Augmented reality glasses based on microprism array

Abstract

The invention discloses augmented reality glasses based on a microprism array. The glasses comprise two groups of optical modules, each group of optical modules is formed by gluing two refractive-diffractive hybrid microprism lenses, the two refractive-diffractive hybrid microprism lenses are respectively a first refractive-diffractive hybrid microprism lens and a second refractive-diffractive hybrid microprism lens, and a micro display is positioned on one side of each optical module; each refractive-diffractive hybrid microprism lens comprises two opposite surfaces, namely a refractive-diffractive hybrid microprism structure surface and an optical continuous smooth surface; based on a refraction-diffraction mixed optical principle, a deep learning algorithm is introduced to carry out joint optimization design on a lens structure so as to improve the light field regulation and control performance. In the imaging process, light emitted by the micro display enters the first refractive-diffractive hybrid microprism lens and then is guided to human eyes through reflection, so that efficient virtual image display is realized; meanwhile, light of an external real scene is not blocked and can directly penetrate through the optical module to enter human eyes, and high permeability and immersive visual experience are guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of wearable devices, relates to augmented reality glasses, and particularly relates to an augmented reality glasses based on a micro prism array. Background Art

[0002] Augmented Reality (AR) technology provides an immersive interactive experience for users by superimposing computer-generated virtual information on the real world. As the core device for implementing this technology, AR glasses have broad application prospects in the fields of education, medical care, industry, military, entertainment, etc. However, the optical systems of existing AR glasses still face many challenges, including system miniaturization, optical efficiency, dispersion control, imaging distortion compensation, and high transmittance of real scenes.

[0003] Currently, the optical solutions for AR glasses mainly include geometric optical waveguides, diffractive optical waveguides, free-form surface optical systems, reflective optical systems, and metasurfaces. Among them, the geometric optical waveguide solution uses the reflection principle to achieve light guiding, with the advantages of high optical efficiency and a large field of view, but has extremely high requirements for the assembly accuracy of optical components and has a large system thickness. The diffractive optical waveguide solution relies on the grating coupling optical path and can achieve a lightweight design, but has obvious dispersion problems. The free-form surface optical solution can improve the imaging quality, but the system structure is complex and the manufacturing difficulty is high. Although metasurfaces have the ability to control light waves at the nanoscale, the current manufacturing cost is high, which limits large-scale applications. In view of the limitations of existing optical solutions, micro prism array optics, as a new optical design solution, exhibits unique advantages. The micro prism array performs light field regulation based on Snell's law and diffractive optics, can effectively reduce the system size, and at the same time maintain high optical efficiency. Compared with traditional geometric optical waveguides or diffractive optical waveguides, the micro prism array has the advantages of high optical efficiency, compactness, lightweight, and low dispersion. Summary of the Invention

[0004] Aiming at the key problems of low miniaturization degree, insufficient optical efficiency, difficult dispersion control, excessive imaging distortion, and insufficient light transmittance of real scenes in the optical system design of existing augmented reality glasses, the purpose of the present invention is to propose an augmented reality glasses based on a micro prism array, which has high optical utilization rate and has more excellent optical performance under a thinner design.

[0005] The technical solution of the present invention is specifically introduced as follows.

[0006] The present invention provides an augmented reality glasses based on a microlens array, which comprises two groups of optical modules and a microdisplay. Each group of optical modules is formed by gluing two diffractive-refractive hybrid microlens lenses. The two diffractive-refractive hybrid microlens lenses are a first diffractive-refractive hybrid microlens lens and a second diffractive-refractive hybrid microlens lens respectively. The microdisplay is located on one side of the optical module;

[0007] Each diffractive-refractive hybrid microlens lens comprises two opposite surfaces, namely a diffractive-refractive hybrid microlens structure surface and an optically continuous smooth surface. The diffractive-refractive hybrid microlens structure surface is provided with a microstructure based on a microlens array; the optically continuous smooth surfaces of the two diffractive-refractive hybrid microlens lenses are butt-jointed to form a gluing surface; a semi-transmissive and semi-reflective film layer is plated on the central area of the gluing surface;

[0008] During operation, the light emitted by the microdisplay enters the first diffractive-refractive hybrid microlens lens close to the human eye through the spherical surface on the side of the optical module, and then passes through the optically continuous surface, the semi-transmissive and semi-reflective film layer and the modulation of the diffractive-refractive hybrid microlens structure surface in sequence. The light is guided to the human eye to present a virtual image. The light of the external real environment passes through the second diffractive-refractive hybrid microlens away from the human eye and the first diffractive-refractive hybrid microlens close to the human eye and enters the human eye in sequence, realizing the fusion of virtual and real and the display of augmented reality virtual images.

