Polarization three-dimensional imaging method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application provide a polarization three-dimensional imaging method and device, electronic equipment and storage medium, which are applied to the field of optical imaging. The method comprises: performing optical imaging on a target object based on a metasurface, wherein the optical imaging result comprises a spatial distribution feature pre-encoded based on the metasurface and a depth feature modulated by a single helix point spread function; and performing three-dimensional topography reconstruction with average absolute depth on the target object based on the spatial distribution feature and the depth feature. The pre-design process of the metasurface comprises: determining an initial single helix phase and an initial diffraction grating phase in a Jones pupil function; performing phase distribution optimization on the initial diffraction grating phase to obtain a target diffraction grating phase satisfying a full Stokes polarization separation requirement, and performing phase distribution optimization on the initial single helix phase to obtain a target single helix phase; and constructing the metasurface based on the target diffraction grating phase and the target single helix phase. The method can improve the accuracy of polarization three-dimensional imaging.

Description

Technical Field

[0001] This application relates to the field of optical imaging, and more particularly to a polarization three-dimensional imaging method, apparatus, electronic device, and storage medium. Background Technology

[0002] Optoelectronic imaging technology is an important method and means for humans to record and perceive environmental information. However, while traditional optoelectronic imaging is a process of information acquisition and recording, it is also a process of information loss. Due to the limitations of the detector, it not only loses information such as phase, polarization, and spectrum in the light field, but also fails to effectively detect the three-dimensional shape information of the object, thus limiting the imaging effect. Polarization three-dimensional imaging, as a passive high-precision three-dimensional sensing technology based on the principles of physical optics, solves the three-dimensional information of the object's surface by constructing a functional relationship between the reflected light information of the object's surface and the surface shape.

[0003] Current polarization-based 3D imaging utilizes Fresnel's law to establish a functional relationship between the polarization characteristics of reflected light and the normal direction of the target surface to reconstruct the 3D topography. However, the polarization characteristics lead to ambiguity when reconstructing the normal vector, thus affecting the accuracy of polarization-based 3D imaging. Summary of the Invention

[0004] The main objective of this application is to propose a polarization three-dimensional imaging method, device, electronic device, and storage medium to improve the accuracy of polarization three-dimensional imaging.

[0005] To achieve the above objectives, this application proposes a polarization three-dimensional imaging method, comprising:

[0006] Optical imaging of a target object is performed based on a pre-designed metasurface. The optical imaging results include polarization-related spatial distribution features pre-encoded by the metasurface and polarization-independent depth features modulated by a single-helix point spread function.

[0007] Based on spatial distribution features and depth features, a three-dimensional shape reconstruction with average absolute depth is performed on the target object.

[0008] The pre-design process for metasurfaces includes:

[0009] Obtain the Jones pupil function of the initial metasurface, and determine the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function;

[0010] The initial diffraction grating phase is optimized by phase distribution to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation. The initial single-helix phase is also optimized by phase distribution to obtain the target single-helix phase with more concentrated energy.

[0011] A metasurface is constructed based on the phase of the target diffraction grating and the phase of the target single helix.

[0012] Phase distribution optimization of the initial diffraction grating phase yields the target diffraction grating phase that satisfies the requirements of full Stokes polarization separation, including:

[0013] The initial diffraction grating phase is spatially sampled to obtain the sampled phase matrix.

[0014] The phase distribution of the initial diffraction grating phase is adjusted based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation;

[0015] Phase distribution optimization of the initial single-helix phase yields a target single-helix phase with more concentrated energy, including:

[0016] The initial single-helix phase is subjected to frequency domain transformation to obtain frequency domain phase information;

[0017] By iteratively correcting the phase distribution of the initial single-helix phase based on frequency domain phase information, a more energy-concentrated target single-helix phase is obtained.

[0018] Optionally, in one embodiment, adjusting the phase distribution of the initial diffraction grating phase based on the phase matrix to obtain the target diffraction grating phase that satisfies the requirements of full Stokes polarization separation includes:

[0019] The propagation phase and geometric phase of the preset grating basic unit are determined based on the initial diffraction grating phase;

[0020] The propagation phase and geometric phase are optimized based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation.

[0021] Optionally, in one embodiment, reconstructing the three-dimensional shape of the target object with an average absolute depth based on spatial distribution features and depth features includes:

[0022] Determine the full Stokes polarization parameters of the target object based on its spatial distribution characteristics;

[0023] Three-dimensional topography reconstruction with average absolute depth is performed on the target object based on the full Stokes polarization parameters and depth features.

[0024] Optionally, in one embodiment, three-dimensional topography reconstruction of the target object with average absolute depth is performed based on the full Stokes polarization parameters and depth features, including:

[0025] A pre-trained deep neural network is invoked to predict the full Stokes polarization parameters, thereby obtaining the three-dimensional reconstructed morphology of the target object.

[0026] The depth features are measured based on a preset reference depth template to obtain the average absolute depth of the 3D reconstructed topography.

[0027] Optionally, in one implementation, a pre-trained deep neural network is invoked to predict the full Stokes polarization parameters to obtain the three-dimensional reconstructed shape of the target object, including:

[0028] Polarization optics modeling and inversion processing is performed on all Stokes polarization parameters to obtain physical prior information;

[0029] A pre-trained deep neural network is invoked to predict the physical prior information, thereby obtaining the three-dimensional reconstructed shape of the target object.

[0030] In another aspect, this application proposes a polarization three-dimensional imaging device, comprising:

[0031] An imaging unit is used to perform optical imaging of a target object based on a pre-designed metasurface. The optical imaging results include polarization-related spatial distribution features pre-encoded by the metasurface and polarization-independent depth features modulated by a single-helix point spread function.

[0032] The reconstruction unit is used to reconstruct the three-dimensional shape of a target object with an average absolute depth based on spatial distribution features and depth features.

[0033] The pre-design process for metasurfaces includes:

[0034] Obtain the Jones pupil function of the initial metasurface, and determine the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function;

[0035] The initial diffraction grating phase is optimized by phase distribution to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation. The initial single-helix phase is also optimized by phase distribution to obtain the target single-helix phase with more concentrated energy.

[0036] A metasurface is constructed based on the phase of the target diffraction grating and the phase of the target single helix.

