Multivariate terahertz information detection method and device

Abstract

A multi-dimensional terahertz information detection method and device are disclosed. The method uses a terahertz detector at a predetermined distance from a terahertz light source to obtain two sets of spatial intensity distribution information: one set is the spatial intensity distribution of the terahertz light to be detected directly collected, and the other set is the spatial intensity distribution of the terahertz light after being regulated by a microstructure array with different phase regulation characteristics. Through a pre-trained neural network, the phase information, spectral information and polarization information of the terahertz light to be detected are reconstructed according to the two sets of spatial intensity distribution information. Through phase regulation of the microstructure array and information reconstruction of the neural network, the disclosure realizes simultaneous detection of multi-dimensional information, avoids complex optical system design; adopts double neural networks to reconstruct different physical mechanisms respectively, improves the detection accuracy; has a fast detection capability of milliseconds, meets the actual application requirements; the system structure is simple, easy to implement, and has good expansibility.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of spectral detection, and in particular to a multi-element terahertz information detection method and device. BACKGROUND

[0002] Under the background of rapid development of science and technology today, the research of terahertz wave band has become the frontier in many fields such as wireless communication, material detection, biomedical and security inspection. Terahertz wave refers to electromagnetic wave with frequency between 0.1-10 THz, between microwave and infrared. Due to its unique ability to penetrate non-conductive materials, low radiation and high sensitivity to chemical substance fingerprint characteristics, it shows great application potential.

[0003] At present, important progress has been made in the application research of terahertz wave. However, there are still many challenges and limitations in the detection and analysis technology of terahertz wave. These challenges include low data acquisition and processing efficiency, insufficient acquisition of multi-element information such as space, phase, spectrum and polarization in complex scenes, which greatly limits the development of terahertz technology in practical applications.

[0004] With the deepening of the research on terahertz detection technology, the demand for terahertz detection devices capable of realizing high-performance and multi-element information detection is increasing. Therefore, how to realize efficient detection of multi-element information such as phase, spectrum and polarization of terahertz wave has become a technical problem to be solved. SUMMARY

[0005] The present disclosure proposes a multi-element terahertz information detection scheme, which can simultaneously acquire multi-dimensional information such as phase, spectrum and polarization of the terahertz light to be detected by setting microstructure arrays with different phase control characteristics in the terahertz detection light path, combined with deep learning prediction, realizing fast and efficient multi-element information detection.

[0006] According to one embodiment of the present disclosure, a multi-element terahertz information detection method is proposed, comprising the following steps:

[0007] providing terahertz light to be detected by a terahertz light source;

[0008] acquiring first spatial intensity distribution information of the terahertz light to be detected by a terahertz detector at a predetermined distance from the terahertz light source;

[0009] setting a control device between the terahertz light source and the terahertz detector, the control device comprising a plurality of microstructure arrays, different microstructure arrays being designed to have different phase control characteristics, and acquiring second spatial intensity distribution information of the terahertz light to be detected after phase control by the terahertz detector;

[0010] The pre-trained neural network is used to reconstruct phase information, spectral information and polarization information of the to-be-detected terahertz light according to the first spatial intensity distribution information and the second spatial intensity distribution information.

[0011] In some possible implementation manners, the plurality of microstructure arrays completely cover an irradiation area of the to-be-detected terahertz light on the regulation device.

[0012] In some possible implementation manners, the regulation device includes MXM microstructure arrays, and the terahertz detector has corresponding MXM detection regions, which one-to-one correspond to the microstructure arrays.

[0013] In some possible implementation manners, the method further includes controlling a position of the regulation device by using a switching device, the switching device including a first gear position and a second gear position, and wherein:

[0014] When the switching device is set to the first gear position, the regulation device is controlled to cut out an optical path of the to-be-detected terahertz light;

[0015] When the switching device is set to the second gear position, the regulation device is controlled to cut into the optical path of the to-be-detected terahertz light between the terahertz light source and the terahertz detector, so that the regulation device performs phase regulation on the to-be-detected terahertz light.