[0009] In the present invention, the design method of the diffractive-refractive hybrid microlens structure surface comprises:

[0010] (1) Introduce a base surface and an original surface. The base surface represents the optically continuous curved surface to which the microstructure array adheres, and the original surface represents the geometric starting surface before the microstructure is added;

[0011] Define the base surface Z base as an even aspheric surface, c b represents the curvature of the base surface, k b is the conic constant, α m represents the even aspheric coefficient of the original surface, r represents the radial coordinate, and M represents the number of even aspheric coefficients of the base surface; the sag of the base surface Z base (x, y) is expressed as:

[0012] ...

[0013] Define

[0014]

[0015] The sag of the original surface Z orin (x, y) is expressed as,

[0016]

[0017] Here, c0 represents the curvature of the original surface, k0 represents the conic constant, r represents the radial coordinate, and the coefficient αn Even aspherical terms corresponding to the original surface, N represents the number of even aspherical coefficients of the original surface, Z(x / R, y / R) represents the Zernike polynomial used to describe the high-order surface shape deviation, and the corresponding Zernike coefficient is γ l , L represents the number of Zernike polynomials;

[0018] (2) Based on the base surface and the original surface, a microprism array structure is superimposed thereon to form the final complex surface, that is, a refractive-diffractive hybrid microprism structure surface is obtained;

[0019] The sagittal height of the refractive-diffractive hybrid microprism structure surface of two refractive-diffractive hybrid microprism lenses is expressed as

[0020]

[0021] Here, Ω i represents the spatial region corresponding to the i-th micro-structure, d is the period length of the microprism, and Z base (x, y) represents the sagittal height of the base surface;

[0022] Above, the sagittal height expressions of the original surface and the base surface are different, and the structures of the first refractive-diffractive hybrid microprism lens and the second refractive-diffractive hybrid microprism lens can be different; the micro-structures on the refractive-diffractive hybrid microprism structure surface are different.

[0023] In the present invention, the arrangement of the micro-structure units on the refractive-diffractive hybrid microprism structure surface is optimized based on the refractive-diffractive hybrid optical principle to improve the light field control performance, increase the optical efficiency, and reduce chromatic aberration.

[0024] In the present invention, the inclination angle of the microprism units on the refractive-diffractive hybrid microprism structure surface in the first refractive-diffractive hybrid microprism lens is optimized based on the total internal reflection constraint to ensure that the light emitted by the micro-display is efficiently transmitted in the waveguide without leakage; specifically, by adjusting the tilt angle of the microprism and the curvature of the optically continuous smooth surface, the incident angles of the light on the refractive-diffractive hybrid microprism structure surface and the optically continuous smooth surface are controlled to be greater than the critical angle of total internal reflection; the expression of the total internal reflection condition is:

[0025]

[0026] where u is the normal vector of the microprism surface, k1 is the wave vector, and n1 and n2 respectively represent the refractive indices of the incident interface and the exit interface.

[0027] In the present invention, a deep learning algorithm is used to optimize the arrangement of the micro - prism units of the first refractive - diffractive hybrid micro - prism lens. Specifically, a deep neural network is used to perform forward - propagation simulation on an image, recording the whole process of the image propagating from the micro - display through the micro - prism array to the human eye. In the system's back - propagation stage, the error gradient between the virtual image and the target image is utilized to jointly optimize the surface parameter θ of the micro - prism and the image pre - processing network. The design parameter θ of the surface of the refractive - diffractive hybrid micro - prism structure described in the present invention mainly includes: (1) surface curvature, position, conic coefficient, and polynomial coefficient, etc., which are used to describe the surface shape parameters of the base surface where the micro - structure adheres; and (2) geometric parameters of the micro - prism array, such as the size, angle, period, etc. of the micro - prism units. The collaborative design of the above - mentioned parameters is used to achieve efficient refractive - diffractive compound regulation of the incident light field. In a specific embodiment, the deep neural network adopts the U - Net network structure.

[0028] In the present invention, the optical continuous smooth surface is optimized by using a deformed aspheric surface to correct imaging distortion and improve the quality of the virtual image. Among them, the deformed aspheric surface is described as

[0029]

[0030] In the formula: Cx is the curvature in the x - direction, Cy is the curvature in the y - direction, Kx is the conic coefficient in the x - direction, Ky is the conic coefficient in the y - direction, AR is the main coefficient of quadratic deformation, AP is the secondary coefficient of quadratic deformation, BR is the main coefficient of cubic deformation, and BP is the secondary coefficient of cubic deformation.

[0031] In the present invention, it is assumed that the light rays with different marginal field - of - view angles emitted from the marginal points P1 and P2 of the micro - display satisfy the total internal reflection condition, and the light rays in the marginal field of view intersect the surface of the refractive - diffractive hybrid micro - prism lens at P a ', P b ', and the range of the semi - transparent and semi - reflective film is larger than the range covered by these two points.