[0037] Phase distribution optimization of the initial diffraction grating phase yields the target diffraction grating phase that satisfies the requirements of full Stokes polarization separation, including:

[0038] The initial diffraction grating phase is spatially sampled to obtain the sampled phase matrix.

[0039] The phase distribution of the initial diffraction grating phase is adjusted based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation;

[0040] Phase distribution optimization of the initial single-helix phase yields a target single-helix phase with more concentrated energy, including:

[0041] The initial single-helix phase is subjected to frequency domain transformation to obtain frequency domain phase information;

[0042] By iteratively correcting the phase distribution of the initial single-helix phase based on frequency domain phase information, a more energy-concentrated target single-helix phase is obtained.

[0043] Another aspect of this application provides an electronic device, comprising:

[0044] Memory, transceiver, processor, and bus system;

[0045] The memory is used to store programs;

[0046] The processor is used to execute programs in memory, including methods for performing the aspects mentioned above;

[0047] Bus systems are used to connect memory and processor to enable communication between them.

[0048] Another aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described above.

[0049] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0050] This method optimizes the phase configuration of the diffraction grating of the metasurface's Jones pupil function, enabling the metasurface to generate spatial distribution features corresponding to the full Stokes polarization state in the spatial domain. This allows information about different polarization states of the target object to be mapped into resolvable spatial location information in a single exposure. Simultaneously, phase optimization of the polarization-independent single-helix point spread function ensures sensitive detection of the depth dimension while maintaining spatial resolution. In the 3D topography reconstruction stage, since both spatial distribution features and depth features originate from the deterministic response of the same metasurface, they possess inherent registration consistency, allowing for direct synchronous reconstruction of 3D coordinates and polarization states. This avoids the problem of 3D coordinate and polarization state mismatch and improves the accuracy of polarization-based 3D imaging. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the architecture of the polarization three-dimensional imaging method provided in the embodiments of this application;

[0052] Figure 2 This is a schematic flowchart of the polarization three-dimensional imaging method provided in the embodiments of this application;

[0053] Figure 3 This is a schematic diagram of the pre-design process of the metasurface provided in the embodiments of this application;

[0054] Figure 4This is a schematic diagram of a three-dimensional topography reconstruction process provided in an embodiment of this application;

[0055] Figure 5 This is a schematic diagram of the metasurface design provided in an embodiment of this application;

[0056] Figure 6 This is a schematic diagram of the far-field light intensity distribution of the metasurface after the design provided in the embodiments of this application;

[0057] Figure 7 This is a schematic diagram of the absolute depth recovery process provided in the embodiments of this application;

[0058] Figure 8 This is a schematic diagram of the structure of the polarization three-dimensional imaging device provided in the embodiments of this application;

[0059] Figure 9 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0061] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0062] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0063] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0064] Furthermore, to better illustrate this application, numerous specific details are provided in the following detailed embodiments. Those skilled in the art should understand that this application can be implemented without certain specific details. In some instances, methods, means, components, and circuits well-known to those skilled in the art have not been described in detail in order to highlight the main points of this application.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0066] First, let's analyze some of the terms used in this application:

[0067] A metasurface is a two-dimensional optical material composed of an array of artificial microstructures at the subwavelength scale. Its thickness is typically much smaller than the operating wavelength, enabling precise manipulation of light wave amplitude, phase, polarization, and frequency at the interface. From a physical perspective, metasurfaces introduce abrupt optical responses at the subwavelength scale by designing nanoresonant units (such as nanopillars, nanopores, and nanodisks) with specific geometries, sizes, and orientations. These units utilize surface plasmon resonance, Mie resonance, or Pancharatnam-Berry phase effects. This interface phase manipulation method, independent of propagation phase accumulation, breaks the design constraints of traditional optical elements limited by refractive index and thickness, making it possible to achieve complex wavefront manipulation within ultrathin planes.

[0068] 3D imaging technology is based on the physical encoding and mathematical decoding mechanism of depth information carried by light fields during propagation in 3D space. From the perspective of geometric optics, light rays emitted from any point in space follow the law of rectilinear propagation, and their projection position on the imaging plane is determined by the 3D coordinates of that point and the geometric parameters of the imaging system. This projection relationship can be rigorously described by perspective transformation or pinhole camera models, thus establishing a dimensional compression mapping from the 3D world to a 2D image. The essence of 3D imaging is to solve the inverse problem of this mapping, recovering the lost depth dimension from degenerate 2D measurements.

[0069] Currently, 3D imaging technology can be divided into two categories based on imaging mode: active 3D imaging based on light source modulation and passive 3D imaging based on stereo vision.

[0070] Active 3D imaging based on light source modulation includes time-of-flight (TOF) imaging, lidar 3D imaging, and structured light 3D imaging. TOF utilizes the spatiotemporal characteristics of light waves to measure 3D space; however, this technology is susceptible to ambient light interference and limited by the temporal resolution of the signal system, resulting in low 3D imaging accuracy. Lidar 3D imaging uses laser ranging to acquire the distance information between the system and target micro-surface elements. Then, it reconstructs the 3D information of the target surface through mechanical scanning or beam deflection. Therefore, this 3D reconstruction method has poor real-time performance for large targets, and its complex mechanical structure leads to a large system size and high cost, hindering widespread adoption. Structured light 3D imaging projects a regularly coded fringe image onto the target surface. By decoding the captured contour image, a one-to-one correspondence is established between points on the camera plane and the projection plane. Combined with camera calibration parameters, the 3D surface is acquired. It has the advantage of high imaging accuracy; however, this system requires active light source illumination and is highly dependent on the illumination mode, resulting in poor resistance to ambient light interference. Furthermore, the accuracy of the coded fringe decreases with increasing imaging distance, severely affecting 3D imaging accuracy.

[0071] Passive 3D imaging based on stereo vision mainly includes binocular 3D imaging and 3D imaging based on light field cameras. Binocular 3D imaging technology, based on the principles of human vision, uses triangulation to construct the 3D surface contour information of an object. However, because its reconstruction accuracy is inversely proportional to the camera baseline length, it is difficult to achieve high-precision 3D surface information acquisition when imaging distant targets. Light field camera-based 3D imaging technology embeds a microlens array between the lens and the detector to determine the direction of light, thereby acquiring 3D information under passive conditions. However, similar to binocular 3D imaging, this technology is limited by the spacing between the microlens arrays, making long-distance 3D imaging impossible and resulting in lower imaging accuracy. Nevertheless, with the increasing demand for long-distance, high-precision, and high-dimensional target information in fields such as security monitoring, deep space exploration, and target detection and recognition, achieving long-distance, non-destructive, and high-precision 3D imaging through in-depth mining and interpretation of the multidimensional physical information of the light field has become the main development direction in this field.