[0016] In some possible implementation manners, the pre-trained neural network is used to reconstruct phase information, spectral information and polarization information of the to-be-detected terahertz light according to the first spatial intensity distribution information and the second spatial intensity distribution information, and the reconstruction includes

[0017] The pre-trained phase reconstruction network is used to reconstruct phase information of the to-be-detected terahertz light according to the first spatial intensity distribution information.

[0018] The pre-trained spectral polarization reconstruction network is used to reconstruct spectral information and polarization information of the to-be-detected terahertz light according to the second spatial intensity distribution information.

[0019] In some possible implementation manners, the training process of the phase reconstruction network includes:

[0020] A training set including a plurality of samples is obtained;

[0021] An initial phase distribution information of terahertz light provided by each sample is measured by using a terahertz time-domain spectroscopy system, and spatial intensity distribution information at a predetermined distance from the sample is obtained by using the terahertz detector;

[0022] The phase reconstruction network is trained based on the measured initial phase distribution information of each sample and the spatial intensity distribution information at the predetermined distance from the sample.

[0023] In some possible embodiments, the training process of the spectral polarization reconstruction network comprises:

[0024] obtaining a training set containing a plurality of samples, and knowing polarization information P xn and polarization information P yn of each sample in the X direction and the Y direction, n being the sample number;

[0025] measuring the transmission spectrum f n of each sample by using a terahertz time-domain spectroscopy system, and irradiating the sample with a broadband terahertz light source with a spectrum S0, to obtain the spectrum S n of the terahertz light provided by the sample according to S n = S0*f n ;

[0026] setting the regulating device between the sample and the terahertz detector, and obtaining the spatial intensity distribution information I mn of the terahertz light after phase regulation by using the terahertz detector, wherein m is the number of the microstructure array, and n is the sample number;

[0027] training the spectral polarization reconstruction network based on the obtained spatial intensity distribution information I mn , and the polarization information P xn , P yn and the spectrum S n of the corresponding sample.

[0028] In some possible embodiments, the substrate of each microstructure array is composed of a material with high terahertz light transmittance, and the elements arranged on the substrate are metal structures.

[0029] According to one embodiment of the present disclosure, a multi-element terahertz information detection device is provided, comprising a terahertz detector, a regulating device, a switching device and a data processing module, wherein:

[0030] the terahertz detector is arranged at a predetermined distance from a terahertz light source, and the terahertz light source is used to provide terahertz light to be detected;

[0031] the regulating device comprises a plurality of microstructure arrays, and different microstructure arrays are designed to have different phase regulation characteristics;

[0032] the switching device comprises a first gear position and a second gear position, when the switching device is set to the first gear position, the regulating device is controlled to cut out the light path of the terahertz light to be detected, and when the switching device is set to the second gear position, the regulating device is controlled to cut into the light path of the terahertz light to be detected between the terahertz light source and the terahertz detector, so as to perform phase regulation on the terahertz light to be detected.

[0033] The data processing module is configured to utilize a pre-trained neural network to reconstruct phase information, spectral information and polarization information of the to-be-detected terahertz light according to spatial intensity distribution information collected by the terahertz detector when the regulating device cuts in and cuts out the light path.

[0034] In some possible implementation manners, the plurality of microstructure arrays completely cover an irradiation area of the to-be-detected terahertz light on the regulating device.

[0035] In some possible implementation manners, the regulating device includes MxM microstructure arrays, and the terahertz detector has corresponding MxM detection areas, which one-to-one correspond to the microstructure arrays.

[0036] In some possible implementation manners, the data processing module includes:

[0037] a phase reconstruction network configured to reconstruct phase information of the to-be-detected terahertz light according to spatial intensity distribution information collected by the terahertz detector when the regulating device cuts out the light path;

[0038] a spectral polarization reconstruction network configured to reconstruct spectral information and polarization information of the to-be-detected terahertz light according to spatial intensity distribution information collected by the terahertz detector when the regulating device cuts in the light path.

