High-resolution light field image reconstruction method, device, electronic device and storage medium

Abstract

The present invention relates to a high-resolution light field image reconstruction method, device, electronic device and storage medium, and belongs to the field of light field imaging technology. The present invention utilizes the electrically controlled focusing characteristics of a liquid crystal microlens array to realize the acquisition of different light field images. The acquisition process does not require frequent movement of the liquid crystal microlens array, which can reduce the acquisition time of the light field image and the required manpower and material resources, thereby improving the efficiency of the above-mentioned high-resolution light field image reconstruction method, and utilizes light field images at different focus positions to reconstruct high-resolution light field images based on a signal recovery method, so that the reconstruction process of the high-resolution light field image does not require the introduction of a high-resolution network, thereby making the above-mentioned high-resolution light field image reconstruction method easier to implement.

Description

Background Art

[0002] Existing high-resolution light field image reconstruction methods are mainly completed through high-resolution networks, but such methods require massive light field image data to train the high-resolution network. Currently, the acquisition of light field images is mainly achieved using a multi-camera array (a glass-type microlens array with a fixed focal length). In order to collect light field images at different focus positions, the multi-camera array needs to be moved multiple times.

[0003] Based on the analysis of the above-mentioned related technologies, the existing high-resolution light field image reconstruction methods have the defects of time-consuming and labor-intensive light field image acquisition and model training. Summary of the invention

[0004] In order to solve the defects of existing high-resolution light field image reconstruction methods that light field image acquisition and model training are time-consuming and labor-intensive, the present invention provides a high-resolution light field image reconstruction method, device, electronic device and storage medium.

[0005] In a first aspect, in order to solve the above technical problems, the present invention provides a high-resolution light field image reconstruction method, comprising:

[0006] The liquid crystal microlens array is used to obtain light field images at different focus positions of the target scene;

[0007] For each light field image, performing signal recovery on the light field image to obtain a target image corresponding to the light field image;

[0008] Based on the target images corresponding to the light field images, a high-resolution light field image corresponding to the target scene is established.

[0009] The beneficial effects of the present invention are as follows: the electrically controlled focusing characteristics of the liquid crystal microlens array are utilized to realize the acquisition of different light field images. The acquisition process does not require frequent movement of the liquid crystal microlens array, which can reduce the acquisition time of the light field image and the required manpower and material resources, thereby improving the efficiency of the above-mentioned high-resolution light field image reconstruction method. In addition, light field images at different focusing positions are used to reconstruct high-resolution light field images based on a signal recovery method, so that the reconstruction process of the high-resolution light field image does not require the introduction of a high-resolution network, thereby making the above-mentioned high-resolution light field image reconstruction method easier to implement.

[0010] Furthermore, the above-mentioned method of establishing a high-resolution light field image corresponding to the target scene based on the target image corresponding to each light field image includes:

[0011] Performing denoising processing on each target image to obtain each target image after denoising processing;

[0012] For each target image after denoising, the pixel visual saliency corresponding to each pixel position in the target image is calculated according to the pixel value of each pixel position in the target image, wherein the pixel visual saliency is used to characterize the visual saliency of the pixel;

[0013] Calculate the mean value of the pixel visual saliency corresponding to the same pixel position in each target image after denoising;

[0014] Based on the mean value of the visual saliency of each pixel, a high-resolution light field image corresponding to the target scene is established.

[0015] The beneficial effect of adopting the above-mentioned improved scheme is: the signal-to-noise ratio of each target image is improved by denoising, and the light field image is reconstructed using the mean of the pixel visual saliency, so as to obtain a high-resolution light field image with a higher signal-to-noise ratio peak and better visual saliency.

[0016] Furthermore, the above-mentioned denoising process is performed on each target image to obtain each target image after denoising, including:

[0017] For each target image, the fusion operation window is used to divide the target image into multiple pixel blocks, and each pixel block is subjected to bilateral filtering to obtain the denoised target image.

[0018] The beneficial effect of adopting the above-mentioned improved scheme is that the denoising method using the fusion operation window combined with the bilateral filtering processing can better achieve denoising of the target image.

[0019] Furthermore, for each pixel position in the target image after denoising, the pixel visual saliency corresponding to the pixel position is determined by the following formula:

[0020] Pv i (x0,y0)=|Pd i (x0,y0)-mean i (i,j)|

[0021] Among them, (x0, y0) represents the pixel position, Pv i (x0, y0) represents the pixel visual saliency corresponding to the pixel position, Pd i (x0, y0) represents the pixel value at the pixel point, mean i (i, j) represents the pixel average value of the pixel block corresponding to the pixel position, and (i, j) represents the size of the pixel block.