[0072] The polarization characteristics of light waves have a functional relationship with the normal to the target surface, directly reflecting the target's morphological characteristics. Therefore, establishing the functional relationship between the polarization characteristics of reflected light and the surface morphology can provide a solution for achieving high-precision, simple, and non-contact 3D reconstruction technology. The core of polarization-based 3D imaging is to use Fresnel's law to establish a functional relationship between the polarization characteristics of reflected light and the direction of the normal to the target surface to reconstruct the 3D morphology. However, the polarization characteristics cause ambiguity when reconstructing the normal vector, thus affecting the quality of the 3D morphology reconstruction.

[0073] Based on this, embodiments of this application provide a polarization three-dimensional imaging method that can solve the above-mentioned technical problems.

[0074] System architecture and scenario description used in the embodiments of this application:

[0075] Figure 1 This is a system architecture diagram of a polarization three-dimensional imaging method according to an embodiment of this application. It includes a terminal 140, an Internet 130, a gateway 120, a server 110, etc.

[0076] Terminal 140 includes various forms of devices with display screens, such as desktop computers, laptops, PDAs (personal digital assistants), mobile phones, in-vehicle terminals, home theater terminals, dedicated terminals, intelligent voice interaction devices, smart home appliances, or aircraft. Furthermore, it can be a single device or a collection of multiple devices. Terminal 140 can communicate with the Internet 130 via wired or wireless means to exchange data.

[0077] Server 110 refers to a computer system that can provide certain services to terminal 140. Compared to ordinary terminal 140, server 110 has higher requirements in terms of stability, security, and performance. Server 110 can be a single high-performance computer in a network platform, a cluster of multiple high-performance computers, a portion of a single high-performance computer (e.g., a virtual machine), or a combination of portions of multiple high-performance computers (e.g., virtual machines).

[0078] Gateway 120, also known as an internetwork connector or protocol converter, is a computer system or device that acts as a translator, enabling network interconnection at the transport layer. It bridges the gap between two systems using different communication protocols, data formats, languages, or even completely different architectures. Gateways can also provide filtering and security functions. Messages sent from terminal 140 to server 110 are forwarded to the corresponding server 110 via gateway 120. Messages sent from server 110 to terminal 140 are also forwarded to the corresponding terminal 140 via gateway 120.

[0079] The polarization three-dimensional imaging method provided in this application embodiment can be implemented alone in terminal 140, alone in server 110, or partially in terminal 140 and partially in server 110.

[0080] When the polarization-based three-dimensional imaging method provided in this application embodiment is implemented alone in terminal 140, terminal 140 performs optical imaging of the target object based on a pre-designed metasurface. The optical imaging result includes polarization-related spatial distribution features pre-encoded by the metasurface and depth features modulated by a polarization-independent single-helix point spread function. Terminal 140 reconstructs the three-dimensional shape of the target object with an average absolute depth based on the spatial distribution features and depth features. The pre-design process of the metasurface includes: obtaining the Jones pupil function of the initial metasurface and determining the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function; optimizing the phase distribution of the initial diffraction grating phase to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation, and optimizing the phase distribution of the initial single-helix phase to obtain... The process involves: obtaining a target single-helix phase with more concentrated energy; constructing a metasurface based on the target diffraction grating phase and the target single-helix phase; optimizing the phase distribution of the initial diffraction grating phase to obtain a target diffraction grating phase that meets the requirements of full Stokes polarization separation, including: sampling the initial diffraction grating phase in the spatial domain to obtain the sampled phase matrix; adjusting the phase distribution of the initial diffraction grating phase based on the phase matrix to obtain a target diffraction grating phase that meets the requirements of full Stokes polarization separation; and optimizing the phase distribution of the initial single-helix phase to obtain a target single-helix phase with more concentrated energy, including: performing frequency domain transformation on the initial single-helix phase to obtain frequency domain phase information; and iteratively correcting the phase distribution of the initial single-helix phase based on the frequency domain phase information to obtain a target single-helix phase with more concentrated energy.

[0081] The polarization three-dimensional imaging method provided in the embodiments of this application will be described below with reference to the accompanying drawings. The execution subject of the polarization three-dimensional imaging method described below is a terminal device, which can be implemented by the terminal device by running the various computer programs mentioned above. Of course, based on the understanding of the following text, it is not difficult to see that the polarization three-dimensional imaging method provided in the embodiments of this application can also be implemented by the terminal device and the server in collaboration.

[0082] Please see Figure 2 ,like Figure 2 The diagram shown is a flowchart of a polarization three-dimensional imaging method provided in this embodiment. The method includes:

[0083] Step 201: Perform optical imaging on the target object based on a pre-designed metasurface. The optical imaging results include polarization-related spatial distribution features pre-encoded based on the metasurface and depth features modulated by a polarization-independent single-helix point spread function.

[0084] Metasurfaces can refer to a two-dimensional artificial electromagnetic material composed of an array of nanostructures at the subwavelength scale. They can flexibly control the amplitude, phase, polarization and other properties of light at the subwavelength scale. Here, metasurfaces specifically refer to planar optical devices that have been pre-designed and integrated with specific optical functions. They simultaneously undertake the dual tasks of polarization control and depth information encoding.

[0085] The target object can refer to the object that needs to be optically imaged, and optical imaging can refer to the process of acquiring an image of the object using an optical system.

[0086] Spatial distribution characteristics refer to the spatial distribution pattern of the electric field vibration direction of light waves. Polarization is an important characteristic of light as a transverse wave, and different polarization states carry different spatial information. In this step, light with different polarization states can be separated into different spatial locations, so that the light intensity distribution on the imaging plane has a one-to-one correspondence with the polarization state of the incident light, thereby converting the polarization information into a directly detectable spatial intensity distribution.

