Holographic reconstruction system and method based on polarization cross-modulation cascaded diffraction network

Abstract

The application discloses a holographic reconstruction system and method based on a polarization cross-modulation cascaded diffraction network, and relates to the field of holographic reconstruction. In the system, a circularly polarized light generation module is arranged in the light path of a coherent light source, and is used for modulating the coherent light source into an input light field with different polarization state components. The polarization cross-modulation cascaded diffraction network comprises a plurality of polarization-sensitive modulation layers arranged in sequence along the light propagation direction. Each polarization-sensitive modulation layer is used for respectively performing phase modulation on different polarization state components in the input light field, and realizing polarization conversion and phase cross-modulation between the different polarization state components. A polarization analysis and detection module is arranged at the output end of the polarization cross-modulation cascaded diffraction network, and is used for separating and detecting the light intensity distribution corresponding to different polarization state components output after modulation by the polarization cross-modulation cascaded diffraction network. The application can improve the information expression capability while keeping the number of diffraction modulation layers limited, and improve the quality of image holographic reconstruction.

Description

Technical Field

[0001] This application relates to the field of holographic reconstruction, and in particular to a holographic reconstruction system and method based on a polarization cross-modulation cascaded diffraction network. Background Technology

[0002] Holographic imaging technology, by recording and reproducing the complex amplitude information of the light field, realizes the expression and reconstruction of the three-dimensional information of the target, and has important application value in optical storage, information processing, and computational imaging. With the development of all-optical computing, utilizing diffraction modulation structures to directly complete information processing during propagation has gradually become an important technical path for achieving high-speed, low-latency optical computing. In existing holographic reconstruction methods, whether it is phase retrieval based on numerical calculation or optical implementation based on diffraction networks, they essentially rely on the modulation and propagation of the spatial distribution of the input light field to reconstruct the target information. These methods are mainly designed around the two degrees of freedom of the light field amplitude and phase, and to a certain extent, they have been able to achieve light field mapping from input to output. However, these methods still have obvious limitations in terms of information expression: on the one hand, existing structures usually adopt a single-channel light field propagation mechanism, and information is modulated and output along independent paths during propagation, lacking effective coupling between channels, which limits the information carrying capacity of the system; on the other hand, the light field intensity, as the final probe, is naturally non-negative, which constrains information expression to a certain extent and is not conducive to the efficient encoding and decoupled expression of complex information. Furthermore, in multilayer diffraction structures, information transmission typically manifests as a cascaded process of independent modulation layer by layer. The relationships between layers are primarily forward propagation, lacking cross-channel information interaction mechanisms. This necessitates increasing structural depth when expressing complex, interconnected information, leading to increased system complexity and higher requirements for manufacturing precision. It is noteworthy that the polarization state of light, as a crucial intrinsic degree of freedom of the optical field, is independent of amplitude and phase, possessing a natural multi-channel information carrying capacity. While existing technologies have explored information multiplexing using polarization, most studies treat different polarization states as independent transmission channels, failing to establish effective information coupling mechanisms between polarization states. This undermines the full potential of polarization degrees of freedom in information modulation and expression. Summary of the Invention

[0003] The purpose of this application is to provide a holographic reconstruction system and method based on a polarization cross-modulation cascaded diffraction network, which can improve information expression capability and enhance the quality of image holographic reconstruction while maintaining a limited number of diffraction modulation layers.

[0004] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a holographic reconstruction system based on a polarization cross-modulation cascaded diffraction network, comprising: A circularly polarized light generation module is set in the optical path of a coherent light source to modulate the coherent light source into an input light field with different polarization state components. A polarization cross-modulation cascaded diffraction network includes multiple polarization-sensitive modulation layers arranged sequentially along the light propagation direction; each polarization-sensitive modulation layer is used to perform phase modulation on different polarization state components in the input light field, and realize polarization conversion and phase cross-modulation between different polarization state components. A polarization analysis and detection module is located at the output end of the polarization cross-modulation cascaded diffraction network, used to separate and detect the light intensity distribution corresponding to different polarization state components output after being modulated by the polarization cross-modulation cascaded diffraction network; The modulation parameters of the polarization cross-modulation cascaded diffraction network are obtained through end-to-end training and optimization, so that the input light field loaded with the hologram is propagated by the polarization cross-modulation of the polarization cross-modulation cascaded diffraction network and the target holographic image is reconstructed on the detection surface of the polarization analysis and detection module.