[0032] Compared with the prior art, the augmented reality glasses of the present invention have significant advantages in the following aspects:

[0033] (1) Ultra - thin and compact design: By adopting the refractive - diffractive hybrid micro - prism lens, the overall system thickness is less than or equal to 8 mm, significantly reducing the volume of the AR optical system and improving the wearing comfort.

[0034] (2) High optical efficiency: Through the total internal reflection and refraction optimization of the micro - prism array, the guiding efficiency of light is improved, light loss is reduced, the light energy utilization rate is high, and the brightness of the virtual image is enhanced.

[0035] (3) Low dispersion: The refractive - diffractive hybrid micro - prism lens structure is adopted to optimize dispersion.

[0036] (4) High permeability: The optimized optical structure enables external light to naturally pass through the glasses and enter the human eye, achieving seamless integration of the augmented reality scene and the real world.

[0037] (5) Suitable for large-scale manufacturing: Compared with free-form optics or metasurface optics, the microprism array structure is easier to mass-produce through injection molding process, reducing the manufacturing cost. Brief Description of the Drawings

[0038] Figure 1 It is a schematic structural diagram of the microprism array augmented reality glasses.

[0039] Figure 2 It is the microstructure topography distribution of the refractive-diffractive hybrid microprism lens.

[0040] Figure 3 It is the process of light being guided by the microprism array augmented reality glasses into the human eye.

[0041] Figure 4 It is the modulation transfer function (MTF) of the microprism array augmented reality glasses for displaying virtual images.

[0042] Figure 5 It is the modulation transfer function (MTF) of the microprism array augmented reality glasses for viewing the real world.

[0043] Figure 6 It is the optical path diagram of the display image and the real scene light in the microprism array augmented reality glasses.

[0044] Figure 7 : (a) The effect diagram of the microprism array augmented reality glasses for displaying virtual images; (b) The effect diagram of the microprism array augmented reality glasses for viewing the real world. Detailed Embodiments

[0045] In order to enable the examiner and the public to more clearly understand the technical essence and beneficial effects of the present invention, the applicant will make a detailed description below in the form of embodiments. However, the description of the embodiments does not constitute a limitation to the technical solution of the present invention, and any equivalent transformation that is merely formal rather than substantial based on the concept of the present invention should be regarded as falling within the scope of the technical solution of the present invention.

[0046] Such as Figure 1 - Figure 2As shown, the present invention discloses an augmented reality glasses based on refractive-diffractive hybrid microprismatic lenses, whose core optical system includes two groups of optical modules, each group of optical modules is composed of two refractive-diffractive hybrid microprismatic lenses: a first refractive-diffractive hybrid microprismatic lens 1-1 and a second refractive-diffractive hybrid microprismatic lens 1-2 glued together, and the central area of the glued surface (i.e., the glued connection between the two lenses) is coated with a semi-transparent and semi-reflective film layer 1-6, which is used to achieve the reflection of light emitted by a micro-display 1-7 arranged on one side of the optical module into the eye and the transmission of external natural light into the eye; each refractive-diffractive hybrid microprismatic lens includes two opposite surfaces, one of which is a refractive-diffractive hybrid microprismatic structure surface 1-5, 1-3, and the other is an optically continuous smooth surface 1-4, and the glued surface is selected to be connected to the optically continuous smooth surface 1-4 to ensure high interface adhesion, no obvious scattering or phase distortion, which is conducive to superimposing film layers and maintaining optical path stability; the semi-transparent and semi-reflective film layer 1-6 is located between the bonding surfaces of the two smooth surfaces. This design enables the film layer to play a dual role in the system: reflecting light from the microdisplay into the human eye and transmitting light from the real scene into the human eye, achieving a fusion of virtual and real. The topography of the continuous surface, as well as the shape and number of the microprism units, can be varied. Secondly, the light propagation path of the optical system: light emitted from the microdisplay 1-7, located on one side of the optical module, is reflected by the refractive-diffractive hybrid microprismatic lens 1-1 near the human eye (after being modulated by the optically continuous smooth surface 1-4, the semi-transparent and semi-reflective film layer 1-6, and the refractive-diffractive hybrid microprismatic structure surface 1-5) before ultimately reaching the human eye, forming a virtual image display. Simultaneously, real-world light can pass through the optical module unobstructed (from the refractive-diffractive hybrid microprismatic lens 1-2, which is far from the human eye, to the refractive-diffractive hybrid microprismatic lens 1-1, which is closer to the human eye) and enter the human eye, ensuring the high transparency of the augmented reality system and achieving a fusion of virtual and real. Secondly, the geometric parameters of the microprism structure, the size, angle, period and other parameters of the microprism unit are based on the refraction-diffraction hybrid optical principle, and a deep learning algorithm is introduced for joint optimization to minimize dispersion and distortion, improve the light field control performance, improve the system optical efficiency and effectively suppress chromatic aberration, thereby improving the imaging quality of the optical system.