[0087] A single-helix point spread function (SFD) can refer to a special form of point spread function that exhibits a single-helix shape. The SFD describes the response of an optical system to a point light source, and the single-helix shape means that the system's impulse response is distributed in a helical pattern on the focal plane. Here, the depth feature refers to the optical encoding related to the object's depth position introduced by the single-helix phase component in the metasurface. When the object is at different depths, the rotation angle of the single-helix point spread function changes, and this rotation angle has a definite mathematical relationship with the depth, thus encoding depth information into the image.

[0088] Step 202: Reconstruct the three-dimensional shape of the target object with average absolute depth based on spatial distribution features and depth features.

[0089] Three-dimensional topography refers to the geometric shape and surface contour information of an object in three-dimensional space, including depth coordinates and lateral coordinates. The three-dimensional topography reconstruction here is a computational inversion process. It refers to using the coded image corresponding to the optical imaging result obtained in step 201, and by analyzing the polarization spatial distribution features and depth features contained within it, the three-dimensional coordinates of each point on the object's surface are deduced, ultimately restoring the three-dimensional topography with an average perceived depth. Specifically, the polarization characteristics of the object's surface can be analyzed using polarization-related spatial distribution features to calculate the relative distances between points on the object's surface, thus reconstructing the three-dimensional topography. Furthermore, by analyzing the depth features decoded from the rotation angle of the single-helix point spread function, the average absolute depth of each point in the three-dimensional topography can be calculated. This allows for the direct synchronous reconstruction of three-dimensional coordinates and polarization states, avoiding the problem of mismatch between three-dimensional coordinates and polarization states, and improving the accuracy of polarization-based three-dimensional imaging.

[0090] In one implementation, reconstructing the three-dimensional shape of a target object with an average absolute depth based on spatial distribution features and depth features includes:

[0091] Determine the full Stokes polarization parameters of the target object based on its spatial distribution characteristics;

[0092] Three-dimensional topography reconstruction with average absolute depth is performed on the target object based on the full Stokes polarization parameters and depth features.

[0093] In this embodiment, the total Stokes polarization parameter specifically refers to the complete polarization characteristics of light reflected or transmitted at each point on the surface of the target object. It not only includes light intensity information, but also all polarization attributes such as polarization direction, degree of polarization, and rotation direction. It is the most complete mathematical representation of the polarization state of light.

[0094] Spatial distribution characteristics here refer to the specific intensity distribution pattern formed on the imaging plane after the metasurface separates different polarization states through its diffraction grating phase components. Since the metasurface achieves full Stokes polarization separation, different spatial regions on the imaging plane correspond to different Stokes component responses. Therefore, by analyzing the light intensity distribution in these regions, the specific values ​​of each Stokes parameter of the incident light can be deduced. The full Stokes polarization parameters can then be jointly processed with depth features: depth features provide longitudinal position information, while spatial distribution characteristics (after being converted to Stokes parameters) provide lateral position constraints and surface polarization properties. The combination of these two features forms a complete three-dimensional spatial coordinate system, ultimately yielding a three-dimensional morphology result containing complete spatial coordinates and polarization characteristics.

[0095] By fusing multimodal information to reconstruct 3D topography, the accuracy and richness of the reconstruction are enhanced.

[0096] In one implementation, a three-dimensional topography reconstruction with average absolute depth is performed on a target object based on full Stokes polarization parameters and depth features, including:

[0097] A pre-trained deep neural network is invoked to predict the full Stokes polarization parameters, thereby obtaining the three-dimensional reconstructed morphology of the target object.

[0098] The depth features are measured based on a preset reference depth template to obtain the average absolute depth of the 3D reconstructed topography.

[0099] In this embodiment, the pre-trained deep neural network can refer to a neural network model specifically designed for the physical characteristics of the metasurface imaging system and pre-trained. This model learns the mapping relationship between the full Stokes polarization parameters and the real three-dimensional shape in a large amount of sample data. The deep neural network receives the full Stokes polarization parameters as input, calculates through forward propagation, and directly outputs the three-dimensional shape estimate of the target object, which can be in the form of a depth map, point cloud, or voxel representation.

[0100] A preset reference depth template can refer to a standard pattern or calibration plate with a specific depth distribution, pre-determined during the design phase of a metasurface imaging system. By comparing the actual acquired depth features with the known depth of the reference depth template, and through methods such as lookup table fitting, multinomial regression, or neural network mapping, a conversion relationship from relative rotation angle to absolute depth value is established. This yields the precise distances, expressed in physical units such as millimeters and centimeters, from each pixel of the target object to the main plane or reference plane of the imaging system. This allows the determination of the true depth position of the entire 3D model in space, resulting in a 3D reconstructed morphology with an average absolute depth.

[0101] By combining polarization network topography prediction and template calibration absolute depth, a 3D topography reconstruction with average absolute depth quantization calibration is achieved.

[0102] In one implementation, a pre-trained deep neural network is invoked to predict the full Stokes polarization parameters to obtain the three-dimensional reconstructed shape of the target object, including:

[0103] Polarization optics modeling and inversion processing is performed on all Stokes polarization parameters to obtain physical prior information;

[0104] A pre-trained deep neural network is invoked to predict the physical prior information, thereby obtaining the three-dimensional reconstructed shape of the target object.

[0105] In this implementation, polarization optics modeling and inversion processing refers to the process of reversing and calculating the collected polarization parameters by establishing a physical law model of polarized light propagation, reflection, and scattering, thereby restoring the underlying physical properties of the object. Here, it can involve establishing a polarization optics model adapted to a pre-designed metasurface imaging system for the full Stokes polarization parameters, and using inversion calculations to remove redundant interference signals during light field propagation, thus restoring the physical information directly related to the three-dimensional morphology of the target object—that is, obtaining prior physical information. Here, Fresnel's law can be used to clarify the intrinsic relationship between the polarization state of the reflected light (i.e., each component of the full Stokes polarization parameters) and the surface normal and refractive index of the target object.

[0106] The pre-trained deep neural network can predict the physical prior information to obtain the relative three-dimensional contour of the target object, restore the undulations, curvature and micro-morphological details of the object's surface, and thus obtain the three-dimensional reconstructed shape.

[0107] By performing polarization optics modeling and inversion processing before inputting the full Stokes polarization parameters into the deep neural network, the original full Stokes polarization parameters are transformed into more targeted physical prior information that is closer to the true shape of the object, thus reducing the processing complexity of the deep neural network.

