Liquid crystal image recognizer, preparation method thereof and image recognition device

Abstract

The application discloses a liquid crystal image recognizer, a preparation method thereof and an image recognition device. The liquid crystal image recognizer comprises a liquid crystal box and a mirror. The liquid crystal box comprises a first substrate and a second substrate arranged oppositely, a first cholesteric phase liquid crystal layer and a second cholesteric phase liquid crystal layer arranged in a stack between the first substrate and the second substrate, and the first cholesteric phase liquid crystal layer and the second cholesteric phase liquid crystal layer have a plurality of orientation structures which are equally spaced and obtained by optimization of a deep learning algorithm. The mirror is located on the side of the first substrate away from the second substrate and is arranged at a preset distance from the first substrate to form a light beam folding architecture between the mirror and the liquid crystal box. The liquid crystal image recognizer provided by the application can realize independent image recognition of two kinds of circularly polarized incident light with different chirality. Through cascading of the liquid crystal box of the device, wavelength multiplexing multifunctional image recognition can be realized, so that the functional diversity and multiplexing dimension of the full-optical image recognizer are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of liquid crystal micro-nano structure manipulation and optical computing, and particularly relates to a liquid crystal image recognizer, a preparation method thereof and an image recognition device. BACKGROUND

[0002] Artificial intelligence is an important driving force for the new round of scientific and technological revolution and industrial transformation. Artificial neural networks are one of the most popular artificial intelligence frameworks, which have been widely used in multiple fields to perform computing tasks such as machine vision, natural language processing, medical diagnosis, etc. However, as Moore's law approaches its physical limit, mainstream electronic neural networks will encounter bottlenecks in processing speed and power consumption, and accelerating the development of the next generation of artificial intelligence has become a strategic issue. Optical neural networks are strong contenders for the next generation of computing platforms, which use photons instead of electrons to realize the mapping between input and output signals. Due to the high propagation speed, high parallelism and multiple degrees of freedom of optical information, optical neural networks have potential advantages such as fast speed, low power consumption and high throughput. In recent years, scholars at home and abroad have proposed various optical neural network frameworks for implementing different intelligent computing. Among them, diffractive neural networks realize the interconnection of neurons in each layer through the diffraction of light, providing an expandable and cost-effective platform for optical computing.

[0003] In 2018, the Ozcan team at the University of California, Los Angeles, proposed a 3D-printed all-optical diffractive neural network that can realize classification and imaging functions in the terahertz band, opening up a research boom in this field. Since then, the Ozcan team and teams from the University of Delaware, Tsinghua University, Southeast University, etc. have applied diffractive neural networks to various computing tasks such as image recognition, optical logic operations, all-optical phase recovery, etc. Among them, all-optical image recognition technology has always been a research hotspot in this field, and is expected to have great potential in applications such as autonomous driving and medical diagnosis. With the increasing demand for image recognition speed, accuracy and multi-functionality, a multi-channel, high-efficiency and easily integrated image recognizer is urgently needed.

[0004] Traditional all-optical image recognizers are mainly based on 3D printing, spatial light modulators, superstructures, etc., and can achieve high-accuracy image recognition. However, most of them have limitations such as non-adjustable functions, single applicable wavelength, limited efficiency, and difficulty in integration, which restricts the development of all-optical image recognition technology. SUMMARY

[0005] The present application provides a liquid crystal image recognizer, a preparation method thereof and an image recognition device, which can realize independent image recognition function for two kinds of chiral circularly polarized incident light, and through the cascading of the liquid crystal cell of the device, it can also realize wavelength multiplexing multi-functional image recognition, so as to improve the functional diversity and multiplexing dimension of the all-optical image recognizer.

[0006] According to one aspect of the present invention, a liquid crystal image recognizer is provided, comprising a liquid crystal cell and a reflector;

[0007] The liquid crystal cell includes:

[0008] A first substrate and a second substrate arranged opposite to each other;

[0009] A first axially oriented cholesteric liquid crystal layer and a second axially oriented cholesteric liquid crystal layer are stacked between the first substrate and the second substrate. A first alignment layer is provided on the side of the first substrate near the first axially oriented cholesteric liquid crystal layer, and a second alignment layer is provided on the side of the second substrate near the second axially oriented cholesteric liquid crystal layer.

[0010] The first and second oriented cholesteric liquid crystal layers have opposite directions of rotation. Both the first and second orientation layers have multiple equally spaced orientation regions. Within the orientation regions, the first and second oriented cholesteric liquid crystal layers have different orientation structures. These orientation structures are used as weight parameters for training a multilayer diffraction neural network and are obtained through iterative optimization using a deep learning algorithm.

[0011] The reflector is located on the side of the first substrate away from the first alignment layer, and the reflector is set at a preset distance from the first substrate to form a beam reflection structure between the reflector and the liquid crystal cell.

[0012] Optionally, the liquid crystal image recognizer includes at least two cascaded liquid crystal cells, wherein the center wavelengths of the Bragg reflection bands of the at least two liquid crystal cells are different, so as to form a wavelength-multiplexed multifunctional liquid crystal image recognizer.

[0013] Optionally, the first alignment layer and the second alignment layer are provided with a plurality of alignment regions of the same size, and the positions of the corresponding alignment regions in the first alignment layer and the second alignment layer are aligned along the direction from the first substrate to the second substrate;

[0014] For the first alignment layer or the second alignment layer, the alignment regions are evenly spaced and distributed along the direction from the first substrate to the second substrate, and there is no overlap between different alignment regions.

[0015] Optionally, in different orientation regions, the first axially oriented cholesteric liquid crystal layer and the second axially oriented cholesteric liquid crystal layer are provided with different orientation structures, and the total number of orientation structures is equal to twice the number of orientation regions;

[0016] Outside the orientation region, the first axially oriented cholesteric liquid crystal layer and the second axially oriented cholesteric liquid crystal layer have random and disordered surface orientations;

[0017] The orientation structures are obtained by optimizing a deep learning algorithm, the deep learning algorithm is based on a multi-layer diffraction neural network architecture, an input layer is an image to be identified, an output layer is light intensity in a plurality of preset detection regions, weights of a hidden layer are set as phases, a number of layers of the hidden layer is equal to a number of times of beam turn-back through the orientation region, and layers are connected by an angular spectrum diffraction formula; according to a cholesteric liquid crystal geometric phase principle, for a left-handed helical structure cholesteric liquid crystal layer, the orientation structure is equal to-0.5 times of corresponding hidden layer weights after iteration, and for a right-handed helical structure cholesteric liquid crystal layer, the orientation structure is equal to+0.5 times of corresponding hidden layer weights after iteration.

[0018] Optionally, the first cholesteric liquid crystal layer is of a left-handed helical structure, and the second cholesteric liquid crystal layer is of a right-handed helical structure, or the first cholesteric liquid crystal layer is of a right-handed helical structure, and the second cholesteric liquid crystal layer is of a left-handed helical structure.

[0019] In the same liquid crystal cell, the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer have a circularly polarized selective Bragg reflection band with the same center wavelength.

[0020] Optionally, the liquid crystal cell further comprises a spacer arranged between the first substrate and the second substrate to control a total thickness of the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer.