[0039] In some possible implementation manners, a substrate of the microstructure array is composed of a material with high terahertz light transmittance, and an element on the substrate is a metal structure.

[0040] The multi-dimensional terahertz information detection technology provided in the scheme can fully utilize spatial intensity distribution information obtained by a terahertz detector, and effectively reconstruct multi-dimensional information of to-be-detected terahertz light through a pre-trained neural network. The phase information can be obtained from the spatial intensity distribution information by using the phase reconstruction network, which makes up for the deficiency that the traditional method is difficult to simultaneously measure the phase and the intensity. The spectral information and the polarization information can be analyzed from the spatial intensity distribution by using the spectral polarization reconstruction network in combination with the regulating effect of the microstructure array on the terahertz light. This information reconstruction method based on the neural network avoids complex optical system design and significantly simplifies the measurement device. Meanwhile, since the neural network has the characteristics of fast calculation, the scheme can realize millisecond-level fast detection, which meets the requirement on detection efficiency in actual application. In addition, the scheme adopts an open architecture design, supports the introduction of new microstructure design and more advanced deep learning model in the future, and has good expansibility. BRIEF DESCRIPTION OF DRAWINGS

[0041] The accompanying drawings, which are incorporated herein and form a part of the specification, illustrate embodiments consistent with the present description and serve to explain the principles

[0042] Figure 1 A flow chart of a method for terahertz spectral information detection is shown according to an embodiment of the present disclosure.

[0043] Figure 2 A number of microstructure arrays and their corresponding X-direction and Y-direction polarization transmission spectra are shown according to an exemplary embodiment of the present disclosure.

[0044] Figure 3 An exemplary schematic diagram of a multi-element terahertz information detection device is shown according to an exemplary embodiment of the present disclosure.

[0045] Figure 4 A workflow schematic diagram of multi-element terahertz information detection is shown according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0046] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The description of the exemplary embodiments is intended to apply to all alternative embodiments, as would be understood by persons skilled in the art. To the extent that they do not particularize to the embodiments as described in the attached claims, they are intended to be illustrative of devices and methods consistent with some aspects of the present disclosure.

[0047] Figure 1 A flow chart of a method for multi-element terahertz information detection is shown according to an embodiment of the present disclosure. As shown, the method includes steps 1-4.

[0048] Step 1, providing a terahertz light to be measured by a terahertz light source.

[0049] Here, the terahertz light source refers to the source of the terahertz light that needs to be detected. For example, in a material detection application, the terahertz light source can be the sample to be measured.

[0050] Step 2, collecting first spatial intensity distribution information of the terahertz light to be measured by a terahertz detector at a predetermined distance from the terahertz light source.

[0051] For example, the terahertz detector can include a face array detector, which can be regarded as an image sensor, and the detection range can include a plurality of detection regions, such as 6X6 detection regions. The terahertz detector can obtain intensity information of each detection region on the face array, i.e., the first spatial intensity distribution information.

[0052] Step 3, a regulating device is arranged between the terahertz light source and the terahertz detector, the regulating device comprises a plurality of microstructure arrays, different microstructure arrays are designed to have different phase regulating characteristics, and the second spatial intensity distribution information of the phase-regulated terahertz light to be measured is collected by using the terahertz detector.

[0053] In some embodiments, the plurality of microstructure arrays completely cover the illumination area of the terahertz light to be measured on the regulating device.

[0054] According to the present embodiment, all spatial information of the terahertz light to be measured can be regulated and detected, and information loss can be avoided.

[0055] In some embodiments, the regulating device comprises MxM microstructure arrays, and the terahertz detector has corresponding MxM detection regions, which correspond to the microstructure arrays one by one.

[0056] According to the present embodiment, the MxM microstructure arrays of the regulating device correspond to the MxM detection regions of the detector one by one, each microstructure array corresponds to a detection region, and the regulation effect of the array on the terahertz light can be accurately recorded. The design according to the present embodiment ensures that the spatial intensity distribution information after regulation can completely reflect all information of the phase-regulated terahertz light, and provides a reliable data basis for subsequent reconstruction of the spectrum, polarization and other information of the terahertz light to be measured. In one example, M=6.