[0047] Figure 2 The geometric construction framework of the refractive-diffractive hybrid microprism structure surface 1-5 is shown. The base surface (Z base ) represents the optical continuous surface to which the microstructure array is attached, i.e., the "large-scale envelope surface of the microprism array", which is usually a continuous optical surface with free-form surface characteristics. Its surface shape is determined by curvature, conic coefficient, polynomial terms, etc.; this surface determines the overall macroscopic morphology of the microprism array and plays a core role in controlling the initial direction of light in the entire system. Figure 2 In the function Z base (x,y) represents the ideal optical reference plane used as a design reference. orinrefers to the "geometric starting surface" before the attachment of microstructures, which is used to represent the optical topography before the processing of the microstructure array; the original surface Z is introduced orin to facilitate the coordinate expansion and normal calculation of the microstructure parameters thereon; finally, with the base surface Z base or the original surface Z orin as the basis, the complex surface formed by superimposing the microprism array structure thereon is the diffractive-refractive hybrid microprism structure surface 1-5, and each microprism unit adheres to a certain point on the original surface Z orin (such as point P o i ), and its prism surface inclination angle, direction, and width are determined according to the design parameter distribution.

[0048] In the present invention, by introducing the base surface and the original surface to design the diffractive-refractive hybrid microprism structure surface, on the one hand, the macroscopic surface shape and the microscopic structure can be separated: since the optical performance comes from the combination of the continuous phase control of the base surface and the local deflection control of the microstructure, it is necessary to clearly distinguish the two in design for separate optimization; on the other hand, it is convenient to establish the local coordinate system and the microstructure definition: using the original surface as the reference surface helps to establish the tangent plane coordinate system at each microprism position, so as to clarify the microscopic parameters such as its normal direction, incident angle, and edge definition. Further, each microprism unit adheres to a certain point on the original surface (such as point P o i ), and its prism surface inclination angle, direction, and width are determined according to the design parameter distribution.

[0049] In the present invention, there may be structural differences in the design of the two diffractive-refractive hybrid microprism lenses 1-1 and 1-2, and their microstructure sag heights are respectively defined by the original curved surface Z orin and the base surface Z base jointly, specifically:

[0050]

[0051] Ω i represents the spatial region corresponding to the i-th microstructure, and d is the period length of the microprism; among them, Z base is defined as an even aspheric surface, c b represents the curvature of the base surface, k b is the conic constant, and α n represents the even aspheric surface coefficient of the original surface; Z base is expressed as:

[0052]

[0053] Definition:

[0054]

[0055] Here, c0 represents the curvature of the original surface, k0 represents the conic constant, and r represents the radial coordinate. The coefficient α n corresponds to the even aspheric terms of the original surface. Z(x / R, y / R) represents the Zernike polynomial used to describe the high-order surface shape deviation, and the corresponding Zernike coefficient is γ l . The original surface and the base surface have the same radial coordinate.

[0056] Above, the surfaces of the two diffractive-refractive hybrid microprism lenses of the present invention are similar in terms of structural design principles, but not exactly the same in actual parameters. The differences stem from the differences in the incident light direction, field of view design, and system compensation requirements: Among them, the first diffractive-refractive hybrid microprism lens 1-1 is more inclined towards incident angle control and light guiding direction modulation; the second diffractive-refractive hybrid microprism lens 1-2 is more inclined towards collimation in the exit direction and imaging matching. The differences in the structures of the two diffractive-refractive hybrid microprism lenses 1-1 and 1-2 ultimately lead to differences in the microprism surfaces.

[0057] In the present invention, the tilt angle of the microprism unit is optimized according to the total internal reflection constraint, so that the light rays at the edge field of view angles all satisfy the total internal reflection condition, avoiding waveguide leakage and improving the edge imaging quality. The "tilt angle" refers to the tilt angle of the microprism unit facet relative to the normal of the base surface, that is, the deflection angle of the microprism acting on the light ray, which is part of the geometric structure parameters of the microprism; specifically, the present invention controls the incident angles of the light rays on the optically continuous smooth surface 1-4 and the diffractive-refractive hybrid microprism structure surface 1-5 by adjusting the tilt angle of the microprism and the curvature of the continuous smooth surface, making it greater than the critical angle of total internal reflection. The expression for the total internal reflection condition is:

[0058]