[0021] The spacer has an extension length greater than or equal to 20 times of a pitch of liquid crystal molecules in the first cholesteric liquid crystal layer.

[0022] According to another aspect of the present application, a preparation method of a liquid crystal image identifier is provided for preparing the above-mentioned liquid crystal image identifier, and the preparation method comprises the following steps:

[0023] A first substrate and a second substrate are provided, a first orientation layer is formed on one side of the first substrate, and a second orientation layer is formed on one side of the second substrate;

[0024] The first substrate and the second substrate are arranged oppositely, a first cholesteric liquid crystal layer and a second cholesteric liquid crystal layer with a plurality of equidistantly distributed orientation structures are prepared between the first substrate and the second substrate to form a liquid crystal cell;

[0025] A mirror is arranged on the first substrate of the liquid crystal cell away from the first orientation layer to form the liquid crystal image identifier.

[0026] The first orientation layer is located on one side of the first substrate close to the first cholesteric phase liquid crystal layer, the second orientation layer is located on one side of the second substrate close to the second cholesteric phase liquid crystal layer, and the orientation structures of the first cholesteric phase liquid crystal layer and the second cholesteric phase liquid crystal layer are different from each other and are trained as weight parameters of a multi-layer diffraction neural network and are obtained through iterative optimization of a deep learning algorithm.

[0027] Optionally, a first orientation layer is formed on one side of the first substrate, and a second orientation layer is formed on one side of the second substrate, comprising:

[0028] A solution containing an orientation material is spin-coated on one side of the first substrate, and after spin-coating, the first substrate is annealed to form the first orientation layer;

[0029] A uniformly mixed solution containing an orientation material and a photoinitiator is spin-coated on one side of the second substrate, and after spin-coating, the second substrate is annealed to form the second orientation layer.

[0030] Optionally, a first cholesteric phase liquid crystal layer and a second cholesteric phase liquid crystal layer with a plurality of equidistantly distributed orientation structures are prepared between the first substrate and the second substrate to form a liquid crystal cell, comprising:

[0031] Linearly polarized ultraviolet exposure is performed on the empty cell composed of the first substrate, the second substrate, the first orientation layer and the second orientation layer, and the orientation structures corresponding to the second cholesteric phase liquid crystal layer are exposed equidistantly in turn;

[0032] A mixture of a second cholesteric phase liquid crystal and a polymer monomer is filled into the space between the first substrate and the second substrate to form the second cholesteric phase liquid crystal layer with the orientation structures;

[0033] Surface-induced ultraviolet polymerization is performed on the formed second cholesteric phase liquid crystal layer, and unreacted molecules are washed away with acetone, the second cholesteric phase liquid crystal layer shrinks to form a second cholesteric polymer network scaffold connected to the surface of the second substrate;

[0034] A second round of linearly polarized ultraviolet exposure is performed, and the orientation structures corresponding to the first cholesteric phase liquid crystal layer are exposed equidistantly in turn, and the orientation directions of the first orientation layer and the second orientation layer are rewritten to be consistent with the orientation structures of the second round of exposure;

[0035] A first-handed cholesteric liquid crystal is filled into a space between the first substrate and the second-handed cholesteric liquid crystal layer, forming the first-handed cholesteric liquid crystal layer with the orientation structure of the second round exposure, while the second-handed cholesteric liquid crystal layer rebounds, forming the liquid crystal cell.

[0036] According to another aspect of the present application, there is also provided an image recognition device, comprising a laser, a first polarizer, a first lens, a second lens, a first quarter wave plate, a beam splitter, a spatial light modulator, a second quarter wave plate, a second polarizer, a third lens, a fourth lens, a third quarter wave plate, the liquid crystal image recognizer and a detector;

[0037] The light beam output by the laser passes through a beam expander system composed of the first polarizer, the first lens and the second lens, the first quarter wave plate and the beam splitter in sequence, is modulated and reflected by the spatial light modulator, is reflected by the beam splitter, and then passes through a 4f imaging system composed of the second quarter wave plate, the second polarizer, the third lens and the fourth lens, and the third quarter wave plate in sequence to obtain circularly polarized incident light of the image to be recognized, which is incident from the first substrate side of the liquid crystal image recognizer and is reflected back and forth between the orientation regions and the mirror distributed at equal intervals, and the reflected light which has undergone multiple geometric phase modulations is incident to the detector after diffraction by a preset distance, and the detector classifies the image to be recognized according to the cumulative light intensity in each detection region.

[0038] Rotating the third quarter wave plate changes the circular polarization chirality of the incident light to switch different functions of the liquid crystal image recognizer.

[0039] The liquid crystal image recognizer provided by the embodiment of the present application comprises a liquid crystal box and a mirror; the liquid crystal box comprises: a first substrate and a second substrate arranged oppositely; a first chiral cholesteric liquid crystal layer and a second chiral cholesteric liquid crystal layer arranged in a stack between the first substrate and the second substrate, the first substrate is provided with a first alignment layer on the side close to the first chiral cholesteric liquid crystal layer, and the second substrate is provided with a second alignment layer on the side close to the second chiral cholesteric liquid crystal layer; wherein the chiral directions of the first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer are opposite, the first alignment layer and the second alignment layer are both provided with a plurality of alignment areas distributed at equal intervals, the first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer have different alignment structures in the alignment areas, the alignment structures are all trained as weight parameters of a multi-layer diffraction neural network, and are obtained through iterative optimization of a deep learning algorithm; the mirror is located on the side of the first substrate away from the first alignment layer, and the mirror and the first substrate are provided with a preset distance to form a light beam folding architecture between the mirror and the liquid crystal box. The technical scheme of the embodiment of the present application can independently perform multiple geometric phase modulations on two kinds of chiral circularly polarized incident light, and further realize the image recognition function of circular polarization multiplexing. In addition, the cholesteric liquid crystal also has wavelength selectivity, and only has reflection and geometric phase modulation effect on light with a wavelength in the Bragg reflection band, and transmits light out of the band without phase modulation effect. The liquid crystal image recognizer provided by the embodiment of the present application can further realize the multifunctional image recognition of wavelength multiplexing by cascading a plurality of liquid crystal boxes with different center wavelengths of Bragg reflection bands, and effectively improves the limitations of the traditional all-optical image recognizer, such as unadjustable function, single applicable wavelength, limited efficiency and difficulty in integration.

[0040] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0042] Figure 1 A y-z side structure schematic diagram of a liquid crystal image recognizer provided by the embodiment of the present application;

[0043] Figure 2 A y-z side structure schematic diagram of another liquid crystal image recognizer provided by the embodiment of the present application;

[0044] Figure 3A schematic diagram of a calculation method of a liquid crystal layer surface orientation structure of a liquid crystal image recognizer provided by the embodiment of the present application is shown in the figure.

[0045] Figure 4 A schematic diagram of a flow of a preparation method of a liquid crystal image recognizer provided by the embodiment of the present application is shown in the figure.

[0046] Figure 5 A schematic diagram of a process flow of a liquid crystal cell in a liquid crystal image recognizer provided by the embodiment of the present application is shown in the figure.

