump patterns. The generated optical pump pattern can include multiple independently controllable regions, each of which can be independently controlled to either an optically pumped state or a non-pumped state. The zoning design proposed in this embodiment facilitates both flexible adjustment of the optical pump pattern and precise control.
[0051] In some embodiments, the optical pump pattern may include N×N square regions, where N is an integer greater than or equal to 2. Each square region can have its optical pump state independently controlled by a spatial light modulator. For example, an optical pump pattern design may include 10×10=100 square regions, each of which can be controlled by a spatial light modulator to have optical pumping or not.
[0052] According to this embodiment, the optical pump pattern is an N×N square area. Its regular structure makes it easy to control, and its clear boundaries make it easier to determine the optical pumping state. Furthermore, the square areas can be seamlessly connected to avoid gaps. This embodiment makes the optical pump modulation effect more controllable and stable.
[0053] Step 3: The terahertz light passes through the area of the pumped material irradiated by the pump light to generate a diffraction pattern.
[0054] As mentioned above, the modulated pump light irradiates the pumped material, forming a diffraction grating corresponding to the optical pumping pattern on the pumped material. The terahertz light passes through the diffraction grating to generate a corresponding diffraction pattern.
[0055] In some embodiments, the irradiation area of the pump light on the pumped material completely covers the irradiation area of the terahertz light on the pumped material. According to this embodiment, it is possible to ensure that the diffraction pattern contains all spectral information of the terahertz light to be measured, avoiding the loss of spectral information.
[0056] Step 4: using a terahertz image sensor to collect a light intensity information sequence of the diffraction pattern corresponding to the optical pump pattern.
[0057] Each time the optical pump pattern changes, the image sensor outputs a corresponding sequence of light intensity signals, which serves as input data for subsequent spectral reconstruction. Terahertz image sensors use array detectors to capture the intensity information sequence of the diffraction pattern, thereby capturing the intensity distribution of the diffraction pattern at different spatial locations and acquiring complete diffraction pattern information.
[0058] In some embodiments, a terahertz image sensor may use an M×M array detector to collect the intensity information sequence of the diffraction pattern, where M is an integer greater than or equal to 2. For example, a 4×4 array detector may be used. In this case, for each optical pump pattern, terahertz light passing through the diffraction grating produces a corresponding diffraction pattern, and the array detector collects the intensity values of each pixel in the diffraction pattern. These intensity values form a two-dimensionally distributed intensity information sequence, reflecting the characteristics of the original terahertz spectrum.
[0059] Figure 2 A plurality of optical pump patterns and corresponding diffraction patterns according to an exemplary embodiment of the present disclosure are shown. Figure 2 In the example shown, the optical pump pattern includes 100 square areas of 10X10, and the light intensity information sequence of the diffraction pattern is collected by a 4X4 array detector.
[0060] Step 5: Using a terahertz image sensor to collect a light intensity information sequence of the diffraction pattern corresponding to the optical pump pattern.
[0061] In principle, there is a physical relationship between the intensity information sequence P of the diffraction pattern and the spectrum S, P = Φ*S, where Φ is the transmission characteristic matrix of the optical pump pattern for different wavelengths. Each row of Φ, φ j T =[φ j1 ,φ j2 …,φ jr ] represents the transmission characteristics of different optical pump patterns for the rth wavelength (j=1, 2, ..., nm). Since the product of the total number of pixels n of the terahertz image sensor and the number of optical pump patterns m (n*m) is usually less than the number of spectral sampling points r, the equation is an underdetermined system of equations. This means that it is difficult to directly solve the spectral information from only a small amount of diffraction pattern intensity information. This embodiment uses a machine learning algorithm to solve this problem. Through the trained machine learning algorithm, the intensity information of multiple diffraction patterns corresponding to multiple optical pump patterns is mapped to the terahertz spectrum transmitted by the terahertz light source to be measured. The terahertz spectrum transmitted by the terahertz light source to be measured can be expressed as S(λ), and the reconstructed terahertz spectrum can be expressed as S'(λ).
[0062] In some embodiments, the light intensity information sequence of multiple diffraction patterns can be arranged into one-dimensional information P; and the spectrum S'(λ) emitted by the terahertz light source can be reconstructed based on the one-dimensional information P using a trained machine learning algorithm.
[0063] For example, if the light intensity information sequence of each diffraction pattern is a two-dimensional information sequence, the light intensity information sequences of all diffraction patterns can be arranged into a one-dimensional information sequence P = [P 11 , P 12 ,…,P 1m , P 21,…,P nm ] T , where n is the number of pixels in the light intensity information sequence corresponding to each diffraction pattern, and m is the number of optical pump patterns, which is convenient for machine learning algorithm processing.