[0059] where u is the normal vector of the microprism surface, k1 is the wave vector, and n1 and n2 represent the refractive indices of the incident interface and the exit interface respectively. This formula is derived based on the geometric relationship between the critical condition of total internal reflection and the angle between the incident angle and the interface normal, and is used for the internal light propagation path design of the entire first diffractive-refractive hybrid microprism lens 1-1, and is used to jointly optimize the tilt angle of the diffractive-refractive hybrid microprism structure surface 1-5 and the curvature of the continuous smooth surface 1-4; by optimizing the tilt angle (as Figure 3 shown), the light rays at different edge field of view angles (such as the positive edge field of view θ f+ and the negative edge field of view θ f- ) emitted from the edge points P1 and P2 of the display all satisfy the total internal reflection condition, thereby constraining the field of view range and avoiding energy loss. The light rays in the edge field of view intersect the diffractive-refractive hybrid microprism lens surface at P a ', P b', The range of the semi-transmissive and semi-reflective film should be at least larger than the range covered by these two points; optimize the curvature of the continuous smooth surface 1-4 and control the incident light direction so that the angle when it enters the microstructure array meets the total reflection condition.

[0060] In the present invention, during the joint optimization process of the optical system, a deep neural network is used to perform forward propagation simulation on the image, recording the whole process of the image propagating from the display through the microprism array to the human eye; in the system backpropagation stage, the error gradient between the virtual image and the target image is utilized to jointly optimize the optical parameter θ of the refractive-diffractive hybrid microprism structure surface and the image preprocessing network; specifically, the arrangement of the microprism units on the refractive-diffractive hybrid microprism structure surface 1-5 is optimized using a deep learning algorithm.

[0061]

[0062] During the forward propagation of the image, the deep neural network θ' first preprocesses the image. The preprocessed image is then displayed on the display and propagated through two refractive-diffractive hybrid microprism lenses, and finally reaches the human eye. During the whole process, the gradient of each optical parameter θ is tracked. In the backpropagation stage, the error between the target image and the perceived image is backpropagated to simultaneously optimize the microprism array waveguide and the image preprocessing network.

[0063] In the present invention, the optical continuous smooth surface 1-4 adopts a deformed aspherical design to correct imaging distortion and improve the quality of the virtual image. The surface can be expressed as a combination of an even-order aspherical function and a Zernike polynomial, and the specific description is as follows:

[0064]

[0065] In the present invention, the optical parameter θ representing the refractive-diffractive hybrid microprism structure surface includes the following two dimensions:

[0066] 1. Surface-related parameters (used to describe the macroscopic free surface to which the microprism array adheres):

[0067] These parameters define the optical base surface to which the microstructure adheres, that is, the morphological basis of the entire lens surface:

[0068] A. Surface Curvature: The initial curvature of the macroscopic base surface;

[0069] B. Position Parameter: The position information of the base surface in space (such as vertex position, center position, etc.);

[0070] C. Conic Constant: Defines the geometric type of an ellipsoid, paraboloid or hyper-sphere;

[0071] D. Polynomial Coefficients: used to represent surface shape deviations or free-form surfaces (such as Zernike or Q-type coefficients).

[0072] These parameters determine the arrangement deformation of the microprism array on the curved surface and the overall optical path control performance.

[0073] 2. Microprism Array Parameters:

[0074] These parameters specifically refer to the geometric structure and array characteristics of the microprism unit itself, including:

[0075] A. Dimensions of the microprism unit (such as bottom width, height);

[0076] B. Microprism edge angle or refraction angle (used to control the deflection / diffraction direction of light);

[0077] C. Arrangement period of the microprisms (determines the spatial frequency of array control);

[0078] In the present invention, further, the design of the refractive-diffractive hybrid microprism lens surface 1-3 on the side far from the human eye is determined by the design of the refractive-diffractive hybrid microprism lens surface 1-5 on the side close to the human eye. After the refractive-diffractive hybrid microprism lens surface 1-5 on the side close to the human eye is determined, the light from the real world first passes through the refractive-diffractive hybrid microprism lens surface 1-3 on the side far from the human eye, then passes through the optically continuous smooth surface 1-4, and then passes through the refractive-diffractive hybrid microprism lens surface 1-5 on the side close to the human eye. Finally, it is best that its propagation direction is parallel to the original propagation direction, thereby further determining the sagittal height expression of the refractive-diffractive hybrid microprism lens surface 1-3 on the side far from the human eye.