[0005] Secondly, this application provides a holographic reconstruction method based on a polarization cross-modulation cascaded diffraction network, applied to the aforementioned holographic reconstruction system based on a polarization cross-modulation cascaded diffraction network, comprising: The coherent light source is modulated into an input light field with different polarization state components using a circularly polarized light generation module; An input light field loaded with a target hologram is incident on a polarization cross-modulation cascaded diffraction network; the polarization cross-modulation cascaded diffraction network includes multiple polarization-sensitive modulation layers arranged sequentially along the light propagation direction; when the input light field propagates in the polarization cross-modulation cascaded diffraction network, each of the polarization-sensitive modulation layers performs phase modulation on different polarization state components in the input light field, and realizes polarization conversion and phase cross-modulation between different polarization state components; The polarization analysis and detection module is used to separate the output light field modulated by the polarization cross-modulation cascaded diffraction network into different polarization state components, and the light intensity distribution corresponding to each polarization state component is detected respectively. Based on the light intensity distribution corresponding to the detected different polarization state components, a holographic image of the target is reconstructed.

[0006] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a holographic reconstruction system and method based on a polarization cross-modulation cascaded diffraction network. By introducing a polarization cross-modulation mechanism through the polarization cross-modulation cascaded diffraction network, left- and right-hand circularly polarized light undergoes continuous polarization separation and channel coupling within the cascaded optical structure provided by the network. This achieves information coupling and dynamic redistribution between different polarization channels, thus overcoming the limitations of independent channels and low information utilization in traditional optical systems. Furthermore, the application significantly improves the hologram reconstruction quality by utilizing polarization degrees of freedom. This application constructs an all-optical computational topology with polarization information channel cross-modulation and establishes a multi-channel information coupling mechanism based on polarization degrees of freedom, enabling information to interact and recombine between different polarization states. This improves information representation capabilities while maintaining a limited number of diffraction modulation layers, thereby overcoming the full connectivity limitation of traditional single-polarization diffraction networks. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of a holographic reconstruction system based on a polarization cross-modulation cascaded diffraction network in one embodiment of this application. Figure 1 Metasurfaces are represented by geometric phase metasurface units. Figure 2 This is a schematic diagram of the training of a polarization cross-modulation cascaded diffraction network. Detailed Implementation

[0009] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0010] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0011] In one exemplary embodiment, such as Figure 1 As shown, a holographic reconstruction system based on a polarization cross-modulation cascaded diffraction network is provided. The system includes: a circularly polarized light generation module 1, a polarization cross-modulation cascaded diffraction network 2, and a polarization analysis and detection module 3. The circularly polarized light generation module 1 is set in the optical path of the coherent light source and is used to modulate the coherent light source into an input light field with different polarization state components; The polarization cross-modulation cascaded diffraction network 2 includes multiple polarization-sensitive modulation layers arranged sequentially along the light propagation direction; each polarization-sensitive modulation layer is used to perform phase modulation on different polarization state components in the input light field, and realize polarization conversion and phase cross-modulation between different polarization state components. The polarization analysis and detection module 3 is located at the output end of the polarization cross-modulation cascaded diffraction network 2, and is used to separate and detect the light intensity distribution corresponding to different polarization state components output after being modulated by the polarization cross-modulation cascaded diffraction network 2. The modulation parameters of the polarization cross-modulation cascaded diffraction network 2 are obtained through end-to-end training and optimization, so that the input light field loaded with the hologram is propagated by polarization cross-modulation of the polarization cross-modulation cascaded diffraction network 2 and the target holographic image is reconstructed on the detection surface of the polarization analysis and detection module 3.

[0012] As a specific embodiment, the circularly polarized light generation module 1 includes a spatial filter, a collimating lens, a linear polarizer, and a quarter-wave plate, used to convert linearly polarized light into left-handed or right-handed circularly polarized light.

[0013] As a specific embodiment, the polarization-sensitive modulation layer includes multiple polarization-sensitive modulation units arranged in a two-dimensional array; each polarization-sensitive modulation unit is used to modulate the incident left-hand circularly polarized light component and outputs a first component that maintains the left-hand circular polarization state and a second component that is converted to the right-hand circular polarization state; each polarization-sensitive modulation unit is also used to modulate the incident right-hand circularly polarized light component and outputs a third component that maintains the right-hand circular polarization state and a fourth component that is converted to the left-hand circular polarization state.

[0014] As a specific embodiment, the polarization-sensitive modulation unit is a geometric phase metasurface unit, i.e. Figure 1 The metasurface in the medium; and by controlling the orientation angle and / or birefringence phase delay of the subwavelength microstructure, same polarization phase modulation and cross polarization conversion are performed.