[0057] In some embodiments, the substrate of each microstructure array is composed of a material with high terahertz light transmittance, and the elements arranged on the substrate are metal structures.

[0058] When the electric field direction of the incident terahertz light is parallel to the long axis of the metal element, strong coupling occurs, and when the electric field direction is perpendicular to the long axis, weak coupling occurs. By designing the direction arrangement of multiple elements on the substrate, different responses to x and y polarization components can be achieved, thereby obtaining specific polarization transmission characteristics and realizing phase regulation. The substrate with high terahertz light transmittance enables the terahertz light to be measured to effectively pass through the material, ensuring signal strength. For example, a high-resistance silicon substrate can be used. The microstructure array can be prepared by a micro-nano processing technology.

[0059] Figure 2 Several microstructure arrays and their corresponding X-direction and Y-direction polarization transmission spectra according to an example embodiment of the present disclosure are shown. As shown in Figure 2 each microstructure array is arranged with 25 metal elements, and each metal element is arranged in a specific direction.

[0060] In some embodiments, the position of the regulating device can be controlled by a switching device, which comprises a first gear position and a second gear position, wherein:

[0061] When the switching device is set at the first gear position, the regulating device is controlled to cut out the optical path of the terahertz light to be measured;

[0062] When the switching device is set at the second gear position, the regulating device is controlled to cut into the optical path of the terahertz light to be measured between the terahertz light source and the terahertz detector, so that the regulating device performs phase regulation on the terahertz light to be measured.

[0063] According to the present embodiment, the system structure is simplified, and two measurement modes can be flexibly implemented. The first gear position makes the regulating device cut out the optical path, so that the first spatial intensity distribution information of the terahertz light to be measured transmitted to the terahertz detector without regulation can be obtained; the second gear position cuts into the phase regulation of the terahertz light by the regulating device, so that the second spatial intensity distribution information of the terahertz light to be measured transmitted to the terahertz detector after phase regulation can be obtained.

[0064] Step 4, using a pre-trained neural network, reconstructing the phase information, spectral information and polarization information of the terahertz light to be measured according to the first spatial intensity distribution information and the second spatial intensity distribution information.

[0065] In some embodiments, using a pre-trained phase reconstruction network, the phase information of the terahertz light to be measured is reconstructed according to the first spatial intensity distribution information;

[0066] Using a pre-trained spectral polarization reconstruction network, the spectral information and polarization information of the terahertz light to be measured are reconstructed according to the second spatial intensity distribution information.

[0067] In some embodiments, the training process of the phase reconstruction network comprises:

[0068] Obtaining a training set containing a plurality of samples;

[0069] Using a terahertz time-domain spectroscopy system to measure the initial phase distribution information of the terahertz light provided by each sample, and using the terahertz detector to obtain the spatial intensity distribution information at a predetermined distance from the sample;

[0070] Based on the measured initial phase distribution information and the spatial intensity distribution information at a predetermined distance from each sample, the phase reconstruction network is trained.

[0071] In some embodiments, the training process of the spectral polarization reconstruction network comprises:

[0072] Obtain a training set containing a plurality of samples, and the polarization information P of each sample in the X direction is known xn and the polarization information P in the Y direction is known yn , n is the sample number;

[0073] Measure the transmission spectrum f of each sample using a terahertz time-domain spectroscopy system n , and use a broadband terahertz light source with a spectrum S0 to illuminate the sample, and obtain the spectrum S provided by the sample according to S n = S0*f n n ;

[0074] Place the regulating device between the sample and the terahertz detector, and obtain the spatial intensity distribution information I of the terahertz light after phase regulation using the terahertz detector mn , where m is the number of the microstructure array, and n is the sample number;

[0075] Based on the obtained spatial intensity distribution information I mn and the polarization information P xn , P yn and the spectrum S n of the corresponding sample, train the spectral polarization reconstruction network.