[0047] Figure 6 A schematic diagram of spectral characterization results of three Bragg reflection band center wavelength different liquid crystal image recognizers provided by the embodiment of the present application is shown in the figure.

[0048] Figure 7 A schematic diagram of a structure of an image recognition device provided by the embodiment of the present application is shown in the figure.

[0049] Figure 8 A schematic diagram of a detection area light spot and recognition result of a liquid crystal image recognizer provided by the embodiment of the present application for multi-wavelength recognition of two types of images is shown in the figure.

[0050] Figure 9 A schematic diagram of a confusion matrix result of a liquid crystal image recognizer provided by the embodiment of the present application for multi-wavelength recognition of all test set images in two data sets. DETAILED DESCRIPTION

[0051] In order to make the personnel in the technical field better understand the present application scheme, the technical scheme in the embodiment of the present application will be described clearly and completely below by combining the figures in the embodiment of the present application. Obviously, the described embodiment is only a part of the embodiment of the present application, not all the embodiments. Based on the embodiment in the present application, all other embodiments obtained by the personnel of ordinary skill in the art without creative labor should belong to the protection scope of the present application.

[0052] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-mentioned figures are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0053] Figure 1 A y-z side structure schematic diagram of a liquid crystal image recognizer provided by an embodiment of the present application.

[0054] Wherein, the x and y directions are parallel to the plane where the first substrate is located, and the z direction is perpendicular to the plane where the first substrate is located. With reference to Figure 1 The liquid crystal image recognizer provided by the embodiment of the present application comprises a liquid crystal cell 100 and a mirror 200. The liquid crystal cell 100 comprises: a first substrate 110 and a second substrate 120 arranged oppositely; a first cholesteric phase liquid crystal layer 130 and a second cholesteric phase liquid crystal layer 140 arranged in a stack between the first substrate 110 and the second substrate 120, a first orientation layer 150 arranged on the side of the first substrate 110 close to the first cholesteric phase liquid crystal layer 130, and a second orientation layer 160 arranged on the side of the second substrate 120 close to the second cholesteric phase liquid crystal layer 140; wherein, the rotation directions of the first cholesteric phase liquid crystal layer 130 and the second cholesteric phase liquid crystal layer 140 are opposite, and the first orientation layer 150 and the second orientation layer 160 are both provided with a plurality of orientation areas distributed at equal intervals Figure 1 Three orientation areas are schematically shown, i.e. the areas corresponding to the dark gray in the orientation layer, which are not a limitation of the embodiment of the present application, in the orientation areas, the first cholesteric phase liquid crystal layer 130 and the second cholesteric phase liquid crystal layer 140 have different orientation structures, the orientation structures are trained as weight parameters of a multi-layer diffraction neural network, and are obtained through iterative optimization of a deep learning algorithm; the mirror 200 is arranged on the side of the first substrate 110 away from the first orientation layer 150, and the mirror 200 and the first substrate 110 are arranged at a preset distance to form a light beam folding architecture between the mirror 200 and the liquid crystal cell 100.

[0055] The first substrate 110 and the second substrate 120 can be flexible substrates such as polyimide, or rigid substrates such as quartz substrates or glass substrates. In some embodiments, the first substrate 110 and the second substrate 120 are flat plates, and the mirror 200 is a planar mirror. The alignment materials in the first alignment layer 150 and the second alignment layer 160 can be at least one of photo-crosslinking materials, photo-degradable materials, and photo-induced cis-trans isomerization materials, such as a photo-sensitive azo material SD1. These materials are photo-controlled alignment materials that can undergo physical or chemical reactions under the irradiation of linearly polarized ultraviolet light, generating anisotropic surface forces, and then inducing the directional arrangement of liquid crystal molecules. The second alignment layer 160 contains not only alignment materials, but also a uniform mixture of photo-initiators, such as benzophenone, which can induce polymerization starting from the surface of the second alignment layer 160. The first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 exhibit periodic helical structures of specific handedness under the anchoring of the alignment layers and the action of chiral agents. In some embodiments, the first cholesteric liquid crystal layer 130 has a left-handed helical structure, and the second cholesteric liquid crystal layer 140 has a right-handed helical structure, or vice versa. The selection can be made according to actual conditions. In the same liquid crystal cell 100, the first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 have circularly polarized selective Bragg reflection bands with the same center wavelength. The first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 have the same refractive index and pitch, and thus have circularly polarized selective Bragg reflection bands with the same center wavelength. In the alignment region, the first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 can be aligned according to different alignment structures. In the non-alignment region, the first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 have chaotic and disordered surface orientations. In the Bragg reflection band, the cholesteric liquid crystal can reflect circularly polarized light of the same handedness as the liquid crystal layer, and impart a geometric phase equal to ±2 times the surface orientation of the cholesteric liquid crystal layer. At the same time, it transmits circularly polarized light of the opposite handedness. Light is reflected back and forth between the cholesteric liquid crystal layer and the mirror 200. The mirror surface reflection of the mirror 200 flips the circular polarization of the light. Therefore, the first circularly polarized incident light is reflected by the first cholesteric liquid crystal layer 130, the mirror 200, the second cholesteric liquid crystal layer 140, the mirror 200, and the first cholesteric liquid crystal layer 130 in turn, and is modulated by multiple corresponding alignment structures in turn. The second circularly polarized incident light is reflected by the second cholesteric liquid crystal layer 140, the mirror 200, the first cholesteric liquid crystal layer 130, the mirror 200, and the second cholesteric liquid crystal layer 140 in turn, and is modulated by multiple corresponding alignment structures in turn.By utilizing the back-and-forth reflection of light between the liquid crystal cell 100 and the reflector 200, the technical solution of this embodiment can independently perform three geometric phase modulations on circularly polarized incident light of two chiral types, thereby realizing the image recognition function of circular polarization multiplexing.

[0056] Optional, continue to refer to Figure 1 The liquid crystal cell 100 also includes a spacer 170 disposed between the first substrate 110 and the second substrate 120 to control the total thickness of the first axial cholesteric liquid crystal layer 130 and the second axial cholesteric liquid crystal layer 140.

[0057] The spacer 170 can be a quartz microsphere or a quartz pillar, and can be disposed at the boundary position of the first substrate 110 and the second substrate 120 to support the first substrate 110 and the second substrate 120, forming the filling space of the first oriented cholesteric liquid crystal layer 130 and the second oriented cholesteric liquid crystal layer 140. Optionally, the extension length of the spacer 170 is greater than or equal to 20 times the pitch of the liquid crystal molecules in the first oriented cholesteric liquid crystal layer 130, so that the thickness of each of the first oriented cholesteric liquid crystal layer 130 and the second oriented cholesteric liquid crystal layer 140 can be greater than or equal to 10 times the pitch of the liquid crystal molecules in the layer, thereby enabling both the first oriented cholesteric liquid crystal layer 130 and the second oriented cholesteric liquid crystal layer 140 to have high reflectivity when circular polarization-selective Bragg reflection occurs. It is understood that... Figure 1 The image shows only the illustrative positional relationship of the spacer 170 for supporting the first substrate 110 and the second substrate 120, and not the actual size and proportion.