[0015] As a specific embodiment, the training and optimization process of the modulation parameters of the polarization cross-modulation cascaded diffraction network 2 is as follows: S1, Construct a training dataset; the training dataset includes multiple input hologram samples and corresponding target reconstructed images; Specifically, given 20,000 handwritten digit target images, each image is uniformly sized to 200*200 pixels and normalized so that pixel values ​​are limited to [0,1]. The target images are then combined in pairs to form composite targets A and B; information is carried by the two polarization channels of a subsequent angular spectral diffraction model. An iterative phase retrieval algorithm (Gerchberg–Saxton) is used to reconstruct the hologram (phase type) of the target at a given propagation distance / image plane. The phase value range of the holographic phase map corresponding to each target image is limited to... The size of the hologram was adjusted to 200*200 pixels and saved in BMP standard format for subsequent full optical cross-modulation polarization cross-modulation cascaded diffraction network 2 to achieve holographic reconstruction.

[0016] As a specific implementation, combining two target images to form a composite target A, B does not mean superimposing pixels, adding grayscale values, or spatially stitching two target images to form a new target image. Rather, at the training data organization level, it means assembling two independent target images into an ordered dual-target training sample according to a predetermined order, and defining them as composite targets. , The specific process is as follows: For 20,000 target images after size standardization and normalization... , ,… Image pairs are matched in order of image number without repetition. ; The first and second images form the first set of composite targets. The third and fourth images form a second set of composite targets. And so on. Composite Objectives , The dimensions remain the same. Pixels and their original normalized grayscale distributions are not fused at the pixel level; during the subsequent training of the polarization cross-modulation cascaded diffraction network, each group... Two target reconstruction images are considered to correspond to the same training sample, where, The target reconstruction image corresponding to the left-hand circular polarization output channel. The target reconstruction image corresponding to the right-hand circular polarization output channel. The left-hand circular polarization reconstruction intensity obtained from the detector surface and... Calculate the corresponding loss, the reconstructed intensity of right-hand circular polarization and Calculate the corresponding loss and optimize the network parameters using the joint loss of the two polarization channels; S2, establish an angular spectrum diffraction model for the propagation of the light field between adjacent polarization-sensitive modulation layers, and a polarization cross-modulation transmission model for each polarization-sensitive modulation layer; the angular spectrum diffraction model uses the angular spectrum method to describe the free-space diffraction propagation between adjacent polarization-sensitive modulation layers; the polarization cross-modulation transmission model uses the Jones matrix to describe the modulation effect of each polarization-sensitive modulation layer on different polarization state components. As a specific example, such as Figure 2 As shown, the process of establishing the angular spectrum diffraction model is as follows: a phase-type liquid crystal spatial light modulator is used to load the holographic dataset prepared by the iterative algorithm, and the loaded phase surface is used as the input surface for light field propagation. The input surface is illuminated by left-hand circularly polarized light to realize the forward transmission of the loaded hologram. The propagation process of the light field is approximated by the angular spectrum method. The distance between the input surface and the projection surface and the first layer of geometric phase metasurface unit is 5 cm. As a specific example, such as Figure 2 As shown, the establishment process of the polarization cross-modulation transmission model is as follows: the polarization-sensitive modulation layer includes multiple polarization-sensitive modulation units arranged in a two-dimensional array; the polarization-sensitive modulation unit is a three-layer cascaded geometric phase metasurface unit; the spacing between layers is 4mm, and each geometric phase metasurface unit is composed of 200*200 pixel neurons. Each neuron, as an independent modulated unit, receives the optical signals emitted by the neurons in the upper layer and simultaneously emits optical signals to all neurons in the lower layer, thereby forming a layered optical information interconnection.