[0076] According to the above embodiment, two independently trained neural networks are used to realize the reconstruction of different information. The phase reconstruction network reconstructs the initial phase through the spatial intensity distribution of the terahertz light, and the training data set contains the spatial intensity distribution information of a plurality of samples at a predetermined distance and the corresponding initial phase distribution information. The spectral polarization reconstruction network uses the regulated spatial intensity distribution to reconstruct the spectrum and polarization information, and the training data includes samples with known polarization information P xn and P yn , the transmission spectrum f n is measured, the spectrum S n is calculated, and the spatial intensity distribution information I mn is obtained in combination with the regulating device.

[0077] The phase reconstruction is based on the propagation characteristics of terahertz waves, and the spectral polarization reconstruction is based on the regulating effect of the microstructure array. The design of such a double neural network is conducive to realizing more accurate information detection and efficiently reconstructing multi-dimensional information. After the training of the two neural networks is completed, they can be directly used for multi-element terahertz information reconstruction in actual measurement, realizing millisecond-level fast detection.

[0078] Figure 3 ​An exemplary schematic diagram of a multi-element terahertz information detection device according to an exemplary embodiment of the present disclosure is shown. As shown in the diagram, the device comprises a modulation device (3), a switching device (4), a terahertz detector (5) and a data processing module (6). The diagram also shows the terahertz light to be detected (2) and a terahertz light source (1) for providing the terahertz light to be detected (2).

[0079] The terahertz light source (1) refers to a terahertz light source of interest in the present disclosure, which can refer to a light source that emits light by itself, or a sample to be detected through which terahertz light passes. The terahertz light source (1) is used to provide the terahertz light to be detected (2).

[0080] The terahertz light to be detected (2) refers to the terahertz light emitted from the terahertz light source (1) and is the detection target of the present disclosure.

[0081] The modulation device (3) is placed behind the terahertz light source (1) in the light path and modulates the terahertz light passing through. Figure 3 The modulation device (3) comprises 6X6 microstructure arrays, numbered m (m = 1, 2, 3, …, 36). The substrate of the array can be made of a high-transmittance material in the terahertz band, such as high-resistance silicon, etc., and a plurality of metal elements with specific directions are arranged on the substrate, Figure 2 An exemplary schematic diagram of a part of the artificial microstructure array is shown. The microstructure array can be prepared by a micro-nano processing technology. Through design, different microstructure arrays have different modulation effects on the x and y polarization directions of the terahertz light. In the present embodiment, the modulation device (3) is composed of 6X6 = 36 different artificial microstructure arrays.

[0082] The switching device (4) is connected to the modulation device (3). The switching device (4) has two different gears, a first gear and a second gear, and can switch the modulation device (3) in or out of the light path of the terahertz light to be detected (2) between the terahertz light source (1) and the terahertz detector (5) according to different information acquisition needs. When the phase information of the terahertz light (2) needs to be acquired, the switching device (4) can be switched to the first gear, i.e. the neutral gear, so that the terahertz light is directly irradiated to the terahertz detector (5) behind; when the spectral and polarization information of the terahertz light needs to be acquired, the switching device (4) can be switched to the second gear, so that the modulation device (3) is switched into the light path for modulation.

[0083] The terahertz detector (5) is at a distance d from the terahertz light source (1) and is used to detect the spatial intensity distribution information of the terahertz light. Figure 3 The terahertz detector (5) shown is an array detector, and its detection surface can be divided into 6X6 = 36 regions in space, each of which can independently output an intensity signal, denoted as I m(m = 1, 2, 3…36). The detection region on the terahertz detector (5) corresponds to the artificial microstructure array of the modulation device (3) one by one. When the modulation device (3) is cut out of the light path, the area array detector of the terahertz detector (5) can be regarded as an image sensor, and the intensity information of each region in space can be obtained; when the modulation device (3) is cut into the light path, each detection region only receives the signal of the terahertz light after passing through the corresponding artificial microstructure array.

[0084] The data processing module (6) is connected between the terahertz detector (5), reads the light intensity information output by the terahertz detector (5), and reconstructs the multi-element terahertz spectrum information through the pre-trained machine learning algorithm.