[0058] Optional, continue to refer to Figure 1 The first orientation layer 150 and the second orientation layer 160 are provided with multiple orientation regions of the same size. Figure 1 Medium-dark gray areas) and unoriented areas ( Figure 1 (Light gray area), along the direction from the first substrate 110 to the second substrate 120, the corresponding orientation regions in the first orientation layer 150 and the second orientation layer 160 are aligned; for the first orientation layer 150 or the second orientation layer 160, the orientation regions are distributed at equal intervals, and there is no overlap between different orientation regions along the direction from the first substrate 110 to the second substrate 120.

[0059] Optionally, the first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 in different orientation regions are provided with different orientation structures, the total number of the orientation structures is equal to twice the number of the orientation regions; outside the orientation regions (non-orientation regions), the first cholesteric liquid crystal layer 130 and the second cholesteric liquid crystal layer 140 have a chaotic and disordered surface orientation; the orientation structures are all obtained by deep learning algorithm, the deep learning algorithm is based on a multi-layer diffraction neural network architecture, the input layer is an image to be identified, the output layer is light intensity in a plurality of preset detection regions, the weight of the hidden layer is set as a phase, the number of layers of the hidden layer is equal to the number of times of beam turn-back through the orientation region, and the layers are connected by an angular spectrum diffraction formula; according to the geometric phase principle of the cholesteric liquid crystal, for the cholesteric liquid crystal layer with a left-handed helical structure, the orientation structure is equal to-0.5 times the corresponding hidden layer weight after iteration, and for the cholesteric liquid crystal layer with a right-handed helical structure, the orientation structure is equal to +0.5 times the corresponding hidden layer weight after iteration.

[0060] Optionally, the liquid crystal image identifier includes at least two liquid crystal boxes arranged in cascade, and the Bragg reflection band center wavelengths of the at least two liquid crystal boxes are different, so as to form a wavelength multiplexing multifunctional liquid crystal image identifier.

[0061] For example, the liquid crystal image identifier includes three liquid crystal boxes arranged in cascade, Figure 2 For another y-z side structure schematic diagram of a liquid crystal image identifier provided by the embodiment of the application. Referring to Figure 2 , the liquid crystal box 100 of the liquid crystal image identifier includes a first liquid crystal box 101, a second liquid crystal box 102 and a third liquid crystal box 103, and the Bragg reflection band center wavelengths of the first liquid crystal box 101, the second liquid crystal box 102 and the third liquid crystal box 103 are different, so as to form a wavelength multiplexing multifunctional liquid crystal image identifier. In other embodiments, the number of liquid crystal boxes arranged in cascade can be set according to actual needs, and the number of liquid crystal boxes arranged in cascade can be designed according to actual conditions in specific implementation.

[0062] For example, Figure 3 For a calculation method schematic diagram of a liquid crystal layer surface orientation structure of a liquid crystal image identifier provided by the embodiment of the application, which can be used to calculate the orientation structure required to realize the above-mentioned image recognition function. Referring to Figure 3 , all the orientation structures are obtained by deep learning algorithm, wherein the deep learning algorithm is based on a multi-layer diffraction neural network architecture, the input layer is an image to be identified, the output layer is light intensity in a plurality of preset detection regions, the weight of the hidden layer is set as a phase, the number of layers of the hidden layer is equal to the number of times of beam turn-back through the orientation region Figure 3 , and the layers are connected by an angular spectrum diffraction formula. According to the first training set Figure 3and the first probe region, the first layer phase weight φ1, the second layer phase weight φ2 and the third layer phase weight φ3 can be obtained by optimization. According to the second training set Figure 3 and the second probe region, the first layer phase weight φ4, the second layer phase weight φ5 and the third layer phase weight φ6 can be obtained by optimization.

[0063] Exemplarily, Figure 3 In the embodiment, the first rotation direction is left-handed and the second rotation direction is right-handed, which is not a limitation of the embodiment. Figure 1 In the embodiment, the three orientation regions arranged from left to right are defined as the first orientation region, the second orientation region and the third orientation region. In the first orientation region, the first rotation direction cholesteric liquid crystal layer 130 is provided with the first orientation structure and the second rotation direction cholesteric liquid crystal layer 140 is provided with the second orientation structure. In the second orientation region, the first rotation direction cholesteric liquid crystal layer 130 is provided with the third orientation structure and the second rotation direction cholesteric liquid crystal layer 140 is provided with the fourth orientation structure. In the third orientation region, the first rotation direction cholesteric liquid crystal layer 130 is provided with the fifth orientation structure and the second rotation direction cholesteric liquid crystal layer 140 is provided with the sixth orientation structure. According to the cholesteric liquid crystal geometric phase principle, for the left-handed cholesteric liquid crystal layer with a left-handed helical structure, the orientation structure is equal to-0.5 times the corresponding hidden layer weight after iteration, and for the right-handed cholesteric liquid crystal layer with a right-handed helical structure, the orientation structure is equal to+0.5 times the corresponding hidden layer weight after iteration. Therefore, the first orientation structure α1=-0.5φ1, the fourth orientation structure α4=0.5φ2, the fifth orientation structure α5=-0.5φ3, the second orientation structure α2=0.5φ4, the third orientation structure α3=-0.5φ5 and the sixth orientation structure α6=0.5φ6. In the embodiment, the first orientation structure, the third orientation structure and the fifth orientation structure are distributed at equal intervals in the first rotation direction cholesteric liquid crystal layer 130, and the second orientation structure, the fourth orientation structure and the sixth orientation structure are distributed at equal intervals in the second rotation direction cholesteric liquid crystal layer 140. It can be understood that the technical solution of the embodiment can independently perform three times of geometric phase modulation on circularly polarized incident light of two kinds of chirality, thereby realizing the circular polarization multiplexing image recognition function. Exemplarily, for the circularly polarized incident light of the first rotation direction, the recognition function of the handwritten digital image can be realized, and for the circularly polarized incident light of the second rotation direction, the recognition function of the fashion image can be realized.

[0064] It should be noted that, Figure 3 The calculation method of the liquid crystal layer surface orientation structure is only exemplarily provided, and is not a limitation of the embodiment. In other embodiments, other calculation methods can be used to obtain a liquid crystal layer surface orientation structure with similar functions according to specific requirements.

[0065] The liquid crystal image recognizer provided by the embodiment of the present application can independently perform multiple geometric phase modulations on two kinds of chiral circularly polarized incident light by arranging a first chiral cholesteric liquid crystal layer and a second chiral cholesteric liquid crystal layer with opposite rotation directions and the same Bragg reflection band center wavelength and forming a plurality of equidistantly distributed orientation structures in the first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer, and further realizing the circular polarization multiplexing image recognition function. In addition, the cholesteric liquid crystal has wavelength selectivity, and only has reflection and geometric phase modulation effect on light with a wavelength in the Bragg reflection band, and is suitable for constructing a multi-wavelength optical device. The liquid crystal image recognizer provided by the embodiment can further realize wavelength multiplexing multifunctional image recognition by cascading a plurality of liquid crystal boxes with different Bragg reflection band center wavelengths. The multifunctional liquid crystal image recognizer can effectively improve the limitations of the conventional full-optical image recognizer, such as unadjustable function, single applicable wavelength, limited efficiency and difficulty in integration.