[0108] The schematic diagram of the pre-design process for metasurfaces can be found by referring to... Figure 3 As shown:

[0109] Step 203: Obtain the Jones pupil function of the initial metasurface, and determine the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function.

[0110] An initial metasurface can refer to a metasurface without corresponding parameters set. The Jones pupil function can refer to the combination of Jones matrix calculation and the concept of pupil function. The Jones matrix is ​​a matrix that describes the transformation of the polarization state of light by a polarization optical element, while the pupil function describes the wavefront phase and amplitude distribution at the entrance pupil of the optical system, describing the modulation effect of the metasurface on light waves with different polarization components at the entrance pupil surface.

[0111] A single-helix phase can refer to a helix phase distribution in polar coordinates where the phase changes linearly with the angle and the period of change is 2π. Its mathematical form usually includes a term proportional to the azimuth angle. This phase distribution will produce a vortex beam with orbital angular momentum. The initial single-helix phase here can refer to the single-helix phase in a metasurface without set parameters.

[0112] The diffraction grating phase refers to the phase delay introduced by diffractive optical elements, used to control the propagation direction and convergence characteristics of light. Here, the initial diffraction grating phase refers to the diffraction grating phase of the metasurface without any parameters set.

[0113] Step 204: Optimize the phase distribution of the initial diffraction grating phase to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation, and optimize the phase distribution of the initial single-helix phase to obtain the target single-helix phase with more concentrated energy.

[0114] Phase distribution optimization can refer to the search process in the phase function space using numerical optimization or iterative algorithms. Here, it can refer to iteratively adjusting the spatial distribution of the initial diffraction grating phase and the initial single-helix phase using optimization algorithms (such as gradient descent, genetic algorithms, or physical model-based inversion algorithms).

[0115] Stokes parameters can be a set of four physical quantities (S0, S1, S2, S3) describing the polarization state of light, where S0 represents the total light intensity, and S1, S2, and S3 describe the components of horizontal / vertical linear polarization, ±45° linear polarization, and circular polarization, respectively, thus fully characterizing the polarization state of light. Here, "full Stokes polarization separation" refers to the ability of a metasurface to simultaneously separate the four Stokes parameters of incident light into four independent spatial channels, allowing different regions of the detector to correspond to different Stokes components, thereby obtaining complete polarization state information of light through a single measurement.

[0116] The phase of the target diffraction grating can be an optimized phase distribution that produces four separate focal spots on the imaging plane, with the intensity of each focal spot proportional to a specific Stokes parameter; the phase of the target single spiral can be an optimized phase distribution with a rotation point spread function and a good linear relationship between the rotation angle and the depth.

[0117] Among these steps, the phase distribution of the initial diffraction grating is optimized to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation, including:

[0118] The initial diffraction grating phase is spatially sampled to obtain the sampled phase matrix.

[0119] The phase distribution of the initial diffraction grating phase is adjusted based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation.

[0120] In this embodiment, the spatial domain can refer to the two-dimensional planar space where the metasurface is located, i.e., the physical layout region of the metasurface structure array, which can be described by a transverse coordinate system in nanometers or micrometers. Spatial domain matrix sampling can refer to sampling the continuous or high-resolution phase distribution function of the initial diffraction grating phase on the discrete structural units of the metasurface to form a phase matrix corresponding to the arrangement of the metasurface units, where each matrix element represents the target phase delay value at the location of a metasurface unit.

[0121] Phase distribution adjustment can refer to systematically modifying the phase values ​​in the sampled phase matrix through an iterative optimization algorithm, so that the diffraction optical function defined by the phase matrix can meet the performance index of full Stokes polarization separation, thereby obtaining the optimized target diffraction grating phase.

[0122] By searching for the optimal solution in the phase distribution design space (the dimension corresponding to the number of metasurface units), a high-performance phase distribution mode that meets the requirements of full Stokes polarization separation is obtained, so as to realize monolithic metasurface integrated polarization imaging and improve the efficiency of three-dimensional imaging.

[0123] In one embodiment, adjusting the phase distribution of the initial diffraction grating phase based on the phase matrix to obtain the target diffraction grating phase that satisfies the requirements of full Stokes polarization separation includes:

[0124] The propagation phase and geometric phase of the preset grating basic unit are determined based on the initial diffraction grating phase;

[0125] The propagation phase and geometric phase are optimized based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation.

[0126] In this embodiment, the basic grating unit can be the structural basis in diffraction optics, referring to micro / nano structural units with specific shapes and optical functions within the grating period, which form a complete grating device through periodic or non-periodic arrangement. Here, the pre-designed basic grating unit can refer to a subwavelength-scale nanostructural unit pre-designed to construct the phase of a metasurface diffraction grating.

[0127] The propagation phase can refer to the phase delay caused by the accumulation of optical path when a light wave propagates in a medium. Here, the phase delay component can be controlled by changing the geometric dimensions (such as height and width) of the nanostructure unit.

[0128] Geometric phase can be a topological phase that describes the evolution of the light field along a closed path in the parameter space. Here, the phase modulation of the conjugate component of circularly polarized light can be achieved by rotating the orientation angle of the nanostructure unit.

[0129] Using the phase matrix as the target response, the size parameters (controlling the propagation phase) and rotation angle (controlling the geometric phase) of the nanostructure units are optimized simultaneously to make the total phase distribution generated by the unit combination match the target phase matrix and meet the performance requirements of full Stokes polarization separation, thereby obtaining the target diffraction grating phase.

[0130] By jointly optimizing the propagation phase and geometric phase, and making full use of the advantages of the two phase modulation mechanisms, the propagation phase provides broadband and efficient phase modulation, while the geometric phase provides polarization-sensitive phase modulation. This allows us to approximate the ideal phase distribution and polarization response within a limited structural parameter space.

[0131] Among these steps, phase distribution optimization of the initial single-helix phase is performed to obtain a target single-helix phase with more concentrated energy, including:

[0132] The initial single-helix phase is subjected to frequency domain transformation to obtain frequency domain phase information;

[0133] By iteratively correcting the phase distribution of the initial single-helix phase based on frequency domain phase information, a more energy-concentrated target single-helix phase is obtained.