[0085] Figure 4 The working flow diagram of the multi-element terahertz information detection according to an example embodiment of the present disclosure is shown, mainly including three parts.

[0086] The first part is to measure the spatial information and intensity information of the terahertz light.

[0087] The modulation device can be cut out of the light path, and the light intensity signal received by each detection region of the terahertz detector is the intensity distribution information of the terahertz light to be measured in space. As shown in the example, there are 6X6 detection regions, and 36 light intensity values can be obtained. Figure 3

[0088] The second part is the phase reconstruction process.

[0089] The training set required for training the phase reconstruction network can be obtained first. The training set can be prepared by using micro-nano processing technology to prepare terahertz super surface devices based on artificial microstructures, so that these devices have different diffraction effects on terahertz waves and have typical spectral transmission characteristics. In this embodiment, the training set contains 1000 samples, numbered as n (n = 1, 2, 3…1000).

[0090] Then the test light path is built, and a broadband terahertz light source is used. The terahertz light emitted by the broadband terahertz light source passes through the sample to obtain the terahertz light as the research target. At this time, the modulation device is cut out of the light path, and the spatial intensity distribution information A(x, y, z = d) and the initial phase distribution information Φ(x, y, z = 0) of the terahertz light after passing through the sample are tested by using the terahertz time domain spectrum system. A(x, y, z = d) is the input of the phase reconstruction network, and the phase distribution information Φ is the output of the phase reconstruction network.

[0091] Then the phase reconstruction algorithm is designed, that is, the spatial intensity distribution information A and the initial phase distribution information Φ of the 1000 samples are used to train the phase reconstruction network to determine the network parameters.​

[0092] On the basis of the training of the phase reconstruction network, the first spatial intensity distribution information A'(x, y, z=d) of the to-be-tested terahertz light, i.e., the spatial intensity distribution information of the to-be-tested terahertz light collected by the terahertz detector when the light path is cut by the regulation device, is input into the trained phase reconstruction network, and the phase information Φ'(x, y, z=0) of the to-be-tested terahertz light can be predicted.

[0093] The third part is the spectral and polarization reconstruction process.

[0094] The transmission spectrum f n of 1000 samples is tested by using a terahertz time-domain spectroscopy system xn The polarization information P yn of the sample in the X direction and the polarization information P xn and P yn of the sample in the Y direction are known. The polarization information P xn and P yn of the sample can be designed by using a polarizer or other device, or the polarization information P k and P n of each sample can be measured by the instrument in advance.

[0095] Then, a test light path is built, a broadband terahertz light source is used, and the emission spectrum of the light source is S0. The terahertz spectrum S k after the light source passes through each sample can be calculated according to S n .

[0096] The regulation device is cut into the light path between the sample and the terahertz detector. The terahertz light after passing through each sample is irradiated onto the regulation device and is completely covered by the multiple artificial microstructure array patterns on the regulation device. The terahertz light regulated by each artificial microstructure array enters the corresponding detection region of the terahertz detector, and the light intensity information is recorded as I mn , where n represents the sample number, m is the number of the microstructure array, m=1, 2, …, 36, and m can also be considered as the number of the detection region of the terahertz detector. I mn is the input of the spectral and polarization reconstruction network, and P xn , P yn , and S n are the outputs of the network.

[0097] Then, the spectral and polarization reconstruction algorithm is designed, i.e., the spatial intensity distribution information I mn and the polarization information P xn , P yn , and S n of all samples in the training set are used to train the spectral and polarization reconstruction network to determine the network parameters.

[0098] Based on the completed training of the spectral polarization reconstruction network, the control device is positioned between the terahertz source and the terahertz detector, and the second spatial intensity distribution information I' of the terahertz light to be measured is acquired through the terahertz detector. m The spectral and polarization information of the terahertz light to be measured can be predicted by inputting it into the trained spectral polarization reconstruction network.