[0022] The beneficial effect of adopting the above-mentioned improved solution is that by calculating the difference between the pixel value of each pixel point position and its corresponding pixel average value, the pixel visual saliency corresponding to each pixel point position can be reasonably obtained.

[0023] Furthermore, for each light field image, performing signal recovery on the light field image to obtain a target image corresponding to the light field image includes:

[0024] For each light field image, a pre-established measurement matrix is ​​used to perform signal recovery on the light field image to obtain a target image corresponding to the light field image, wherein the measurement matrix is ​​used to characterize a conversion relationship between the light field image and the target image corresponding to the light field image;

[0025] Among them, for each light field image, the target image corresponding to the light field image is determined by the following formula:

[0026] P(i)=ΦX(i)

[0027] Among them, P(i) represents the light field image, X(i) represents the target image, and Φ represents the measurement matrix.

[0028] The beneficial effect of adopting the above-mentioned improved solution is that: by using the above-mentioned signal recovery method, a target image with more detailed information can be obtained based on the light field image, so that a high-resolution light field image can be obtained by using each target image.

[0029] Furthermore, the liquid crystal microlens array includes a substrate conductive layer for aligning the liquid crystal in a predetermined direction, and the substrate conductive layer includes a zinc oxide film with a grid structure.

[0030] The beneficial effect of adopting the above-mentioned improvement scheme is that: by using the zinc oxide film with a grid structure, the liquid crystal orientation function based on the semiconductor material can be realized, thereby improving the orientation effect on the liquid crystal.

[0031] Furthermore, the substrate conductive layer further comprises a glass substrate, and the substrate conductive layer is prepared by the following method:

[0032] Zinc oxide thin films were prepared on glass substrates using pulsed laser deposition technology;

[0033] The zinc oxide film on the glass substrate is etched into a grid structure by using photolithography and etching technology to obtain a substrate conductive layer.

[0034] The beneficial effect of adopting the above-mentioned improvement scheme is: preparing zinc oxide film through pulse laser deposition technology and photolithography etching technology to achieve liquid crystal orientation can overcome the defects such as tiny impurities, fluff and static electricity caused by traditional friction orientation and reduce physical damage to the device.

[0035] In a second aspect, the present invention provides a high-resolution light field image reconstruction device, comprising:

[0036] An acquisition module, used to acquire light field images at different focus positions of a target scene using a liquid crystal microlens array;

[0037] A processing module, used for performing signal recovery on each light field image to obtain a target image corresponding to the light field image;

[0038] The reconstruction module is used to establish a high-resolution light field image corresponding to the target scene based on the target image corresponding to each light field image.

[0039] In a third aspect, the present invention provides a computer-readable storage medium having instructions stored therein. When the instructions are executed on a terminal device, the terminal device executes all or part of the steps of the high-resolution light field image reconstruction method of the first aspect.

[0040] In a fourth aspect, the present invention provides an electronic device, comprising a memory, a processor, and a program stored in the memory and running on the processor, wherein when the processor executes the program, all or part of the steps of the high-resolution light field image reconstruction method of the first aspect are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of a process for reconstructing a high-resolution light field image provided by an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of the structure of a substrate conductive layer provided in an embodiment of the present invention;

[0043] Figure 3 A schematic structural diagram of a high-resolution light field image reconstruction device provided by an embodiment of the present invention;

[0044] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following examples are provided to further explain and supplement the present invention and do not constitute any limitation to the present invention.

[0046] A high-resolution light field image reconstruction method according to an embodiment of the present invention is described below with reference to the accompanying drawings.

[0047] Reference Figure 1 As shown, the present invention provides a high-resolution light field image reconstruction method, comprising the following steps S1 to S3, wherein:

[0048] In step S1, a liquid crystal microlens array is used to acquire light field images at different focus positions of a target scene.

[0049] It should be noted that liquid crystal material is a birefringent material. When the liquid crystal material is sealed in a specific electric field, the different swinging angles of the liquid crystal molecules at different positions will cause different refractive indices, and different voltages will form different electric fields, which in turn affect the distribution of the refractive index.

[0050] In this embodiment, liquid crystal materials are applied to light field imaging, and a light field imaging system based on a liquid crystal microlens array is used. The spatial and angular information of the outgoing light beam of the target scene is captured by accessing the timing electrical signal of the light field imaging system to realize the detection of four-dimensional light field information, thereby utilizing the characteristics of the liquid crystal microlens array's electrically controlled focusing to collect light field images at different focusing positions.