[0064] In some embodiments, the machine learning algorithm may be trained by:
[0065] Construct a training set containing L samples, where the samples are numbered k, k = 1, 2, 3, ..., L, where L is an integer greater than 1;
[0066] The transmission spectrum f of each sample in the training set is measured using a terahertz time-domain spectroscopy system. k ;
[0067] A broadband terahertz light source is used to illuminate each sample. According to the emission spectrum S0 and transmission spectrum f k The terahertz spectrum S of each sample was calculated k =S0*f k ;
[0068] Collect the light intensity information sequence P of the diffraction pattern formed by each sample under the plurality of pump lights with different optical pump patterns mk , m is the number of the optical pump pattern;
[0069] Using the light intensity information sequence P mk and terahertz spectroscopy S k The machine learning algorithm is trained.
[0070] A large number of training samples (e.g. L = 1000) can be constructed to obtain enough spectrum-diffraction pattern data pairs (S k , P mk ), enabling the machine learning algorithm to learn the mapping relationship between the diffraction pattern intensity information sequence and the spectrum. According to this embodiment, there is no need to explicitly solve a complex set of underdetermined equations. Instead, the machine learning algorithm automatically learns the corresponding relationship between the two from a large amount of training data. By providing sufficient training data, the accuracy and robustness of the reconstruction results can be further improved. The same optical pump pattern with the same appearance order is used in the training and inference processes.
[0071] Using a trained machine learning algorithm, the original terahertz spectrum can be reconstructed from the light intensity information sequence of the diffraction pattern, achieving efficient and accurate spectral detection.
[0072] Figure 3 FIG. 1 shows a schematic diagram of a workflow for detecting terahertz spectral information according to an exemplary embodiment of the present disclosure. Figure 3As shown in the figure, the workflow mainly includes three parts: generating optical pump patterns, algorithm learning process and spectrum reconstruction process.
[0073] The first step is to generate an optical pump pattern. A spatial light modulator can be used to generate 100 optical pump patterns, numbered m (m = 1, 2, 3...100). Each pattern consists of 100 regions, each of which can be independently controlled to have or not have optical pumping. Since the optical pump patterns are different from each other, the terahertz diffraction patterns after optical pumping are also different.
[0074] Next comes the algorithm learning process. This process first obtains 1,000 designed training samples, numbered k (k = 1, 2, 3, ... 1,000). Samples with known typical spectral characteristics in the terahertz band, such as lactose and nicotinic acid, can be used. Alternatively, terahertz metasurface devices with different spectral transmission characteristics can be fabricated using micro-nanofabrication techniques.
[0075] The transmission spectra of these samples were then measured using a terahertz time-domain spectroscopy system, denoted as f k .
[0076] Then, a broadband terahertz light source (with an emission spectrum of S0) was used to illuminate each sample. k The terahertz spectrum S of each sample was calculated k =S0*f k .
[0077] The terahertz light after passing through each sample is irradiated onto the pumped material. The optical pump pattern on the pumped material completely covers the irradiation area of the terahertz light. The terahertz light after passing through the pumped material enters the image sensor and records its diffraction pattern information P mk , where k corresponds to the kth sample, and m represents the spectrum modulated by the mth optical pump pattern.
[0078] Therefore, all spectral signals in the training set (S k ) and diffraction pattern information (P mk ) to train the spectral reconstruction algorithm.
[0079] Finally, the spectral reconstruction process involves measuring the terahertz spectrum under test. The same optical pump modulation as used in the training process is used during this measurement. The light intensity information sequences collected by the terahertz image sensor at different times correspond to different optical pump patterns on the pumped material. The terahertz spectrum S'(λ) is reconstructed using the trained spectral reconstruction algorithm.
[0080] The terahertz spectral information detection method proposed in this embodiment forms a controllable electrical conductivity spatial distribution as a diffraction grating on the pumped material, causing the terahertz light to produce a diffraction pattern containing spectral information. The spectral information is then reconstructed from the diffraction pattern using a machine learning algorithm, thereby achieving efficient and accurate detection in the terahertz band. It avoids the reliance on mechanical scanning and complex optical systems in traditional terahertz spectral detection technology, making the overall structure simple and compact, which not only significantly reduces manufacturing costs but also facilitates miniaturization and portable design, and can meet practical application needs such as real-time measurement.
[0081] The present disclosure also proposes a terahertz spectrum information detection device, comprising:
[0082] The pumped material is arranged on the transmission path of the terahertz light to be measured;
[0083] A spatial light modulator, configured to generate a plurality of pump lights with different optical pump patterns, wherein the pump lights are irradiated on the pumped material to form a conductivity spatial distribution corresponding to the optical pump patterns;
[0084] A terahertz image sensor for collecting a sequence of light intensity information of a diffraction pattern, the diffraction pattern being generated after the terahertz light passes through the area of the pumped material illuminated by the pump light;
[0085] A data processing module is used to reconstruct the spectrum of the terahertz light emitted by the terahertz light source based on the light intensity information sequence of the multiple diffraction patterns corresponding to the multiple optical pump patterns using a trained machine learning algorithm.