[0066] Figure 4 The flowchart of the preparation method of the liquid crystal image recognizer provided by the embodiment of the present application is used for preparing the liquid crystal image recognizer provided by the above embodiment, and the preparation method is described with reference to Figure 4 The preparation method comprises the following steps.

[0067] S110, a first substrate and a second substrate are provided, a first orientation layer is formed on one side of the first substrate, and a second orientation layer is formed on one side of the second substrate.

[0068] The first substrate and the second substrate can be a flexible substrate or a rigid substrate with high light transmittance (greater than or equal to 85%). For example, the material of the first substrate and the second substrate can be glass, and the thickness of the substrate can be 1mm-2mm. The first orientation layer is located on the side of the first substrate close to the second substrate, and the second orientation layer is located on the side of the second substrate close to the first substrate.

[0069] Optionally, the first orientation layer is formed on one side of the first substrate, and the second orientation layer is formed on one side of the second substrate, comprising the following steps.

[0070] A solution containing an orientation material is spin-coated on one side of the first substrate, and after spin-coating, the first substrate is annealed to form the first orientation layer.

[0071] A uniform mixed solution containing an orientation material and a photoinitiator is spin-coated on one side of the second substrate, and after spin-coating, the second substrate is annealed to form the second orientation layer.

[0072] Exemplarily, in the embodiment, the orienting material is selected as a photosensitive azo material SD1, the spin coating solution without photoinitiator contains 0.35% of SD1 and 99.65% of dimethylformamide; the photoinitiator is selected as benzophenone, and the spin coating solution with photoinitiator contains 0.35% of SD1, 0.15% of benzophenone and 99.5% of dimethylformamide.

[0073] Exemplarily, the spin coating process can include: first, adjusting the rotation speed to 600-900 rpm, controlling the first stage spin coating time to 5-10 s, so that the material is uniformly distributed on the substrate surface; then adjusting the rotation speed to 2500-3500 rpm, controlling the second stage spin coating time to 30-50 s, so that the material is coated according to a specific thickness. Optionally, the thickness of the first orienting layer and the second orienting layer can be 30-50 nm.

[0074] Exemplarily, the annealing process can include: the annealing atmosphere is air, the annealing temperature is 80-120°C, and the annealing time is 8-12 min.

[0075] It should be noted that the spin coating solution composition, spin coating parameters and annealing parameters described above are only exemplary descriptions, and in other embodiments, they can be adjusted according to actual needs.

[0076] S120, oppositely arranging the first substrate and the second substrate, preparing the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer with a plurality of equally spaced orientation structures between the first substrate and the second substrate, and forming a liquid crystal cell.

[0077] Wherein, when the first substrate and the second substrate are opposite, the first orienting layer and the second orienting layer are oppositely arranged. In specific implementation, the second cholesteric liquid crystal layer with the second orientation structure, the fourth orientation structure and the sixth orientation structure can be prepared between the first substrate and the second substrate first, and then the first cholesteric liquid crystal layer with the first orientation structure, the third orientation structure and the fifth orientation structure can be prepared between the first substrate and the second cholesteric liquid crystal layer. The second orientation structure, the fourth orientation structure and the sixth orientation structure are equally spaced, and are obtained by deep learning algorithm optimization. The first orientation structure, the third orientation structure and the fifth orientation structure are also equally spaced, and are obtained by deep learning algorithm optimization and are aligned with the positions of the second orientation structure, the fourth orientation structure and the sixth orientation structure. The sizes of the first orientation structure, the second orientation structure, the third orientation structure, the fourth orientation structure, the fifth orientation structure and the sixth orientation structure are the same, but the structures are different.

[0078] Optionally, when the first substrate and the second substrate are oppositely arranged, the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer with a plurality of equally spaced orientation structures are prepared between the first substrate and the second substrate, and a liquid crystal cell is formed, further comprising:

[0079] A spacer is arranged between the first substrate and the second substrate.

[0080] The spacer can be a quartz microsphere or a quartz column, and is arranged at a boundary position of the first substrate and the second substrate, and is used for supporting the first substrate and the second substrate, and forming a filling space of the first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer. The extension length of the spacer in a direction perpendicular to the first substrate and the second substrate is greater than or equal to 20 times of a pitch of liquid crystal molecules in the first chiral cholesteric liquid crystal layer.

[0081] S130, a mirror is arranged on a side of the liquid crystal cell away from the first alignment layer, and a liquid crystal image identifier is formed.

[0082] The mirror is arranged at a preset distance from the first substrate, so as to form a light beam folding architecture between the mirror and the liquid crystal cell.

[0083] Figure 5 A process flow diagram for preparing a liquid crystal cell in a liquid crystal image identifier is provided for the embodiment of the present application. Figure 5 Optionally, the second chiral cholesteric liquid crystal layer with the second alignment structure, the fourth alignment structure and the sixth alignment structure is prepared between the first substrate and the second substrate first, and then the first chiral cholesteric liquid crystal layer with the first alignment structure, the third alignment structure and the fifth alignment structure is prepared between the first substrate and the second chiral cholesteric liquid crystal layer, comprising:

[0084] Linearly polarized ultraviolet light is exposed to the empty cell composed of the first substrate, the second substrate, the first alignment layer and the second alignment layer, and the second alignment structure, the fourth alignment structure and the sixth alignment structure corresponding to the second chiral cholesteric liquid crystal layer are exposed in sequence at equal intervals, so that the alignment directions of the first alignment layer and the second alignment layer are consistent with the three alignment structures;

[0085] A mixture of the second chiral cholesteric liquid crystal and a polymer monomer is filled into a space between the first substrate and the second substrate, and the second chiral cholesteric liquid crystal layer with the second alignment structure, the fourth alignment structure and the sixth alignment structure is formed;

[0086] Surface-induced ultraviolet polymerization is performed on the formed second chiral cholesteric liquid crystal layer, and molecules that do not undergo polymerization reaction are washed away with acetone, the second chiral cholesteric liquid crystal layer shrinks, and a second chiral polymer network support connected with a surface of the second substrate is formed;

[0087] The second round of linear polarized ultraviolet exposure is performed, and the first, third and fifth alignment structures corresponding to the first handedness (for example, left-handed) cholesteric liquid crystal layer are exposed in sequence at equal intervals, so that the alignment directions of the first and second alignment layers are both rewritten to be consistent with the three alignment structures of the second round of exposure;

[0088] The space between the first substrate and the second handedness cholesteric liquid crystal layer is filled with the first handedness cholesteric liquid crystal, and the first handedness cholesteric liquid crystal layer with the first, third and fifth alignment structures of the second round of exposure is formed, while the second handedness cholesteric liquid crystal layer rebounds.

[0089] After the above-mentioned two photo-orientation, surface-induced photopolymerization, and washing and refilling processes, the first and second handedness cholesteric liquid crystal layers with opposite handedness and the same Bragg reflection band center wavelength are prepared between the first and second substrates, and a plurality of alignment structures are formed in the first and second handedness cholesteric liquid crystal layers at equal intervals, thereby forming a liquid crystal cell in the liquid crystal image identifier.