[0051] Optionally, in one embodiment, the liquid crystal microlens array includes a substrate conductive layer for aligning the liquid crystal in a predetermined direction, and the substrate conductive layer includes a zinc oxide film with a grid structure.

[0052] It should be noted that zinc oxide is a new type of wide bandgap compound semiconductor material. Due to its excellent photoelectric properties and high chemical stability, it can be used as a transparent electrode in the field of optoelectronic display.

[0053] The device mechanism of the liquid crystal microlens array oriented by zinc oxide thin film with a grid structure is as follows:

[0054] The zinc oxide film has a single C-axis crystal preferential orientation, which determines that it has good orientation (orientation) effect and electromechanical coupling. This property makes it possible to arrange liquid crystal molecules in an orderly manner over a large area.

[0055] The groove theory assumes that the polyimide layer forms a sinusoidal periodic surface after friction orientation, and the bending elastic constant and the splay elastic constant of the liquid crystal are the same (K11 = K33 = K), then the anchoring energy per unit area generated by the surface is Where A and q (q>0) are the amplitude and wave function of the sinusoidal periodic surface, is the angle between the groove and the horizontal direction.

[0056] In the Cassie-Baxter wetting model, the contact angle between the liquid crystal droplet and the zinc oxide film with a grid microstructure is cosθ r =p s (cosθ s +1)-1. Among them, p s It is expressed as the area ratio of the liquid crystal droplet in contact with the zinc oxide film, θ s It represents the intrinsic contact angle of the liquid crystal droplet at the zinc oxide film interface, which can be measured using a water drop angle tester.

[0057] Combining the above groove theory and Cassie-Baxter wetting model, it can be known that the contact angle of the liquid crystal droplet is closely related to the surface tension between the droplet and the solid. The liquid crystal molecule is a long chain molecule, and each part of the molecule has different polarity and surface tension, that is, the van der Waals force of the film can affect the orientation of the liquid crystal molecules on the microstructure film. Therefore, the zinc oxide film with a periodic grid microstructure has a better orientation ability for the liquid crystal molecules.

[0058] Optionally, in one embodiment, the substrate conductive layer further comprises a glass substrate, and the substrate conductive layer is prepared by the following method:

[0059] Zinc oxide thin films were prepared on glass substrates using pulsed laser deposition technology;

[0060] The zinc oxide film on the glass substrate is etched into a grid structure by using photolithography and etching technology to obtain a substrate conductive layer.

[0061] It should be noted that the existing preparation methods for liquid crystal microlens arrays include modal addressing, polymer network, surface-relief profile and curved electrode. These methods use the birefringence property of liquid crystal to induce liquid crystal to form a gradient refractive index distribution under the action of an electric field, and the incident light generates an optical path difference after passing through the liquid crystal layer, thereby achieving an electrically controlled focusing effect.

[0062] The existing method usually uses the friction polyimide method to achieve anchoring of liquid crystal molecules, that is, to prepare the substrate conductive layer based on the friction orientation method, but this method will produce orientation defects caused by mechanical friction. In this embodiment, the zinc oxide film with a grid structure is prepared by using pulsed laser deposition technology and photolithography etching technology to achieve anchoring of liquid crystal molecules, which can overcome the defects such as tiny impurities, fluff and static electricity caused by traditional friction orientation, and can also reduce physical damage to the device and improve the orientation effect of liquid crystal.

[0063] A large-area zinc oxide film is prepared by pulsed laser deposition technology, and the prepared large-area zinc oxide film is etched into a grid structure using photolithography and etching technology. The zinc oxide film microstructure can achieve large-area orderly arrangement of liquid crystal molecules using a single-sided induced orientation method. Therefore, by preparing a zinc oxide-based liquid crystal microlens array (ZnO-LCMLA) and coupling the ZnO-LCMLA with an imaging detector, the light field image at different focus positions can be collected.

[0064] In this embodiment, the liquid crystal microlens array with zinc oxide thin film orientation is composed of a substrate conductive layer with zinc oxide thin film and a nematic liquid crystal layer. Figure 2 As shown, the substrate conductive layer includes a glass substrate, and a zinc oxide thin film layer with a grid microstructure prepared on the glass substrate by pulsed laser deposition technology; the nematic liquid crystal layer includes a glass substrate, an ITO (indium tin oxide) layer and a liquid crystal layer, the ITO layer is arranged between the glass substrate and the liquid crystal layer, and the liquid crystal molecules in the liquid crystal layer are horizontally and uniformly arranged by the zinc oxide thin film orientation layer.