[0086] Figure 4 An exemplary schematic diagram of terahertz spectrum detection using a terahertz spectrum information detection device according to an exemplary embodiment of the present disclosure is shown, comprising a pumped material (8), a spatial light modulator (6), a terahertz image sensor (9) and a data processing module (10). Figure 4 Also shown are: a terahertz light source to be measured (1), terahertz light to be measured (2), a lens (3), a visible light source (4), a lens (5) and a pump light (7).
[0087] The terahertz light source to be measured (1) refers to the terahertz light source concerned in the present disclosure, that is, the sample to be detected, which is used to provide the terahertz light to be measured (also called terahertz wave) (2).
[0088] The spectrum of the terahertz light (2) emitted from the terahertz light source (1) to be measured is the detection target of the present disclosure.
[0089] The lens (3) is placed behind the terahertz light source (1) to be measured in the optical path and is used to expand the terahertz light so as to increase the irradiation area of the terahertz light.
[0090] The visible light source (4) is placed on the side of the optical path and is used to provide pump light. The lens (5) is placed behind the visible light source (4) and is used to expand the pump light and increase the coverage area of the pump light.
[0091] The pumped material (8) is placed behind the lens (3) in the optical path and is used to drive diffraction gratings corresponding to different optical pump patterns. The pumped material (8) can be made of a material with high terahertz transmittance and a conductivity that changes significantly before and after optical pumping, such as high-resistance silicon, and can be prepared through micro-nano processing technology according to the requirements of use.
[0092] The spatial light modulator (6) is located behind the lens (5) and generates 10 different light pumping patterns in a certain order to irradiate the pumped material (8) to produce a light pumping effect. The light pumping patterns are numbered m (m = 1, 2, 3, ..., 10). Each light pumping pattern consists of 10×10=100 square areas, where each square area can be controlled by the spatial light modulator to have light pumping or no light pumping. The light pumping patterns are different, and the diffraction patterns generated after the terahertz light passes through are also different. See Figure 2 Examples of optical pump patterns and their corresponding terahertz wave diffraction patterns are shown.
[0093] Pump light (7) is generated from a spatial light modulator (6) and irradiated onto a pumped material (8). The pattern of the pump light (7) can be adjusted by setting parameters of the spatial light modulator (6). The irradiation area of the pumped material (8) by the pump light completely covers the irradiation area of the pumped material (8) by the terahertz light.
[0094] The terahertz image sensor (9) is located behind the pumped material (8) in the optical path and is used to detect the light intensity information of the diffraction pattern of the terahertz light. In this exemplary embodiment, the terahertz image sensor (9) includes an array detector arranged in a 4*4 arrangement, and the pixel points in the light intensity information sequence collected each time are numbered n (n=1, 2, 3, ..., 16). There are a total of 10 groups of light pump patterns. Each time the spatial light modulator (6) switches the pattern, the image sensor (9) outputs a two-dimensional light intensity information sequence, which can be recorded as P m (m=1, 2, 3,…, 10).
[0095] The data processing module (10) is connected to the terahertz image sensor (9), reads the light intensity information sequence output by the terahertz image sensor (9), and reconstructs the terahertz spectrum through a pre-trained machine learning algorithm. The reconstructed terahertz spectrum is output through the data processing module (10) as the final terahertz spectrum detection result.
[0096] For other details and beneficial effects of this embodiment, please refer to the relevant introduction above and will not be repeated here.
[0097] The various embodiments in this specification are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other. Each embodiment focuses on the differences between the other embodiments. In particular, the data processing device embodiment is generally similar to the method embodiment, so its description is relatively simple. For relevant portions, refer to the description of the method embodiment.
[0098] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0099] Although this specification includes many specific implementation details, these should not be interpreted as limiting the scope of any invention or the scope of protection claimed, but are mainly used to describe the features of specific embodiments of specific inventions. Certain features described in multiple embodiments within this specification may also be implemented in combination in a single embodiment. On the other hand, the various features described in a single embodiment may also be implemented separately in multiple embodiments or in any suitable sub-combination. In addition, although features may work in certain combinations as described above and even initially claimed as such, one or more features from the claimed combination may be removed from the combination in some cases, and the claimed combination may point to a sub-combination or a variation of the sub-combination.
[0100] Similarly, although operations are depicted in a particular order in the accompanying drawings, this should not be understood as requiring that these operations be performed in the particular order shown or performed sequentially, or that all illustrated operations be performed to achieve the desired results. In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various system modules and components in the above-described embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product, or packaged into multiple software products.