[0099] The above examples provide information on the spatial intensity distribution, total intensity, phase, spectrum, and polarization of the terahertz light to be measured.

[0100] For other details and beneficial effects of this embodiment, please refer to the relevant description above, which will not be repeated here.

[0101] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the data processing device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0102] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] While this specification contains numerous specific implementation details, these should not be construed as limiting the scope of any invention or the scope of the claims, but rather are primarily intended to describe features of specific embodiments of a particular invention. Certain features described in the various embodiments herein may also be implemented in combination in a single embodiment. Conversely, various features described in a single embodiment may also be implemented separately in various embodiments or in any suitable sub-combination. Furthermore, while features may function in certain combinations as described above and even initially claimed in this way, one or more features from a claimed combination may be removed from that combination in some cases, and a claimed combination may refer to a sub-combination or a variation thereof.

[0104] Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring such order nor limiting of all illustrations to that order, nor requiring that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing can be advantageous. Moreover, the separation of various system modules and components in the embodiments described above 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.

[0105] Thus, particular embodiments of the subject matter have been described. Other embodiments are within the scope of the following claims. In some cases, actions recited in the claims can be performed in a different order and still achieve desirable results. In addition, the processes depicted in the accompanying figures need not be performed in the particular order described or in sequential order at all. In certain implementations, multitasking and parallel processing can be advantageous.

[0106] The above descriptions are only preferred embodiments of one or more embodiments of the present specification, and are not intended to limit one or more embodiments of the present specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of the present specification should be included in the protection scope of one or more embodiments of the present specification.

Claims

1. A multi-element terahertz information detection method, characterized in that, The method comprises: providing a to-be-measured terahertz light by a terahertz light source; collecting first spatial intensity distribution information of the to-be-measured terahertz light by a terahertz detector at a predetermined distance from the terahertz light source; setting a regulating device between the terahertz light source and the terahertz detector, the regulating device comprising a plurality of microstructure arrays, different microstructure arrays being designed to have different phase regulating characteristics, and collecting second spatial intensity distribution information of the to-be-measured terahertz light after phase regulation by the terahertz detector; reconstructing phase information, spectral information and polarization information of the to-be-measured terahertz light according to the first spatial intensity distribution information and the second spatial intensity distribution information by using a pre-trained neural network, comprising: reconstructing phase information of the to-be-measured terahertz light according to the first spatial intensity distribution information by using a pre-trained phase reconstruction network; reconstructing spectral information and polarization information of the to-be-measured terahertz light according to the second spatial intensity distribution information by using a pre-trained spectral polarization reconstruction network, wherein: the training process of the phase reconstruction network comprises: obtaining a training set containing a plurality of samples; measuring initial phase distribution information of terahertz light provided by each sample by a terahertz time-domain spectroscopy system, and obtaining spatial intensity distribution information at a predetermined distance from the sample by using the terahertz detector; training the phase reconstruction network based on the measured initial phase distribution information of each sample and the spatial intensity distribution information at a predetermined distance from the sample; the training process of the spectral polarization reconstruction network comprises: A training set comprising a plurality of samples is acquired, each sample having known polarization information P in the X direction xn and polarization information P in the Y direction yn n is the sample number; The transmission spectrum f of each sample is measured using a terahertz time domain spectroscopy system n and the sample is illuminated with a broadband terahertz light source with spectrum S0, and the spectrum S provided by the sample is obtained from n =S0*f n n ;​ The regulating device is arranged between the sample and the terahertz detector, and the terahertz light spatial intensity distribution information I after phase regulation is acquired by using the terahertz detector mn Wherein m is the number of the microstructure array, and n is the number of the sample. based on the acquired spatial intensity distribution information I mn and polarization information P of the corresponding sample xn , P yn and spectrum S n , the spectral polarization reconstruction network is trained.

2. The multi-element terahertz information detection method of claim 1, wherein, the plurality of microstructure arrays completely cover the irradiation area of the to-be-measured terahertz light on the regulating device.