[0134] In this implementation, frequency domain transformation refers to converting a function in the spatial or time domain to a frequency domain representation through orthogonal transformations such as Fourier transform and cosine transform, revealing the distribution characteristics of the signal at different frequency components. Here, frequency domain transformation processing can refer to performing a two-dimensional discrete Fourier transform on the initial single-helix phase spatial distribution function, converting it from a spatial phase value distribution to a spatial frequency spectrum distribution in the frequency domain, i.e., frequency domain phase information.

[0135] Iterative correction can refer to the process of gradually approximating the target solution through multiple iterations, with each iteration adjusting based on the deviation between the current solution and the target. Here, iterative correction can refer to repeatedly optimizing and adjusting the spatial distribution of the initial single-helix phase, guided by frequency domain phase information, so that the final phase distribution maintains helical symmetry while optimizing its spatial frequency components, thereby obtaining a more energy-concentrated target single-helix phase.

[0136] Through frequency domain analysis and iterative correction, while maintaining the basic topological structure of the spiral phase, its spatial distribution is finely controlled to suppress unfavorable frequency components and enhance favorable frequency responses, thereby optimizing the quality of the point spread function.

[0137] Step 205: Construct a metasurface based on the target diffraction grating phase and the target single-helix phase.

[0138] Based on the spatial distribution of the phase of the target diffraction grating and the phase of the target single helix, specific metasurface nanostructures (such as nanopillars, nanopores, nanofins, etc.) are designed, and the geometric parameters (height, width, length, rotation angle, etc.) of the structure at each location are determined so that the structure can generate the required phase delay at the target working wavelength. Then, these nanostructure arrays are fabricated on the substrate material through semiconductor micro-nano fabrication processes (such as electron beam lithography, reactive ion etching, etc.) to finally form a metasurface with predetermined optical functions.

[0139] In one example, a flowchart of 3D topography reconstruction can be shown as follows: Figure 4 As shown, firstly, the Jones pupil function of the initial metasurface is obtained, which function... Represented as:

[0140]

[0141] This function includes the polarization-dependent Jones matrix. and polarization-independent amplitude and phase By utilizing polarization-dependent and polarization-independent terms in metasurfaces, polarization control of vector gratings and wavefront shaping properties of single-spiral point spread functions (SH-PSFs) can be achieved. Furthermore, by utilizing propagation phase and geometric phase, Jones pupil functions can be directly modulated to generate metasurfaces.

[0142] Specifically, to improve the polarization measurement efficiency of metasurfaces and the depth measurement accuracy of SH-PSF, the metasurfaces can be optimized, among which... It consists of two components: the diffraction grating phase and the single-helix phase.

[0143] Secondly, phase optimization is performed on the diffraction grating phase and single-helix phase in the Jones pupil function. Spatial matrix sampling of the diffraction grating phase is conducted using the Fourier matrix method to optimize the phase of the diffraction metasurface, achieving full Stokes polarization separation in the first-order diffraction order. This allows the designed metasurface to measure the full Stokes polarization state. The optimized phase consists of the propagation phase and geometric phase distribution generated by periodically arranged titanium dioxide nanopillars. To improve optimization efficiency, an 11×11 grating unit is selected for optimization. The diffraction grating is designed by optimizing two orthogonal propagation phases and one geometric phase. The propagation phase can be controlled by changing the length and width of the unit, while the geometric phase can be controlled by changing the rotation angle of the unit.

[0144] During the optimization process, the energy distribution and polarization contrast in the first-order diffraction order are kept uniform. , where I and Let S be the energy of the diffraction order when the target's polarization state and its orthogonal polarization state are incident, respectively. The total energy in the first-order diffraction order is optimized under the condition of being close to 1. The propagation phase and geometric phase are optimized through a multi-parameter optimization algorithm. The optimized phase distribution enables the detection of the designed polarization state in a specific diffraction order. Subsequently, full Stokes polarization imaging is achieved using S=A*I, where S is the detected full Stokes parameters, A is a 4×4 matrix composed of the Stokes parameters of the designed polarization state, and I is the energy of the corresponding diffraction order.

[0145] Single-helix spread function engineering can achieve absolute depth measurement by utilizing the rotation angle of the focal point. The focal point generated on the focal plane rotates with changes in object distance, and both the rotation angle and depth are monotonic. Theoretically, the single-helix spread function can achieve 360-degree measurements, giving it a wider measurement range. The fundamental phase of the single-helix spread function can be expressed as...

[0146]

[0147] Where u represents the normalized radial coordinate, L represents the azimuth angle located within the plane of the incident pupil, while L and To adjust the design parameters and improve focusing efficiency at different depths, the single-spiral point spread function is optimized using frequency domain transformation and iterative correction. This maximizes the intensity within the main lobe of the rotating PSF across the entire 360° range. The optimization of the single-spiral point spread function is achieved by iterating the focusing intensity values ​​at different depths.

[0148] Subsequently, a metasurface is constructed based on the optimized diffraction grating phase and single-helix phase. The metasurface design is achieved by superimposing the optimized SH-PSF phase and diffraction grating phase; the designed metasurface can achieve full Stokes polarization imaging while simultaneously combining SH-PSF for absolute depth measurement. For an example, please refer to [link to example]. Figure 5 The schematic diagram of the metasurface design shown illustrates a convolution operation (or phase superposition in the frequency domain) between a vector grating 501 (a spatial intensity distribution map, presenting a cross-shaped pattern, with the horizontal and vertical axes representing normalized spatial positions, achieving full Stokes polarization separation) and an optimized single-spiral point spread function 502 (a spiral stripe circular pattern, with the horizontal and vertical axes representing the spatial positions of the incident pupil plane, achieving depth information encoding). This yields the final encoded metasurface 503 (four separate focal points carrying depth encoding, with the horizontal and vertical axes representing the spatial positions of the focal plane). The optimization of the vector grating 501 is primarily achieved through two propagation phases (faix 1 and faiy 2) and one geometric phase (theta 1), with the horizontal and vertical axes representing the spatial arrangement of the grating's basic units. The optimization of the single-spiral point spread function 502 is primarily achieved through propagation phases (faix 3 and faiy 4) with the horizontal and vertical axes representing normalized radial coordinates. This involves constructing a spiral phase distribution by changing the length and width of the structural units, ultimately convolving to obtain the two propagation phases (faix 5 and faiy 4) corresponding to the encoded metasurface 503. 6) and a geometric phase (theta 2), where the x and y axes represent the complete metasurface aperture. The propagation phase simultaneously includes periodic modulation and helical variation, and the geometric phase integrates polarization separation and helical modulation. The far-field light intensity distribution of the designed metasurface can be referenced. Figure 6 As shown, the four separate diffraction-order subfoci correspond to the four polarization channels (horizontal, vertical, 45-degree linear polarization, and circular polarization) after full Stokes polarization separation. Each focal point exhibits the characteristic morphology of a single-helix point spread function (SH-PSF), i.e., a helical intensity distribution rather than a traditional Airy disk, indicating that the depth coding function has been successfully integrated. The spatial separation of the four focal points confirms the effectiveness of polarization separation, while the helical morphology of each focal point verifies the realization of depth-sensitive rotation characteristics.