[0079] Example 1

[0080] In this embodiment, the optical module of the augmented reality glasses is composed of two refractive-diffractive hybrid microprism lenses glued together ( Figure 1 ), and the specific parameters are as follows: the aperture of the optical module is 55 mm, and the thickness is 8 mm; the period of the microprism unit: 50 μm ( Figure 2 marked as d)); the continuous surface of the refractive-diffractive hybrid lens adopts a second-order deformed aspherical surface; the transmittance and reflectivity ratio of the semi-transparent and semi-reflective film layer on the glued surface: 50% / 50%; the size of the micro display: 0.7-inch OLED display screen; the field of view angle is 36°.

[0081] To verify the feasibility of the present invention, the following gives the specific structural parameters and optical configurations of two refractive-diffractive hybrid microprism lenses in a group of typical embodiments. The parameters are based on the actual modeling optimization results and have manufacturability and optical consistency:

[0082] 1. Design parameters of the microprism array:

[0083] Dimensions of the microprism unit: The width is approximately 50 μm and the height is approximately 20 μm;

[0084] Number of microprism units: 1260, covering the entire optical incident window area;

[0085] Material: The microprism substrate is a high - refractive - index optical glass (such as K26R).

[0086] The design of the refractive - diffractive hybrid microprism lenses on surfaces 1 - 3 and 1 - 5 is as described above.

[0087] 2. Parameters of the optically continuous and smooth surface:

[0088] For the deformed aspheric surface, cx is - 2.061646775342683E - 002

[0089] For the deformed aspheric surface, cy is - 1.309862433380350E - 002

[0090] For the deformed aspheric surface, kx is - 2.625102242387300E - 002

[0091] For the deformed aspheric surface, ky is - 2.625102242387300E - 002

[0092] For the deformed aspheric surface, AR is 6.462711856227429E - 007

[0093] For the deformed aspheric surface, AP is 4.161068204184557E - 001

[0094] For the deformed aspheric surface, BR is 1.499135200723314E - 009

[0095] For the deformed aspheric surface, BP is - 1.452409722681883E - 001

[0096] 3. Parameters of the semi - transparent and semi - reflective film:

[0097] Film layer size: 7.140 mm×7.140 mm, located in the central area between two lenses;

[0098] Reflection / transmission ratio: R:T = 50%:50%, and the working wavelength range is 450 - 650 nm;

[0099] In the embodiments of the present invention, a classical U-Net encoding-decoding structure is adopted for the collaborative optimization of image preprocessing and microstructure lens parameters. This network can retain both global structure information and local high-frequency details, and is suitable for tasks such as image reconstruction and structure inversion in an augmented reality (AR) system. The U-Net network structure is specifically as follows:

[0100] 1) Overall network architecture:

[0101] U-Net consists of a symmetric encoder and decoder, and fuses multi-scale information through skip connections in the middle.

[0102] 2) Input layer:

[0103] Input data: A single-channel grayscale image with a size of 128×128 pixels (such as a simulated point spread image or an original microdisplay image);

[0104] Input dimension: 128×128×3.

[0105] 3) Encoder (downsampling path):

[0106] It contains 4 encoding modules, and each module consists of:

[0107] A. Two layers of convolution (kernel size = 3, stride = 1, padding = 1) + ReLU;

[0108] B. One layer of max pooling (max pooling, size = 2);

[0109] The number of convolution channels increases layer by layer to: 64→128→256→512.

[0110] 4) Bottleneck layer (Bridge):

[0111] The convolution channel is 1024, and the structure also consists of two layers of convolution and ReLU activation.

[0112] 5) Decoder (upsampling path):

[0113] It contains 4 decoding modules, and each module includes:

[0114] A. Upsampling (transposed convolution or bilinear upsampling);

[0115] B. Fusing the skip connection with the output of the same layer of the encoder;

[0116] C. Two layers of convolution + ReLU;

[0117] The number of channels is successively: 512 → 256 → 128 → 64.

[0118] 6) Output layer:

[0119] The output of the last convolution:

[0120] A. The number of channels is 1, which is used to generate the optimized image (such as the structured light image or the phase map of the adjusted lens);

[0121] B. Or output the structural parameters (such as polynomial coefficients, micro - prism tilting angles, etc.) through the fully - connected sub - module;

[0122] Activation function: Select to use sigmoid (for image normalized output) or linear activation (for regression of structural parameters) according to the task.

[0123] Figure 2 The micro - structure topography distribution of the refractive - diffractive hybrid micro - prism lens of this embodiment is also shown. Point P 0 represents the point on the original surface; point P b represents the point on the base surface; among them, the micro - prism structures in different regions have different tilting angles (for example, the tilting - angle distribution of the refractive - diffractive hybrid micro - prism lens 1 - 1 near the human eye is 59.95° - 83.23°, and the tilting - angle distribution of the refractive - diffractive hybrid micro - prism lens far from the human eye here is 57.65° - 81.17°) to optimize the optical path and chromatic aberration.