[0017] The light field of the phase target before it reaches the first geometric phase metasurface unit under illumination by left-handed circularly polarized light. It can be represented as: ; in, Indicates that it is located at The target light field at the location, x and y represent the horizontal and vertical coordinates of the complex amplitude light field, respectively, and the subscript L indicates left-handed circularly polarized light illumination. This represents the impulse response function. Its specific expression is: ; in, Indicates the distance between neurons. This represents the axial distance between the input surface and the first layer of geometric phase metasurface elements, where j is a complex number and π is the mathematical constant pi. Indicates the wavelength of illumination from a coherent light source. nm. The polarization-sensitive modulation layer neuron structure is a metasurface array unit, and the output of each unit after phase modulation of the incident left-handed circularly polarized light. It can be represented as: ; in, This represents the incident light field containing only a left-handed circularly polarized component at the input of the first-layer geometric phase metasurface unit; specifically, the wavefront of the left-handed circularly polarized light. It can be written as complex amplitude The fundamental function of left-handed circularly polarized light The product of The Jones matrix representing the first layer of geometric phase metasurface units can be specifically represented as: ; In the above formula, i represents the imaginary number, e represents the exponent, and the left-handed circularly polarized light, after being geometrically phase modulated by the first layer of geometric phase metasurface unit, outputs... This can be represented as follows: ; As can be seen from the above, polarization occurs; the first term represents the right-handed circularly polarized light generated due to polarization after passing through the first layer of geometric phase metasurface unit, and the second term represents the left-handed circularly polarized light, that is, the left-handed circularly polarized light generates additional right-handed circularly polarized light after passing through the geometric phase metasurface unit. ; First item The second term represents the phase modulation of right-handed circularly polarized light generated after polarization separation by the geometric phase metasurface unit. This represents the phase modulation coefficient of the generated separately polarized left-handed circularly polarized light; where, Indicates the location at coordinates The delay at the location, The Sigmoid function represents the optimized phase parameters based on BP modulation. It is limited to between 0 and 2π.

[0018] In this context, BP modulation refers to geometric phase modulation based on the Berry phase effect. In the polarization-sensitive anisotropic metasurface used in this application, each modulation unit has a defined local optical axis orientation angle. When circularly polarized light passes through this modulation unit, a portion of the incident polarized light undergoes a polarization state conversion, changing from one circularly polarized state to a circularly polarized state with the opposite rotation direction under a circularly polarized substrate. Simultaneously, the polarization-converted optical field acquires a geometric phase related to the local orientation angle of the modulation unit.

[0019] For example, let the first The diffraction modulation layer is located in spatial position The local orientation angle of the anisotropic metasurface at that location is The birefringence phase retardation is ; When left-handed circularly polarized light is incident, after passing through this modulation unit, the output light field can be decomposed into a co-polarized component that retains the original polarization state and a cross-polarized component that is converted into right-handed circularly polarized light. The cross-polarized component obtains the Berry geometric phase. for; ; Correspondingly, when right-handed circularly polarized light is incident, the resulting left-handed circularly polarized component acquires a geometric phase with opposite sign. for: ; Therefore, when two circularly polarized lights with opposite rotations pass through the same anisotropic metasurface unit, they obtain geometric phase modulations with opposite signs, thus providing an important physical basis for realizing cross-modulation of left- and right-hand circularly polarized lights.

[0020] Furthermore, left-handed circularly polarized light and right-handed circularly polarized light The optical field corresponding to the left-handed circularly polarized light before it reaches the second geometric phase metasurface unit during forward propagation. The light field corresponding to right-hand circularly polarized light It can be represented as: ; ; in, The impulse response function between the first layer of geometric phase metasurface units and the second layer of geometric phase metasurface units; The polarization information of the wavefronts carried by left-handed and right-handed circularly polarized light after being cross-modulated by the second layer of geometric phase metasurface units can be expressed as: and ; Similarly, following the aforementioned polarization rules, when the light field passes through the third geometric phase metasurface unit and is generated on the detector surface after the third modulation layer, the light field can be expressed as follows: and ; S3, Design a dual-channel fusion loss function; the dual-channel fusion loss function is constructed based on the differences between the reconstructed image and the target image corresponding to the left-hand and right-hand polarization channels on the detector surface, respectively; The specific training process is as follows: B prepared holographic datasets were selected and loaded into the polarization cross-modulation cascaded diffraction network 2. To ensure training consistency, the following preprocessing was performed uniformly: the size was uniformly adjusted to 200*200 pixels, and the value range of each sample was compressed to... Meanwhile, lightweight random augmentations (rotation / translation / contrast perturbation) are performed to improve training robustness.

[0021] To simplify algorithm construction, the propagation between layers uses the angular spectrum approximation. For any polarization component... , The operator for propagation distance z is: ; In the above formula and These represent the inverse Fourier transform and the Fourier transform, respectively. This represents the complex optical field transmitted to the z-position in the plane after diffraction propagation, where, The angular spectral transfer function for free-space propagation is given by the following formula: ; in, Indicates wave number, The wavelength of the coherent light source, This represents the spatial frequency. The angular spectral transfer function can be implemented using FFT in deep learning frameworks (such as PyTorch), is a differentiable operator, and can be used for backpropagation.