[0149] Next, the constructed metasurface was validated. Polarization characterization and depth-sensitive SH-PSF characterization were performed on the designed metasurface to verify that the monolithic metasurface simultaneously achieved SH-PSF encoding and polarization separation. Specifically, incident light of known polarization states (such as horizontal linear polarization, vertical linear polarization, 45-degree linear polarization, circular polarization, etc.) was irradiated onto the metasurface, and its energy distribution in each diffraction order was measured to verify whether different polarization states were correctly separated to the predetermined spatial positions, and whether the separation contrast, efficiency, and uniformity met the design specifications. A point source or collimated beam was placed at different axial depth positions of the metasurface imaging system, and its point spread function shape and orientation on the focal plane were measured. The monotonic relationship between rotation angle and depth, linearity, dynamic range (whether it reaches 360 degrees), main lobe concentration, and side lobe suppression level were analyzed.

[0150] Next, optical imaging of the target object is performed based on the validated metasurface. This optical imaging includes pre-encoded polarization-related spatial distribution features and depth features modulated by a single-helix point spread function based on the metasurface. Specifically, after the light reflected or transmitted from the target object passes through the metasurface, the target diffraction grating phase component in the metasurface separates different polarization states to different spatial positions on the imaging plane, forming polarization-related spatial distribution features. Simultaneously, the target single-helix phase component in the metasurface performs helical phase modulation on the light field, generating a point spread function that rotates with the object's depth, forming depth features. The detector receives the light field modulated by the metasurface, obtaining a raw image that simultaneously contains polarization spatial distribution information and depth-encoded information.

[0151] Finally, the neural network is used to reconstruct the 3D topography with average absolute depth from the optical imaging image. The network architecture consists of 5 downsampling encoders, 5 upsampling decoders, and corresponding skip connection residual blocks. After upsampling, the residual blocks are composed of two 3×3 convolutional layers. First, the network aligns and segments the original image, then performs bilinear interpolation to improve resolution, resulting in four sub-images with different polarization states as the initial input to the network. These four sub-images correspond to the four polarization channel responses after full Stokes polarization separation achieved by the metasurface, containing the polarization spatial distribution and depth features of the object.

[0152] The full Stokes polarization parameters can be calculated based on four sub-images, and combined with the physical definitions of different reflection types, nine-channel normal priors for different types of reflections can be obtained. The nine-channel normal priors are spatially scaled and introduced as conditional inputs into the normalization layers of the five encoder stages of the neural network. The neural network can recover the surface normal vectors of the object, and thus recover the three-dimensional shape of the object.

[0153] The device rotates the metasurface by 180 degrees to obtain a reference image; the original image and the corresponding reference image are segmented into the same region to obtain the target region; then edge detection is performed using this pair of images, and the translation vector is calculated through template matching; finally, the average absolute depth of each pixel of the three-dimensional shape is calculated using the established functional relationship between the angle and depth of the translation vector.

[0154] In one example, a schematic diagram of the absolute depth recovery process can be found here. Figure 7 As shown, the original image is taken directly from the metasurface towards the target object, while the reference image is taken after the metasurface has been rotated 180 degrees. Both images contain SH-PSF encoded focal points of four polarization channels. After segmenting the two images into their respective regions, edge detection and template matching algorithms are used to calculate the translation vector. The angle of this translation vector reflects the relative rotation difference of the SH-PSF focal points in the two images. Using the established functional relationship between the translation vector angle and depth, the relative rotation is converted into an absolute depth value, yielding a depth reference value within the range of 10cm to 80cm.

[0155] The polarization three-dimensional imaging method has been described above. The apparatus for performing this method will be described below.

[0156] Please see Figure 8 The diagram shows a structural schematic of a polarization three-dimensional imaging device. The device 800 includes:

[0157] Imaging unit 801 is used to perform optical imaging of a target object based on a pre-designed metasurface. The optical imaging results include polarization-related spatial distribution features pre-encoded based on the metasurface and polarization-independent depth features modulated by a single-helix point spread function.

[0158] Reconstruction unit 802 is used to reconstruct the three-dimensional shape of the target object with average absolute depth based on spatial distribution features and depth features.

[0159] The pre-design process for metasurfaces includes:

[0160] Obtain the Jones pupil function of the initial metasurface, and determine the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function;

[0161] The initial diffraction grating phase is optimized by phase distribution to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation. The initial single-helix phase is also optimized by phase distribution to obtain the target single-helix phase with more concentrated energy.

[0162] A metasurface is constructed based on the phase of the target diffraction grating and the phase of the target single helix.

[0163] Phase distribution optimization of the initial diffraction grating phase yields the target diffraction grating phase that satisfies the requirements of full Stokes polarization separation, including:

[0164] The initial diffraction grating phase is spatially sampled to obtain the sampled phase matrix.

[0165] The phase distribution of the initial diffraction grating phase is adjusted based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation;

[0166] Phase distribution optimization of the initial single-helix phase yields a target single-helix phase with more concentrated energy, including:

[0167] The initial single-helix phase is subjected to frequency domain transformation to obtain frequency domain phase information;

[0168] By iteratively correcting the phase distribution of the initial single-helix phase based on frequency domain phase information, a more energy-concentrated target single-helix phase is obtained.

[0169] Optionally, in one embodiment, the imaging unit 801 is specifically used for:

[0170] The propagation phase and geometric phase of the preset grating basic unit are determined based on the initial diffraction grating phase;

[0171] The propagation phase and geometric phase are optimized based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation.