3. The multi-element terahertz information detection method of claim 1, wherein, The regulating device comprises MXM microstructure arrays, and the terahertz detector has corresponding MXM detection regions, which correspond one-to-one to the microstructure arrays.

4. The multi-element terahertz information detection method of claim 1, wherein, The method further comprises controlling the position of the regulating device by a switching device, the switching device comprising a first gear position and a second gear position, wherein: when the switching device is set to the first gear position, the regulating device is controlled to cut out the light path of the to-be-measured terahertz light; when the switching device is set to the second gear position, the regulating device is controlled to cut into the light path of the to-be-measured terahertz light between the terahertz light source and the terahertz detector, so that the regulating device performs phase regulation on the to-be-measured terahertz light.

5. The multi-element terahertz information detection method of claim 1, wherein, The substrate of each microstructure array is composed of a material with high terahertz light transmittance, and the cells arranged on the substrate are metal structures.

6. A multi-element terahertz information detection device, characterized in that, The system comprises a terahertz detector, a regulating device, a switching device and a data processing module, wherein: the terahertz detector is arranged at a predetermined distance from a terahertz light source, and the terahertz light source is used to provide a to-be-measured terahertz light; the regulating device comprises a plurality of microstructure arrays, different microstructure arrays being designed to have different phase regulating characteristics; The switching device comprises a first gear position and a second gear position, when the switching device is set in the first gear position, the control device is controlled to cut out the optical path of the to-be-tested terahertz light, when the switching device is set in the second gear position, the control device is controlled to cut in the optical path of the to-be-tested terahertz light between the terahertz light source and the terahertz detector, so as to perform phase control on the to-be-tested terahertz light; The data processing module is configured to: reconstruct the phase information, the spectral information and the polarization information of the to-be-tested terahertz light by using a pre-trained neural network according to the spatial intensity distribution information collected by the terahertz detector when the control device cuts in and cuts out the optical path, and the data processing module is configured to: reconstruct the phase information of the to-be-tested terahertz light by using a pre-trained phase reconstruction network according to the spatial intensity distribution information collected by the terahertz detector when the control device cuts out the optical path; reconstruct the spectral information and the polarization information of the to-be-tested terahertz light by using a pre-trained spectral polarization reconstruction network according to the spatial intensity distribution information collected by the terahertz detector when the control device cuts in the optical path; The training process of the phase reconstruction network comprises: obtaining a training set containing a plurality of samples; measuring the initial phase distribution information of the terahertz light provided by each sample by using a terahertz time-domain spectroscopy system, and obtaining the spatial intensity distribution information at a predetermined distance away from the sample by using the terahertz detector; training the phase reconstruction network based on the measured initial phase distribution information of each sample and the spatial intensity distribution information at a predetermined distance away from the sample; The training process of the spectral polarization reconstruction network comprises: A training set comprising a plurality of samples is acquired, each sample having known polarization information P in the X direction xn and polarization information P in the Y direction yn n is the sample number; The transmission spectrum f of each sample is measured using a terahertz time domain spectroscopy system n and the sample is illuminated with a broadband terahertz light source with spectrum S0, and the spectrum S provided by the sample is obtained from n =S0*f n n ;​ The regulating device is arranged between the sample and the terahertz detector, and the terahertz light spatial intensity distribution information I after phase regulation is acquired by using the terahertz detector mn Wherein m is the number of the microstructure array, and n is the sample number. based on the acquired spatial intensity distribution information I mn and polarization information P of the corresponding sample xn , P yn and spectrum S n , the spectral polarization reconstruction network is trained.

7. The multi-element terahertz information detection device according to claim 6, wherein The plurality of microstructure arrays completely cover the illumination area of the to-be-tested terahertz light on the control device.

8. The multi-element terahertz information detection device according to claim 6, characterized in that, The control device comprises MxM microstructure arrays, and the terahertz detector has corresponding MxM detection areas, which correspond to the microstructure arrays one by one.

9. The multi-element terahertz information detection device of claim 6, wherein, The substrate of the microstructure array is composed of a material with high terahertz light transmittance, and the elements on the substrate are metal structures.

Images