[0022] Each layer of geometric phase metasurface unit can be modeled as Each unit can be equivalent to one Jones Matrix , The layer number is indicated. Under a circularly polarized substrate, to represent the cross-modulation mechanism of "same polarization modulation + opposite polarization conversion," the following equivalent form can be adopted (a feasible parameterized expression is given): ; in, , They represent the incident on the first Complex amplitude components of left-handed and right-handed circularly polarized light at the layered geometric phase metasurface unit. , These represent the left-hand and right-hand polarization components of the output after passing through the geometric phase metasurface unit, respectively. express The right-hand to left-hand phase modulation on the layered geometric phase metasurface unit. express The phase modulation amount from left-handed to right-handed on the layered geometric phase metasurface unit. To indicate Right-hand to right-hand phase modulation on a layered geometric phase metasurface unit. To indicate The left-to-left phase modulation amount on the layered geometric phase metasurface unit.

[0023] In the fabrication of metasurface units, the orientation angle typically corresponds to that of "nanopillars". Phase delay with birefringence Each layer learns only one orientation angle per pixel. The training variables are set to unconstrained real numbers. Phase is constrained by mapping. .

[0024] The process of determining the dual-channel fusion loss function specifically includes: The detector output is a reconstructed image with two polarization components: ; In the above formula, This represents the complex amplitude optical field corresponding to the left-hand circularly polarized light component on the detector surface, and similarly... This represents the complex amplitude optical field corresponding to the right-hand circularly polarized light component on the detection surface. Dual-channel fusion monitoring fully utilizes the information capacity gained from cross-modulation. This represents the reconstructed image corresponding to the left-hand circular polarization channel. This represents the reconstructed image corresponding to the right-hand circular polarization channel.

[0025] That is, the dual-channel fusion loss function It can be represented as: ; Among them, the left channel loss function and right channel loss function They are respectively: ; Right now ; in, For left-hand circularly polarized light channels, the pixel loss function is... , For right-hand circularly polarized light channels, the pixel loss function is... , For the frequency domain loss function of the left-hand circularly polarized light channel, , For the frequency domain loss function of the right-hand circularly polarized light channel, , This is the diffraction efficiency constraint term for the left-hand circularly polarized light channel. This is the diffraction efficiency constraint term for the right-hand circularly polarized light channel. , , These are non-negative weighting coefficients used to adjust the weight of the frequency domain loss function in the total loss function. These are non-negative weighting coefficients used to adjust the weight of the diffraction efficiency constraint term in the total loss function. This represents the predicted light intensity under a left-handed circularly polarized light channel. These are the training labels for the corresponding channels. This represents the number of pixels corresponding to the P channel. This represents the predicted light intensity in a right-hand circularly polarized light channel. The training labels are for right-handed circularly polarized light. Target light intensity distribution for left-hand circularly polarized light channel After Fourier transform, at the th p The value at each frequency domain sampling point Target light intensity distribution for left-hand circularly polarized light channel After Fourier transform, at the th p The value at each frequency domain sampling point Indicates the target light intensity distribution in a right-hand circularly polarized light channel. After Fourier transform, at the th p The value at each frequency domain sampling point Indicates the target light intensity distribution in a right-hand circularly polarized light channel. After Fourier transform, at the th p The value at each frequency domain sampling point The total input light intensity power. To represent the output optical power in the target reconstruction region or effective output region of a left-handed circularly polarized light channel, This represents the output optical power within the target reconstruction region or effective output region of a right-hand circularly polarized light channel.

[0026] S4. The modulation parameters of each polarization-sensitive modulation layer are jointly optimized through the backpropagation algorithm until the dual-channel fusion loss function converges.

[0027] Backpropagation is performed on the dual-channel fusion loss function to obtain the gradients of the parameters of each layer. An optimizer is then used to update the variables to be learned in each layer. Since the polarization Jones matrix has the function of polarization channel transformation, it represents the Jones matrix of the polarization-sensitive modulation layer, used to describe the modulation effect of this polarization-sensitive modulation layer on different polarization components of the incident light field. Because the polarization Jones matrix contains a cross-polarization transformation term, the gradient can be backpropagated simultaneously along both the same-polarization modulation path and the cross-polarization transformation path, thereby achieving end-to-end joint optimization of the polarization cross-modulation cascaded diffraction network.