[0090] Optionally, the two rounds of linear polarized ultraviolet exposure include: using a digital micromirror projection system, synchronously controlling the exposure pattern and the angle of the polarizer according to the exposure sequence, and performing ultraviolet exposure on the first, second and third alignment regions of the first and second alignment layers in sequence, so that the first and second alignment layers form alignment structures distributed at equal intervals.

[0091] Illustratively, the cholesteric liquid crystal can be a mixture of nematic liquid crystal and chiral agent. In this embodiment, the nematic liquid crystal is selected to be mixed crystal E7, and the chiral agent is selected to be left-handed chiral agent S5011 or right-handed chiral agent R5011. By changing the proportion of the chiral agent, the Bragg reflection band center wavelength of the liquid crystal image identifier can be arbitrarily controlled. In the mixture of the second handedness cholesteric liquid crystal and the polymer monomer, the polymer monomer is selected to be RM257, and the mass proportion of the polymer monomer is selected to be 20%.

[0092] It should be noted that the above-mentioned cholesteric liquid crystal components and proportions are only illustrative and are not limited to the embodiments of the present application. In other embodiments, other components and proportions can be used according to specific needs.

[0093] Illustratively, Figure 6 The spectral characterization results of the three liquid crystal image identifiers with different Bragg reflection band center wavelengths provided by the embodiments of the present application are shown in the schematic diagram. Referring to Figure 6 , the abscissa represents the wavelength of incident light in nanometers (nm), and the ordinate represents the reflectivity; the reflection spectrum corresponding to the left circularly polarized light is represented by a dashed line, and the reflection spectrum corresponding to the right circularly polarized light is represented by a solid line. From Figure 6It can be seen that the liquid crystal image recognizer in the embodiment can simultaneously reflect left-handed circularly polarized light and right-handed circularly polarized light within the Bragg reflection band, and has high reflectivity. This is because the first-handed cholesteric liquid crystal layer and the second-handed cholesteric liquid crystal layer prepared have opposite handedness and the same Bragg reflection band center wavelength, and the thicknesses of the first-handed cholesteric liquid crystal layer and the second-handed cholesteric liquid crystal layer can satisfy the Bragg reflection condition. Exemplarily, Figure 6 The three graphs in FIG. 8 correspond to Figure 2 The three liquid crystal cells in FIG. 8 have a Bragg reflection band center wavelength of about 485 nm, a Bragg reflection band center wavelength of about 560 nm, and a Bragg reflection band center wavelength of about 650 nm, respectively.

[0094] It should be noted that Figure 6 The spectra corresponding to the three liquid crystal cells in FIG. 8 are only exemplary and are not a limitation on the embodiments of the present application. In other embodiments, other Bragg reflection band positions can be selected for the liquid crystal image recognizer according to actual needs. Figure 2

[0095] The liquid crystal image recognizer prepared in the embodiment of the present application comprises a first-handed cholesteric liquid crystal layer and a second-handed cholesteric liquid crystal layer which are arranged in a stack and have multi-region orientations. The first-handed cholesteric liquid crystal layer and the second-handed cholesteric liquid crystal layer are provided with different orientation structures, each of which is trained as a weight parameter of a multi-layer diffraction neural network and is obtained through iterative optimization by a deep learning algorithm. By using the processes of secondary light orientation, surface-induced light polymerization, washing and refilling, the first-handed cholesteric liquid crystal layer and the second-handed cholesteric liquid crystal layer having opposite handedness and the same Bragg reflection band center wavelength can be prepared, and a plurality of equally spaced orientation structures can be formed in the first-handed cholesteric liquid crystal layer and the second-handed cholesteric liquid crystal layer, respectively. Within the Bragg reflection band, the cholesteric liquid crystal can reflect circularly polarized light having the same handedness as the liquid crystal layer and impart a geometric phase related to the surface orientation structure of the cholesteric liquid crystal layer, while transmitting circularly polarized light having opposite handedness. Therefore, by means of the back-and-forth reflection of light between the cholesteric liquid crystal layer and the mirror, the liquid crystal image recognizer provided in the embodiment of the present application can independently perform multiple geometric phase modulations on two kinds of chiral circularly polarized incident light, thereby realizing the function of circularly polarized multiplexing image recognition. In addition, the cholesteric liquid crystal also has wavelength selectivity, and only has reflection and geometric phase modulation effects on light with a wavelength within the Bragg reflection band, and has no phase modulation effect on out-of-band light. The liquid crystal image recognizer provided in the embodiment of the present application can further realize wavelength-multiplexing multi-functional image recognition by cascading a plurality of liquid crystal cells with different Bragg reflection band center wavelengths, thereby effectively improving the limitations of traditional full-light image recognizers, such as unadjustable function, single applicable wavelength, limited efficiency, and difficulty in integration.

[0096] ​Figure 7 This is a schematic diagram of the structure of an image recognition device provided in an embodiment of the present invention. (Reference) Figure 7 The image recognition device includes a laser 1, a first polarizer 2, a first lens 3, a second lens 4, a first quarter-wave plate 5, a beam splitter 6, a spatial light modulator 7, a second quarter-wave plate 8, a second polarizer 9, a third lens 10, a fourth lens 11, a third quarter-wave plate 12, a liquid crystal image recognizer 13 provided in the above embodiment, and a detector 14. The beam output from the laser 1 passes sequentially through the beam expanding system composed of the first polarizer 2, the first lens 3, and the second lens 4, the first quarter-wave plate 5, and the beam splitter 6. It is then modulated and reflected by the spatial light modulator 7, and after being reflected by the beam splitter 6, it passes sequentially through the second... The 4f imaging system, consisting of a quarter-wave plate 8, a second polarizer 9, a third lens 10, and a fourth lens 11, along with a third quarter-wave plate 12, obtains circularly polarized incident light of the image to be identified. This light is incident from one side of the first substrate of the liquid crystal image recognizer 13, and is reflected back and forth between the equally spaced orientation regions and the reflector. After undergoing multiple geometric phase modulations, the reflected light is diffracted at a preset distance and then incident on the detector 14. The detector 14 classifies the image to be identified based on the cumulative light intensity in each detection region. Rotating the third quarter-wave plate 12 changes the circular polarization chirality of the incident light, thereby switching different functions of the liquid crystal image recognizer 13.

[0097] In this system, the first lens 3 and the second lens 4 serve as a laser beam expander to generate an incident beam that matches the area of ​​the spatial light modulator 7. The first polarizer 2, the first quarter-wave plate 5, the second quarter-wave plate 8, and the second polarizer 9 are used to change the polarization state of the light. Combined with the phase-modulated spatial light modulator, spatial light intensity modulation can be achieved to generate images of arbitrary intensity. A first plane mirror 15 and a second plane mirror 16 are also provided in the optical path to change the beam direction. The third lens 10 and the fourth lens 11 serve as a 4f imaging system to image the generated image of arbitrary intensity onto the input surface of the liquid crystal image recognizer 13. The third quarter-wave plate 12 is used to change the polarization state of the input beam and is set at an angle of ±45° to the second polarizer 9. By switching the circular polarization chirality of the input beam, different functions of the liquid crystal image recognizer 13 can be switched. In this embodiment, the liquid crystal image recognizer schematically includes three cascaded liquid crystal cells. The three cascaded liquid crystal cells are used to achieve multifunctional image recognition at different wavelengths. For example, the center-to-center distance between the devices is set to 0.5 cm. In a liquid crystal image recognizer, the reflector is arranged parallel to and opposite to the cascaded liquid crystal cell to cause the light beam to reflect back and forth between the cholesteric liquid crystal layer and the reflector. It is understandable that... Figure 7 The images shown are merely illustrative of the relative positions of the optical path elements and do not represent their actual dimensions and proportions. In other embodiments, optical elements can be designed according to actual needs. For example, lenses can be replaced with lens groups, and mirrors can be added or removed.