[0101] Thus, specific embodiments of the subject matter have been described. Other embodiments are within the scope of the following claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the particular order shown or sequential sequence to achieve the desired results. In some implementations, multitasking and parallel processing may be advantageous.
[0102] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of this specification shall be included in the scope of protection of one or more embodiments of this specification.
Claims
1. A terahertz spectroscopy information detection method, characterized in that: include: Allowing the terahertz light emitted by the terahertz light source to be measured to irradiate the pumped material; Generate a plurality of pump lights with different optical pump patterns and irradiate the pumped material to form a conductivity spatial distribution corresponding to the optical pump patterns in the pumped material; The terahertz light passes through the area of the pumped material irradiated by the pump light, generating a diffraction pattern; Using a terahertz image sensor to collect a light intensity information sequence of a diffraction pattern corresponding to the optical pump pattern; The terahertz spectrum emitted by the terahertz light source to be tested is reconstructed based on the light intensity information sequence of multiple diffraction patterns corresponding to multiple optical pump patterns using a trained machine learning algorithm.
2. The method according to claim 1, characterized in that Generating a plurality of pump lights with different optical pump patterns includes: A spatial light modulator is used to generate a plurality of pump lights with different optical pump patterns, wherein the optical pump pattern includes a plurality of independently controllable regions, and each region is independently controlled to be in an optical pump state or a non-optical pump state.
3. The method according to claim 2, characterized in that The optical pump pattern includes N×N square areas, where N is an integer greater than or equal to 2, and / or the terahertz image sensor uses an M×M array detector to collect a two-dimensional light intensity information sequence of the diffraction pattern, where M is an integer greater than or equal to 2.
4. The method according to claim 1, characterized in that The irradiation area of the pump light on the pumped material completely covers the irradiation area of the terahertz light on the pumped material.
5. The method according to claim 1, characterized in that The pumped material is composed of a material with high terahertz light transmittance and a conductivity change before and after optical pumping that is greater than a set condition.
6. The method according to claim 1, characterized in that Using a trained machine learning algorithm, a terahertz spectrum emitted by a terahertz light source to be tested is reconstructed based on a light intensity information sequence of a plurality of diffraction patterns corresponding to a plurality of optical pump patterns, including: Arranging light intensity information sequences of a plurality of diffraction patterns corresponding to a plurality of optical pump patterns into one-dimensional information P; The terahertz spectrum emitted by the terahertz light source is reconstructed based on the one-dimensional information P using a trained machine learning algorithm.
7. The method according to claim 1, characterized in that The method also includes training the machine learning algorithm by the following steps: Construct a training set containing L samples, where the sample number is k, k = 1, 2, 3...L, where L is an integer greater than 1; The transmission spectrum f of each sample in the training set is measured using a terahertz time-domain spectroscopy system. k ; A broadband terahertz light source was used to illuminate each sample. According to the emission spectrum S0 and transmission spectrum f k The terahertz spectrum S of each sample was calculated. k =S0*f k ; Collect the light intensity information sequence P of the diffraction pattern formed by each sample under the plurality of pump lights with different optical pump patterns mk , m is the number of the optical pump pattern; Using the light intensity information sequence P mk and terahertz spectroscopy S k The machine learning algorithm is trained.
8. A terahertz spectrum information detection device, characterized in that: include: The pumped material is arranged on the transmission path of the terahertz light to be measured; A spatial light modulator, used to generate a plurality of pump lights with different optical pump patterns, wherein the pump lights are irradiated on the pumped material to form a conductivity spatial distribution corresponding to the optical pump patterns; A terahertz image sensor for collecting a sequence of light intensity information of a diffraction pattern, wherein the diffraction pattern is generated after the terahertz light passes through the area of the pumped material irradiated by the pumping light; The data processing module is used to reconstruct the spectrum of the terahertz light emitted by the terahertz light source based on the light intensity information sequence of multiple diffraction patterns corresponding to multiple optical pump patterns using a trained machine learning algorithm.
9. The device according to claim 8, characterized in that The optical pump pattern generated by the spatial light modulator includes a plurality of independently controllable regions, each region being independently controlled to be in an optical pump state or a non-optical pump state.
10. The device according to claim 9, characterized in that The optical pump pattern includes N×N square areas, where N is an integer greater than or equal to 2, and / or the terahertz image sensor uses an M×M array detector to collect a two-dimensional light intensity information sequence of the diffraction pattern, where M is an integer greater than or equal to 2.
11. The device according to claim 8, characterized in that The irradiation area of the pump light on the pumped material completely covers the irradiation area of the terahertz light on the pumped material.
12. The device according to claim 8, characterized in that The pumped material is composed of a material with high terahertz light transmittance and a conductivity change before and after optical pumping that is greater than a set condition.