[0028] After each training epoch, random perturbations are disabled on the validation set, and validation metrics (Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), correlation coefficient, etc.) are calculated. The parameters with the optimal validation metrics are selected as the final model. This model is used for subsequent experimental validation (such as loading a liquid crystal spatial light modulator (SLM)) or for metasurface unit structure lookup and layout generation.

[0029] As a specific embodiment, after the above training is completed, all parameters are fixed. Based on the modulation parameters obtained from the training, the structural design of the geometric phase metasurface unit, femtosecond laser processing, and physical solidification of the multilayer diffraction modulation layer are completed. The phase modulation parameters or equivalent orientation parameters (e.g., the orientation angle corresponding to the geometric phase) of each geometric phase metasurface unit are then used as the basis for the training. and delay Establish the mapping relationship between modulation parameters and microstructure geometric parameters.

[0030] The mapping relationship is determined through a pre-established structural response library, which consists of phase response and polarization conversion characteristics corresponding to different microstructure sizes, orientation angles, and heights. This ensures that the modulation parameters obtained during training can be accurately converted into machinable microstructure geometric parameters. Based on the mapping relationship, the two-dimensional modulation parameter distribution of each diffraction modulation layer is converted into the corresponding metasurface (geometric phase metasurface unit) structural layout.

[0031] Each pixel in the metasurface structure layout corresponds to a microstructure unit, whose geometric parameters (including but not limited to orientation angle, length, width, and height) are used to achieve co-polarization modulation and cross-polarization conversion of incident circularly polarized light.

[0032] High-purity quartz glass was selected as the processing substrate, and a femtosecond laser direct writing process was used to form anisotropic microstructures on the surface of the quartz glass at the subwavelength scale.

[0033] By controlling the pulse energy, repetition rate, scanning speed, and focusing conditions of a femtosecond laser, microstructure units with birefringent properties can be formed inside or on the surface of quartz glass, thereby achieving phase modulation and polarization cross-conversion of left- and right-hand circularly polarized light. Due to the high spatial resolution and low heat-affected zone characteristics of the femtosecond laser processing, the dimensional accuracy and structural consistency of the microstructures can be guaranteed.

[0034] Following the diffraction network structure set during training, the three-layer metasurface structure patterns were fabricated and sequentially fixed at predetermined axial positions, ensuring that the spacing between adjacent metasurface structure patterns matched the propagation distance used in the training phase. A high-precision displacement and alignment system was used to achieve coaxial alignment of each metasurface structure pattern, and physical fixing or encapsulation methods were employed to solidify the three-layer metasurface structure patterns, thereby forming a stable, multi-layered polarization cross-modulation diffraction network.

[0035] After the device is fabricated, the microstructure morphology of each layer of metasurface structure and its phase modulation and polarization conversion characteristics can be verified by microscopic imaging and polarization optics testing to ensure that its optical response is consistent with the modulation model used in the training phase.

[0036] As a specific implementation, a single longitudinal mode continuous wave laser was selected as the illumination source, and the output wavelength was consistent with the working wavelength set during the training phase to ensure the consistency between the experimental and design parameters.

[0037] As a specific embodiment, the polarization analysis and detection module 3 includes: a linear polarizer, a quarter-wave plate, and an imaging sensor; The linear polarizer and quarter-wave plate are used to convert circular polarization to linear polarization and to select the polarization state. The imaging sensor is used to record the light intensity distribution at the detection surface.

[0038] The imaging sensor uses a high-resolution CMOS or CCD imaging sensor; As a specific embodiment, the alignment and initial calibration process is as follows: Turn on the coherent light source and adjust the beam center and optical axis direction through beam shaping; Adjust the circularly polarized light generation module 1 so that the incident light is verified as the target circularly polarized state by the polarization analysis and detection module 3 when no sample is placed. Three layers of metasurface modulation devices (geometric phase metasurface units) were inserted one by one, and the lateral position and tilt angle of each layer were adjusted to ensure that the center of each microstructure was aligned with the optical axis. The output spot is recorded without any input modulation to check the alignment and overall transmission stability.

[0039] As a specific example, the imaging test of the polarization cross-modulation diffraction network is as follows: All the trained and processed three-layer geometric phase metasurface units are fixed in the predetermined positions; The input terminal is illuminated by left-hand circularly polarized light, so that the light field passes through three layers of geometric phase metasurface units in sequence; The optical field undergoes phase modulation and polarization cross-conversion sequentially in each modulation layer and propagates in the free space between layers; At the detection surface behind the last layer, the left-hand circular polarization channel and the right-hand circular polarization channel are selected respectively by the polarization analysis and detection module 3; The imaging sensor in the analysis and detection module is used to record the output light intensity distribution under the corresponding polarization channel.