[0098] Exemplarily, in the embodiment, a supercontinuum laser with continuously tunable output wavelength is selected, and according to actual requirements, lasers with wavelengths of 485 nm, 560 nm and 650 nm are outputted, and the incident angle of the laser to the liquid crystal image recognizer 13 can be 5°-15°.

[0099] Exemplarily, in the embodiment, a CMOS camera is selected as the detector of the output surface, and according to the preset detection area, the cumulative light intensity in each detection area can be calculated, and then the image to be recognized is classified.

[0100] Exemplarily, Figure 8 The liquid crystal image recognizer provided in the embodiment is used for multi-wavelength recognition of the detection area light spot and the recognition result of two types of images, Figure 9 The confusion matrix result schematic diagram of the liquid crystal image recognizer provided in the embodiment for multi-wavelength recognition of all test set images in two data sets is shown in the following figure. Figure 8 When a handwritten digital image in the MNIST data set is inputted, and the incident light is set to left circularly polarized state (the polarization direction is shown by the arrow in the lower right corner of the figure), the three liquid crystal boxes cascaded in the liquid crystal image recognizer in the embodiment can accurately recognize the inputted handwritten digital image at three wavelengths of 485 nm, 560 nm and 650 nm, and according to the cumulative light intensity in each detection area, it can be known that the cumulative light intensity in the detection area corresponding to class 4 is the strongest at the three wavelengths, so the handwritten digital recognition result at the three wavelengths is class 4 (all correct); when a fashion image in the Fashion-MNIST data set is inputted, and the incident light is set to right circularly polarized state, the three liquid crystal boxes cascaded in the liquid crystal image recognizer in the embodiment can accurately recognize the inputted fashion image at three wavelengths of 485 nm, 560 nm and 650 nm, and according to the cumulative light intensity in each detection area, it can be known that the cumulative light intensity in the detection area corresponding to class 1 is the strongest at the three wavelengths, so the fashion recognition result at the three wavelengths is class 1 (all correct).

[0101] Reference Figure 9At 485nm, 560nm and 650nm, the liquid crystal image recognizer in the embodiment was used to recognize all the test set images in the MNIST dataset and the Fashion-MNIST dataset in turn, and the confusion matrix between the predicted labels and the real labels and the recognition accuracy of the two types of images at the three wavelengths were obtained. For the MNIST test set (the first test set), the matching of the predicted labels and the real labels was extremely high, and the recognition accuracies at 485nm, 560nm and 650nm were 91.09%, 91.11% and 91.18% respectively; for the Fashion-MNIST test set (the second test set), the matching of the predicted labels and the real labels was generally high, and the matching of the labels of categories 2 and 6 was relatively poor, and the recognition accuracies at 485nm, 560nm and 650nm were 82.29%, 82.42% and 82.61% respectively.

[0102] It should be noted that the bifunctional image recognition effect at 485nm, 560nm and 650nm is only exemplarily shown in the embodiment, and the liquid crystal image recognizer provided by the present application is not limited thereto. In other embodiments, according to actual needs, the pitch or orientation structure of the first chiral nematic liquid crystal layer and the second chiral nematic liquid crystal layer can be adjusted, so that the liquid crystal image recognizer provided by the present application can be applied to different wavelengths or have different image recognition functions.

[0103] The technical scheme of the embodiment of the present application can independently perform multiple geometric phase modulations on two types of chiral circularly polarized incident light by arranging the first chiral nematic liquid crystal layer and the second chiral nematic liquid crystal layer with opposite chiralities and the same Bragg reflection band center wavelength, and forming a plurality of equidistantly distributed orientation structures in the first chiral nematic liquid crystal layer and the second chiral nematic liquid crystal layer. In addition, the chiral nematic liquid crystal also has wavelength selectivity, and only has reflection and geometric phase modulation effect on light with a wavelength within the Bragg reflection band, which is suitable for constructing a multi-wavelength optical device. The liquid crystal image recognizer provided by the present application can realize circular polarization multiplexing image recognition function, and can further realize wavelength multiplexing multifunctional image recognition by cascading a plurality of devices with different Bragg reflection band center wavelengths, thereby effectively improving the limitations of the traditional all-optical image recognizer, such as unadjustable function, single applicable wavelength, limited efficiency and difficulty in integration.

[0104] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A liquid crystal image identifier, characterized by, The liquid crystal box and the mirror are included. The liquid crystal box includes: The first substrate and the second substrate are oppositely arranged; The first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer are arranged in a stack between the first substrate and the second substrate, the first substrate is provided with a first orientation layer on the side close to the first cholesteric liquid crystal layer, and the second substrate is provided with a second orientation layer on the side close to the second cholesteric liquid crystal layer; The first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer are opposite in rotation direction, the first orientation layer and the second orientation layer are each provided with a plurality of orientation areas distributed at equal intervals, the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer have different orientation structures in the orientation areas, the orientation structures are trained as weight parameters of a multi-layer diffraction neural network, and the orientation structures are obtained through iterative optimization of a deep learning algorithm; The mirror is located on the side of the first substrate away from the first orientation layer, and the mirror is provided with a preset distance from the first substrate to form a light beam folding architecture between the mirror and the liquid crystal box.

2. The liquid crystal image identifier of claim 1, wherein, The liquid crystal image recognizer includes at least two liquid crystal boxes arranged in cascade, and the Bragg reflection band center wavelengths of the at least two liquid crystal boxes are different to form a wavelength multiplexing multifunctional liquid crystal image recognizer.

3. The liquid crystal image recognizer according to claim 1, wherein The first orientation layer and the second orientation layer are provided with a plurality of orientation areas of the same size, and the corresponding orientation area positions of the first orientation layer and the second orientation layer are aligned in the direction of the first substrate pointing to the second substrate. For the first orientation layer or the second orientation layer, the orientation areas are distributed at equal intervals, and there is no overlap between different orientation areas in the direction of the first substrate pointing to the second substrate.

4. The liquid crystal image recognizer according to claim 1, characterized in that, In different orientation areas, the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer are provided with different orientation structures, and the total number of the orientation structures is equal to twice the number of the orientation areas. Outside the orientation areas, the first cholesteric liquid crystal layer and the second cholesteric liquid crystal layer have chaotic and disordered surface orientations. The orientation structures are all obtained through optimization of a deep learning algorithm, the deep learning algorithm is based on a multi-layer diffraction neural network architecture, the input layer is an image to be identified, the output layer is light intensity in a plurality of preset detection areas, the weight of a hidden layer is a phase, the number of layers of the hidden layer is equal to the number of times of light beam folding through the orientation areas, and the layers are connected by an angular spectrum diffraction formula; according to a cholesteric geometric phase principle, for a left-handed helical structure cholesteric layer, the orientation structure is equal to-0.5 times the corresponding hidden layer weight after iteration, and for a right-handed helical structure cholesteric layer, the orientation structure is equal to+0.5 times the corresponding hidden layer weight after iteration.