[0172] Optionally, in one embodiment, the reconstruction unit 802 is specifically used for:

[0173] Determine the full Stokes polarization parameters of the target object based on its spatial distribution characteristics;

[0174] Three-dimensional topography reconstruction with average absolute depth is performed on the target object based on the full Stokes polarization parameters and depth features.

[0175] Optionally, in one embodiment, the reconstruction unit 802 is specifically used for:

[0176] A pre-trained deep neural network is invoked to predict the full Stokes polarization parameters, thereby obtaining the three-dimensional reconstructed morphology of the target object.

[0177] The depth features are measured based on a preset reference depth template to obtain the average absolute depth of the 3D reconstructed topography.

[0178] Optionally, in one embodiment, the reconstruction unit 802 is specifically used for:

[0179] Polarization optics modeling and inversion processing is performed on all Stokes polarization parameters to obtain physical prior information;

[0180] A pre-trained deep neural network is invoked to predict the physical prior information, thereby obtaining the three-dimensional reconstructed shape of the target object.

[0181] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described polarization three-dimensional imaging method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0182] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0183] The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0184] The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the polarization three-dimensional imaging method of the embodiments of this application.

[0185] The input / output interface 903 is used to implement information input and output;

[0186] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0187] Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904);

[0188] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0189] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described polarization three-dimensional imaging method.

[0190] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0191] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0192] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0193] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0194] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0195] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0196] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0197] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0198] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0199] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0200] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0201] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A polarization three-dimensional imaging method, characterized in that, include: Optical imaging of a target object is performed based on a pre-designed metasurface. The optical imaging results include polarization-related spatial distribution features pre-encoded based on the metasurface and polarization-independent depth features modulated by a single-helix point spread function. Based on the spatial distribution features and the depth features, a three-dimensional topography reconstruction with an average absolute depth is performed on the target object; The pre-design process for the metasurface includes: Obtain the Jones pupil function of the initial metasurface, and determine the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function; The initial diffraction grating phase is optimized in phase distribution to obtain a target diffraction grating phase that meets the requirements of full Stokes polarization separation. The initial single-helix phase is also optimized in phase distribution to obtain a target single-helix phase with more concentrated energy. The metasurface is constructed based on the target diffraction grating phase and the target single-helix phase; The step of optimizing the phase distribution of the initial diffraction grating phase to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation includes: The initial diffraction grating phase is spatially sampled to obtain the sampled phase matrix. The phase distribution of the initial diffraction grating phase is adjusted based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation; The step of optimizing the phase distribution of the initial single-helix phase to obtain a target single-helix phase with more concentrated energy includes: The initial single-helix phase is subjected to frequency domain transformation to obtain frequency domain phase information; Based on the frequency domain phase information, the phase distribution of the initial single-helix phase is iteratively corrected to obtain a target single-helix phase with more concentrated energy.

2. The method according to claim 1, characterized in that, The step of adjusting the phase distribution of the initial diffraction grating phase based on the phase matrix to obtain the target diffraction grating phase that satisfies the requirements of full Stokes polarization separation includes: The propagation phase and geometric phase of the preset grating basic unit are determined based on the initial diffraction grating phase; The propagation phase and the geometric phase are optimized based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation.

3. The method according to claim 1, characterized in that, The method of reconstructing the three-dimensional shape of the target object with an average absolute depth based on the spatial distribution features and the depth features includes: The full Stokes polarization parameters of the target object are determined based on the spatial distribution characteristics. Based on the full Stokes polarization parameters and the depth features, a three-dimensional topography reconstruction with average absolute depth is performed on the target object.

4. The method according to claim 3, characterized in that, The three-dimensional shape reconstruction of the target object with average absolute depth based on the full Stokes polarization parameters and the depth features includes: A pre-trained deep neural network is invoked to predict the full Stokes polarization parameters, thereby obtaining the three-dimensional reconstructed morphology of the target object; The depth features are measured based on a preset reference depth template to obtain the average absolute depth of the three-dimensional reconstructed topography.

5. The method according to claim 4, characterized in that, The step of calling a pre-trained deep neural network to predict the full Stokes polarization parameters to obtain the three-dimensional reconstructed shape of the target object includes: The full Stokes polarization parameters are subjected to polarization optics modeling and inversion processing to obtain physical prior information; A pre-trained deep neural network is invoked to predict the physical prior information, thereby obtaining the three-dimensional reconstructed shape of the target object.

6. A polarization three-dimensional imaging device, characterized in that, include: An imaging unit is used to perform optical imaging of a target object based on a pre-designed metasurface. The optical imaging results include polarization-related spatial distribution features pre-encoded on the metasurface and polarization-independent depth features modulated by a single-helix point spread function. A reconstruction unit is used to reconstruct the three-dimensional shape of the target object with an average absolute depth based on the spatial distribution features and the depth features. The pre-design process for the metasurface includes: Obtain the Jones pupil function of the initial metasurface, and determine the initial single-helix phase and the initial diffraction grating phase in the Jones pupil function; The initial diffraction grating phase is optimized in phase distribution to obtain a target diffraction grating phase that meets the requirements of full Stokes polarization separation. The initial single-helix phase is also optimized in phase distribution to obtain a target single-helix phase with more concentrated energy. The metasurface is constructed based on the target diffraction grating phase and the target single-helix phase; The step of optimizing the phase distribution of the initial diffraction grating phase to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation includes: The initial diffraction grating phase is spatially sampled to obtain the sampled phase matrix. The phase distribution of the initial diffraction grating phase is adjusted based on the phase matrix to obtain the target diffraction grating phase that meets the requirements of full Stokes polarization separation; The step of optimizing the phase distribution of the initial single-helix phase to obtain a target single-helix phase with more concentrated energy includes: The initial single-helix phase is subjected to frequency domain transformation to obtain frequency domain phase information; Based on the frequency domain phase information, the phase distribution of the initial single-helix phase is iteratively corrected to obtain a target single-helix phase with more concentrated energy.

7. An electronic device, characterized in that, include: Memory, transceiver, processor, and bus system; The memory is used to store programs; The processor is configured to execute a program in the memory, including performing the method as described in any one of claims 1 to 5; The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

8. A computer-readable storage medium, characterized in that, Includes instructions that, when run on a computer, cause the computer to perform the method as described in any one of claims 1 to 5.

Images