[0040] Based on the same inventive concept, this application also provides a holographic reconstruction method based on a polarization cross-modulation cascaded diffraction network, the method comprising: S201, The circularly polarized light generation module 1 is used to modulate the coherent light source into an input light field with different polarization state components; S202, the input light field loaded with the target hologram is incident on the polarization cross-modulation cascaded diffraction network 2; the polarization cross-modulation cascaded diffraction network 2 includes multiple polarization-sensitive modulation layers arranged sequentially along the light propagation direction; when the input light field propagates in the polarization cross-modulation cascaded diffraction network 2, each of the polarization-sensitive modulation layers performs phase modulation on different polarization state components in the input light field, and realizes polarization conversion and phase cross-modulation between different polarization state components; S202 specifically includes: S21, For the left-hand circularly polarized light component, after being modulated by any polarization-sensitive modulation layer, the output includes a first component that maintains the left-hand circularly polarized state and a second component that is converted to the right-hand circularly polarized state; S22, For a right-hand circularly polarized light component, after being modulated by any polarization-sensitive modulation layer, the output includes a third component that maintains the right-hand circular polarization state and a fourth component that is converted to a left-hand circular polarization state. The phase modulation process satisfies the polarization coupling relationship described by the Jones matrix.

[0041] S203, using the polarization analysis and detection module 3, the output light field modulated by the polarization cross-modulation cascaded diffraction network 2 is separated into different polarization state components, and the light intensity distribution corresponding to each polarization state component is detected respectively. S204 reconstructs a holographic image of the target based on the light intensity distribution corresponding to the detected different polarization state components.

[0042] The target holographic image includes a two-dimensional amplitude image, a phase image, or a complex amplitude image.

[0043] In an exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a holographic reconstruction method based on a polarization cross-modulation cascaded diffraction network.

[0044] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0045] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0046] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0047] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0048] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0049] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0050] In this application, all actions to acquire signals, information, or data are carried out in compliance with the relevant data protection laws and policies of the country where the location is situated, and with the authorization granted by the owner of the relevant device.

[0051] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0052] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A holographic reconstruction system based on a polarization cross-modulation cascaded diffraction network, characterized in that, include: A circularly polarized light generation module is set in the optical path of a coherent light source to modulate the coherent light source into an input light field with different polarization state components. A polarization cross-modulation cascaded diffraction network includes multiple polarization-sensitive modulation layers arranged sequentially along the light propagation direction; each polarization-sensitive modulation layer is used to perform phase modulation on different polarization state components in the input light field, and to realize polarization conversion and phase cross-modulation between different polarization state components. A polarization analysis and detection module is located at the output end of the polarization cross-modulation cascaded diffraction network, used to separate and detect the light intensity distribution corresponding to different polarization state components output after being modulated by the polarization cross-modulation cascaded diffraction network; The modulation parameters of the polarization cross-modulation cascaded diffraction network are obtained through end-to-end training and optimization, so that the input light field loaded with the hologram is propagated by polarization cross-modulation of the polarization cross-modulation cascaded diffraction network and the target holographic image is reconstructed on the detection surface of the polarization analysis and detection module.

2. The holographic reconstruction system of the polarization cross-modulation cascaded diffraction network according to claim 1, characterized in that, The circularly polarized light generation module includes a spatial filter, a collimating lens, a linear polarizer, and a quarter-wave plate, used to convert linearly polarized light into left-handed or right-handed circularly polarized light.

3. The holographic reconstruction system of the polarization cross-modulation cascaded diffraction network according to claim 1, characterized in that, The polarization-sensitive modulation layer includes multiple polarization-sensitive modulation units arranged in a two-dimensional array; each polarization-sensitive modulation unit is used to modulate the incident left-hand circularly polarized light component and outputs a first component that maintains the left-hand circular polarization state and a second component that is converted to the right-hand circular polarization state; each polarization-sensitive modulation unit is also used to modulate the incident right-hand circularly polarized light component and outputs a third component that maintains the right-hand circular polarization state and a fourth component that is converted to the left-hand circular polarization state.

4. The holographic reconstruction system of the polarization cross-modulation cascaded diffraction network according to claim 3, characterized in that, The polarization-sensitive modulation unit is a geometric phase metasurface unit; and it performs co-polarization phase modulation and cross-polarization conversion by controlling the orientation angle and / or birefringence phase delay of the subwavelength microstructure.