[0124] Figure 3 The beam propagation of the light of the augmented reality glasses in this embodiment inside and outside the refractive - diffractive hybrid micro - prism lenses 1 - 1 and 1 - 2 is given. As Figure 3 shown, the light emitted by the micro - display first passes through the spherical surface 3 and is refracted into the refractive - diffractive hybrid micro - prism lens 1 - 1. The light undergoes total internal reflection successively on the continuous surface (deformed aspherical surface) 2 and the surface 1 - 5 of the refractive - diffractive hybrid micro - prism structure in the micro - prism waveguide, and part of the light is adjusted by the semi - transparent and semi - reflective film layer 1 - 6 to be incident on the human eye to achieve virtual - image display. Among them, 1’ and 2’ represent the total - reflection regions. At the same time, the light of the external real scene can directly pass through the optical module and enter the human eye, and is fused with the virtual image to form an augmented reality picture.

[0125] Figure 4 The modulation transfer function (MTF) curve of the augmented reality glasses system based on the refractive - diffractive hybrid micro - prism array in the present invention when displaying the virtual - image channel is shown. This figure reflects the response ability of the system to different spatial - frequency components, especially the ability to retain image details in the medium - and high - frequency bands. As Figure 4As shown, at a spatial frequency of 30 lp / mm (line pairs per mm), the MTF remains above 0.2, demonstrating a strong ability to retain high-frequency details; within the main visual frequency band of 10 - 20 lp / mm, the MTF is maintained above 0.4, indicating that the virtual image has good clarity and contrast; the overall change of the curve is stable without obvious fluctuations, indicating that there is no modulation attenuation caused by diffraction artifacts or microstructural interference fringes in the image; compared with traditional planar prisms or total reflection waveguides, better off-axis field imaging uniformity is maintained. Figure 4 It verifies the high spatial frequency fidelity and edge sharpness of the micro prism array when transmitting virtual images, indicating that the system of the present invention has good image restoration ability, and is especially suitable for the projection requirements of high-precision content on small displays.

[0126] Figure 5 It then shows the modulation transfer function performance when viewing the real environment channel (transmission path) through the system of the present invention, for verifying the imaging influence of the micro prism array structure on the light of the real scene. As Figure 5 shown, within the typical human eye visual sensitive frequency range of 10 - 15 lp / mm, the MTF is maintained above 0.5, indicating that the human eye will not perceive obvious blurring when observing the real scene; at a high frequency of 30 lp / mm, it still remains at 0.1, indicating that the diffraction and scattering effects caused by the light transmission structure (micro prisms, film layers, etc.) of the system are small; the results show that the system of the present invention not only has the ability to output high-quality virtual images, but also maintains high imaging quality in terms of the real scene through-view field, and is suitable for application scenarios such as daily wear, augmented reality navigation, and medical assistance. Compared with the total reflection structure, the structure of the present invention has less interference with the real light field and higher transmittance, and the comprehensive imaging performance of the system is better.

[0127] Figure 6 It gives the optical path diagram of the display image and the real scene light in the micro prism array augmented reality glasses.

[0128] Figure 7 It gives the effect diagrams of the micro prism array augmented reality glasses displaying virtual images and viewing the real world.

Claims

1. An augmented reality glasses based on a micro prism array, characterized in that, It includes two sets of optical modules and a microdisplay. Each set of optical modules is glued together by two refractive-diffractive hybrid microprism lenses, which are the first refractive-diffractive hybrid microprism lens and the second refractive-diffractive hybrid microprism lens respectively. The microdisplay is located on one side of the optical module. Each refractive-diffractive hybrid microprism lens includes two opposite surfaces, namely the refractive-diffractive hybrid microprism structure surface and the optically continuous smooth surface. The refractive-diffractive hybrid microprism structure surface is provided with a microstructure based on a microprism array; the optically continuous smooth surfaces of the two refractive-diffractive hybrid microprism lenses are butt-jointed to form a glued surface; a semi-transmissive and semi-reflective film layer is deposited on the central area of the glued surface. During operation, the light emitted by the microdisplay enters the first refractive-diffractive hybrid microprism lens close to the human eye through the spherical surface on the side of the optical module, and then passes through the optically continuous surface, the semi-transmissive and semi-reflective film layer, and the modulation of the refractive-diffractive hybrid microprism structure surface in sequence. The light is guided to the human eye to present a virtual image. The light of the external real environment passes through the second refractive-diffractive hybrid microprism far from the human eye and the first refractive-diffractive hybrid microprism close to the human eye and enters the human eye in sequence, realizing the fusion of virtual and real and achieving augmented reality virtual image display.