5. The liquid crystal image identifier of claim 1, wherein, The first cholesteric liquid crystal layer is of a left-handed helical structure, the second cholesteric liquid crystal layer is of a right-handed helical structure, or the first cholesteric liquid crystal layer is of a right-handed helical structure, and the second cholesteric liquid crystal layer is of a left-handed helical structure. The first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer have the same center wavelength of the circular polarization selective Bragg reflection band in the same liquid crystal cell.

6. The liquid crystal image identifier of claim 1, wherein, The liquid crystal cell further comprises a spacer arranged between the first substrate and the second substrate to control the total thickness of the first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer. The extension length of the spacer is greater than or equal to 20 times the pitch of the liquid crystal molecules in the first chiral cholesteric liquid crystal layer.

7. A method of preparing a liquid crystal image identifier, characterized by, The preparation method for the liquid crystal image recognizer of any one of claims 1-6 comprises: providing a first substrate and a second substrate, forming a first alignment layer on one side of the first substrate, and forming a second alignment layer on one side of the second substrate; arranging the first substrate and the second substrate oppositely, preparing a first chiral cholesteric liquid crystal layer and a second chiral cholesteric liquid crystal layer with a plurality of equidistantly distributed alignment structures between the first substrate and the second substrate to form a liquid crystal cell; arranging a mirror on the side of the first substrate of the liquid crystal cell away from the first alignment layer to form the liquid crystal image recognizer; wherein the first alignment layer is located on the side of the first substrate close to the first chiral cholesteric liquid crystal layer, the second alignment layer is located on the side of the second substrate close to the second chiral cholesteric liquid crystal layer, and the alignment structures of the first chiral cholesteric liquid crystal layer and the second chiral cholesteric liquid crystal layer are different and are trained as weight parameters of a multi-layer diffraction neural network, and are obtained through iterative optimization of a deep learning algorithm.

8. The method of claim 7, wherein the liquid crystal image identifier is prepared by the steps of: forming a first alignment layer on one side of the first substrate and a second alignment layer on one side of the second substrate comprises: spinning a solution containing an alignment material on one side of the first substrate, and after completing the spinning, annealing the first substrate to form the first alignment layer; spinning a uniformly mixed solution containing an alignment material and a photoinitiator on one side of the second substrate, and after completing the spinning, annealing the second substrate to form the second alignment layer.

9. The method for manufacturing a liquid crystal image recognizer according to claim 7, characterized in that, preparing a first chiral cholesteric liquid crystal layer and a second chiral cholesteric liquid crystal layer with a plurality of equidistantly distributed alignment structures between the first substrate and the second substrate to form a liquid crystal cell comprises: linearly polarized ultraviolet exposure is performed on the empty cell composed of the first substrate, the second substrate, the first alignment layer, and the second alignment layer, and the alignment structures corresponding to the second chiral cholesteric liquid crystal layer are exposed equidistantly in turn; a mixture of a second chiral cholesteric liquid crystal and a polymer monomer is filled into the space between the first substrate and the second substrate to form the second chiral cholesteric liquid crystal layer with the alignment structures; surface-induced ultraviolet polymerization is performed on the formed second chiral cholesteric liquid crystal layer, and molecules that have not undergone polymerization are washed away with acetone, the second chiral cholesteric liquid crystal layer shrinks, and a second chiral polymer network scaffold connected to the surface of the second substrate is formed; performing second round linear polarization ultraviolet exposure, exposing the orientation structure corresponding to the first chiral nematic liquid crystal layer in turn at equal intervals, rewriting the orientation direction of the first orientation layer and the second orientation layer to be consistent with the orientation structure of the second round exposure; filling the space between the first substrate and the second chiral nematic liquid crystal layer with the first chiral nematic liquid crystal, forming the first chiral nematic liquid crystal layer with the orientation structure of the second round exposure, while the second chiral nematic liquid crystal layer rebounds to form the liquid crystal cell.

10. An image recognition apparatus characterized by comprising: The liquid crystal image recognizer comprises a laser, a first polarizer, a first lens, a second lens, a first quarter-wave plate, a beam splitter, a spatial light modulator, a second quarter-wave plate, a second polarizer, a third lens, a fourth lens, a third quarter-wave plate, the liquid crystal image recognizer of any one of claims 1-6, and a detector. The light beam output by the laser passes through the beam expander system composed of the first polarizer, the first lens and the second lens, the first quarter-wave plate and the beam splitter in turn, is modulated and reflected by the spatial light modulator, is reflected by the beam splitter, and then passes through the 4f imaging system composed of the second quarter-wave plate, the second polarizer, the third lens and the fourth lens in turn, and the third quarter-wave plate to obtain circularly polarized incident light of the image to be recognized, which is incident from the first substrate side of the liquid crystal image recognizer and is reflected back and forth between the orientation regions and the mirror at equal intervals, and the reflected light after experiencing multiple geometric phase modulations is incident to the detector after diffraction by a preset distance, and the detector classifies the image to be recognized according to the cumulative light intensity in each detection region. Rotating the third quarter-wave plate changes the circular polarization chirality of the incident light to switch different functions of the liquid crystal image recognizer. performing second round linear polarization ultraviolet exposure, exposing the orientation structure corresponding to the first chiral nematic liquid crystal layer in turn at equal intervals, rewriting the orientation direction of the first orientation layer and the second orientation layer to be consistent with the orientation structure of the second round exposure; filling the space between the first substrate and the second chiral nematic liquid crystal layer with the first chiral nematic liquid crystal, forming the first chiral nematic liquid crystal layer with the orientation structure of the second round exposure, while the second chiral nematic liquid crystal layer rebounds to form the liquid crystal cell. The liquid crystal image recognizer comprises a laser, a first polarizer, a first lens, a second lens, a first quarter-wave plate, a beam splitter, a spatial light modulator, a second quarter-wave plate, a second polarizer, a third lens, a fourth lens, a third quarter-wave plate, the liquid crystal image recognizer of any one of claims 1-6, and a detector. The light beam output by the laser passes through the beam expander system composed of the first polarizer, the first lens and the second lens, the first quarter-wave plate and the beam splitter in turn, is modulated and reflected by the spatial light modulator, is reflected by the beam splitter, and then passes through the 4f imaging system composed of the second quarter-wave plate, the second polarizer, the third lens and the fourth lens in turn, and the third quarter-wave plate to obtain circularly polarized incident light of the image to be recognized, which is incident from the first substrate side of the liquid crystal image recognizer and is reflected back and forth between the orientation regions and the mirror at equal intervals, and the reflected light after experiencing multiple geometric phase modulations is incident to the detector after diffraction by a preset distance, and the detector classifies the image to be recognized according to the cumulative light intensity in each detection region. Rotating the third quarter-wave plate changes the circular polarization chirality of the incident light to switch different functions of the liquid crystal image recognizer.

Images