5. The holographic reconstruction system of the polarization cross-modulation cascaded diffraction network according to claim 1, characterized in that, The training and optimization process of the modulation parameters of the polarization cross-modulation cascaded diffraction network is as follows: Construct a training dataset; the training dataset includes multiple input hologram samples and corresponding target reconstructed images; An angular spectrum diffraction model for the propagation of the light field between adjacent polarization-sensitive modulation layers and a polarization cross-modulation transmission model for each polarization-sensitive modulation layer are established. The angular spectrum diffraction model uses the angular spectrum method to describe the free-space diffraction propagation between adjacent polarization-sensitive modulation layers. The polarization cross-modulation transmission model uses the Jones matrix to describe the modulation effect of each polarization-sensitive modulation layer on different polarization state components. A dual-channel fusion loss function is designed; the dual-channel fusion loss function is constructed based on the differences between the reconstructed image and the target image corresponding to the left-hand and right-hand polarization channels on the detector plane, respectively. The modulation parameters of each polarization-sensitive modulation layer are jointly optimized using the backpropagation algorithm until the dual-channel fusion loss function converges.

6. The holographic reconstruction system of the polarization cross-modulation cascaded diffraction network according to claim 5, characterized in that, The dual-channel fusion loss function specifically includes: Using formula Determine the dual-channel fusion loss function ; in, For left-hand circularly polarized light channels, the pixel loss function is... For right-hand circularly polarized light channels, the pixel loss function is... For the frequency domain loss function of the left-hand circularly polarized light channel, For the frequency domain loss function of the right-hand circularly polarized light channel, This is the diffraction efficiency constraint term for the left-hand circularly polarized light channel. This is the diffraction efficiency constraint term for the right-hand circularly polarized light channel. These are non-negative weighting coefficients used to adjust the weight of the frequency domain loss function in the total loss function. These are non-negative weighting coefficients used to adjust the weight of the diffraction efficiency constraint term in the total loss function.

7. The holographic reconstruction system of the polarization cross-modulation cascaded diffraction network according to claim 1, characterized in that, The polarization analysis and detection module includes: a linear polarizer, a quarter-wave plate, and an imaging sensor; The linear polarizer and quarter-wave plate are used to convert circular polarization to linear polarization and to select the polarization state. The imaging sensor is used to record the light intensity distribution at the detection surface.

8. A holographic reconstruction method based on a polarization cross-modulation cascaded diffraction network, applied to the holographic reconstruction system based on a polarization cross-modulation cascaded diffraction network as described in any one of claims 1-7, characterized in that, include: The coherent light source is modulated into an input light field with different polarization state components using a circularly polarized light generation module; An input light field loaded with a target hologram is incident on a polarization cross-modulation cascaded diffraction network; the polarization cross-modulation cascaded diffraction network includes multiple polarization-sensitive modulation layers arranged sequentially along the light propagation direction; when the input light field propagates in the polarization cross-modulation cascaded diffraction network, each of the polarization-sensitive modulation layers performs phase modulation on different polarization state components in the input light field, and realizes polarization conversion and phase cross-modulation between different polarization state components; The polarization analysis and detection module is used to separate the output light field modulated by the polarization cross-modulation cascaded diffraction network into different polarization state components, and the light intensity distribution corresponding to each polarization state component is detected respectively. Based on the light intensity distribution corresponding to the detected different polarization state components, a holographic image of the target is reconstructed.

9. The holographic reconstruction method of the polarization cross-modulation cascaded diffraction network according to claim 8, characterized in that, When the input light field propagates in the polarization cross-modulation cascaded diffraction network, each polarization-sensitive modulation layer performs phase modulation on different polarization state components in the input light field, and realizes polarization conversion and phase cross-modulation between different polarization state components, specifically including: For a left-hand circularly polarized light component, after being modulated by any polarization-sensitive modulation layer, the output includes a first component that maintains the left-hand circular polarization state and a second component that is converted to the right-hand circular polarization state; For a right-hand circularly polarized light component, after being modulated by any polarization-sensitive modulation layer, the output includes a third component that maintains the right-hand circular polarization state and a fourth component that is converted to a left-hand circular polarization state. The phase modulation process satisfies the polarization coupling relationship described by the Jones matrix.

10. The holographic reconstruction method of the polarization cross-modulation cascaded diffraction network according to claim 8, characterized in that, The target holographic image includes a two-dimensional amplitude image, a phase image, or a complex amplitude image.

Images