2. The augmented reality glasses according to claim 1, characterized in that, The design method of the refractive-diffractive hybrid microprism structure surface includes: (1) Introduce a base surface and an original surface. The base surface represents the optically continuous curved surface to which the microstructure array adheres, and the original surface represents the geometric starting surface before the microstructure is added. Define the base surface Z base is an even aspheric surface, c b represents the curvature of the base surface, k b is the conic constant, α m represents the even aspheric coefficient of the original surface, r represents the radial coordinate, M represents the number of even aspheric coefficients of the base surface; the sag Z base (x, y) is expressed as: Define The sagittal height Z of the original surface orin (x, y) is expressed as Here, c0 represents the curvature of the original surface, k0 represents the conic constant, r represents the radial coordinate, and the coefficient α n corresponds to the even aspheric terms of the original surface, N represents the number of even aspheric coefficients of the original surface, Z(x / R, y / R) represents the Zernike polynomial used to describe the high-order surface shape deviation, and the corresponding Zernike coefficient is γ l , L represents the number of Zernike polynomials; (2) Based on the base surface and the original surface, superimpose a microprism array structure on them to form a final complex surface, that is, obtain the refractive-diffractive hybrid microprism structure surface. The sag height of the refractive-diffractive hybrid microprism structure surfaces of the two refractive-diffractive hybrid microprism lenses is expressed as Here, Ω i represents the spatial region corresponding to the i-th microstructure, d is the period length of the microprism, and Z base (x, y) represents the sagittal height of the base surface.

3. The augmented reality glasses according to claim 2, characterized in that, The arrangement of the microstructure units on the refractive-diffractive hybrid microprism structure surface is optimized based on the principle of refractive-diffractive hybrid optics to improve the light field regulation performance, increase the optical efficiency, and reduce chromatic aberration.

4. The augmented reality glasses according to claim 3, characterized in that, The inclination angle of the microprism units on the refractive-diffractive hybrid microprism structure surface in the first refractive-diffractive hybrid microprism lens is optimized based on the total internal reflection constraint to ensure the efficient transmission of the light emitted by the microdisplay in the waveguide without leakage; specifically, by adjusting the inclination angle of the microprisms and the curvature of the optically continuous smooth surface, the incident angles of the light on the refractive-diffractive hybrid microprism structure surface and the optically continuous smooth surface are controlled to be greater than the critical angle of total internal reflection; the expression of the total internal reflection condition is: where u is the normal vector of the microprism surface, k1 is the wave vector, and n1 and n2 represent the refractive indices of the incident interface and the exit interface respectively.

5. The augmented reality glasses according to claim 4, characterized in that, Use a deep learning algorithm to optimize the arrangement of the microprism units of the first refractive-diffractive hybrid microprism lens. Specifically: use a deep neural network to perform forward propagation simulation on the image, and record the whole process of the image propagating from the microdisplay through the microprism array to the human eye; in the system backpropagation stage, use the error gradient between the virtual image and the target image to jointly optimize the optical parameters of the microprism surface and the image preprocessing network.

6. The augmented reality glasses according to claim 5, characterized in that, The microprism surface parameters include surface-related parameters and microprism array parameters. Among them, the surface-related parameters include surface curvature, position parameters, conic coefficients, and polynomial coefficients; the microprism array parameters include microprism unit size, microprism edge angle or refractive angle, and microprism arrangement period.

7. The augmented reality glasses according to claim 5, characterized in that, After determining the surface structure of the refractive-diffractive hybrid microprism lens in the first refractive-diffractive hybrid microprism lens, control the propagation direction of the light in the real external environment after passing through the second refractive-diffractive hybrid microprism lens and the first refractive-diffractive hybrid microprism lens to be parallel to the original propagation direction, so as to determine the sag height expression of the surface of the refractive-diffractive hybrid microprism lens in the second refractive-diffractive hybrid microprism lens.

8. The augmented reality glasses according to claim 1, characterized in that, The optical continuous smooth surface is optimized by using a deformed aspheric surface to correct the imaging distortion and improve the virtual image quality; Among them, the deformed aspheric surface is described as In the formula: Cx is the curvature in the x direction, Cy is the curvature in the y direction, Kx is the conic coefficient in the x direction, Ky is the conic coefficient in the y direction, AR is the main coefficient of the quadratic deformation, AP is the auxiliary coefficient of the quadratic deformation, BR is the main coefficient of the cubic deformation, and BP is the auxiliary coefficient of the cubic deformation.

9. The augmented reality glasses according to claim 1, characterized in that The light rays with different marginal field angles emitted from the marginal points P1 and P2 of the microdisplay satisfy the total internal reflection condition, and the light rays in the marginal field of view intersect the surface of the diffractive-refractive hybrid microprism lens at P a ', P b ', and the range of the semi-transmissive and semi-reflective film is larger than the range covered by these two points.

Images