[0065] In the absence of an external voltage, the single nematic liquid crystal (liquid crystal molecules) in the liquid crystal layer tend to align parallel to the preset anchoring direction. Under the action of an external voltage, the liquid crystal molecules begin to rotate, and then gradually deflect in the direction of the electric field, gradually becoming perpendicular to the top zinc oxide film glass substrate and the bottom ITO glass substrate of the liquid crystal microlens. At the same time, the transmittance of the liquid crystal device is gradually improved.

[0066] The molecular structure of liquid crystal determines its optical anisotropy, that is, there are two refractive indices n e With n o When the electric field acts on the liquid crystal microlens array, the liquid crystal director rotates to an angle θ, at which point its effective refractive index is generated by extraordinary light. The focus can thus be changed by influencing a change in the refractive index.

[0067] As a possible implementation, the lithography line width ratio of the grid structure is set to 1:5.

[0068] In step S2, for each light field image, signal restoration is performed on the light field image to obtain a target image corresponding to the light field image.

[0069] It should be noted that the light field corresponding to the target scene is usually characterized by a dual-plane model, and the light field data can be expressed as L(u,v,s,t), where (u,v) represents the angle information of the light field and (s,t) represents the spatial position information of the light field. The light field image collected from the target scene can be regarded as mapping the 4D light field information to the irradiance of the 2D image sensor I(x,y)=∫∫L(u,v,s,t)dudv, where (x,y) represents the horizontal and vertical coordinates of the image sensor that collects the light field image.

[0070] Signal recovery is performed on light field images at different focus positions, that is, light field data (target image) with more detailed information (such as more angle information of the light field and spatial position information of the light field) is obtained based on the collected light field images.

[0071] In this embodiment, the liquid crystal microlens array is coupled with an image sensor to form a light field imaging system. When different voltages are applied, the transmittance of the liquid crystal microlens changes, and the focal length changes accordingly. In order to establish a high-resolution light field image, it is necessary to obtain the corresponding light field images under different voltages to reconstruct the light field image.

[0072] Exemplarily, according to the electrically controlled focusing characteristics of the ZnO-LCMLA device, different voltages are adjusted to obtain light field images at corresponding voltages. The light field images obtained from the starting voltage to the threshold voltage of the liquid crystal device are arranged in the order of the voltage values ​​to obtain light field images P(i) at different focus positions, 0≤i≤N, where i represents the light field image with sequence number i, and N represents the number of light field images.

[0073] Optionally, in one embodiment, for each light field image, performing signal recovery on the light field image to obtain a target image corresponding to the light field image includes:

[0074] For each light field image, a pre-established measurement matrix is ​​used to perform signal recovery on the light field image to obtain a target image corresponding to the light field image, wherein the measurement matrix is ​​used to characterize a conversion relationship between the light field image and the target image corresponding to the light field image;

[0075] Among them, for each light field image, the target image corresponding to the light field image is determined by the following formula:

[0076] P(i)=ΦX(i)

[0077] Among them, P(i) represents the light field image, X(i) represents the target image, and Φ represents the measurement matrix.

[0078] It can be understood that in the compressed sensing theory P(i)=ΦX(i), the measurement matrix is ​​used to reduce the dimension of the original signal (target image) X(i) while minimizing the information loss of the original signal. Therefore, under the condition that the measurement matrix and the light field image are known, the compressed sensing theory can be used to reasonably restore the original signal with higher dimension and more information from the light field image. The original signal has a specific representation of more detailed information and has better visual saliency (clarity) and spatial resolution.

[0079] As a possible implementation, the measurement matrix is ​​set to a Gaussian matrix, and the Gaussian matrix is ​​constructed as follows: a m×n matrix Φ that obeys a Gaussian distribution is randomly generated, and each element in the matrix Φ obeys a Gaussian distribution with a mean of 0 and a variance of 1 / m. The size of the matrix Φ m×n is freely set according to actual needs, and this is not limited in this embodiment.

[0080] In step S3, a high-resolution light field image corresponding to the target scene is established based on the target images corresponding to the light field images.

[0081] Among them, high-resolution light field images refer to light field images that have better visual significance and higher peak signal-to-noise ratio than light field images directly collected by devices.

[0082] Optionally, in one embodiment, the implementation process of establishing a high-resolution light field image corresponding to a target scene based on the target image corresponding to each light field image includes:

[0083] Performing denoising processing on each target image to obtain each target image after denoising processing;

[0084] For each target image after denoising, the pixel visual saliency corresponding to each pixel position in the target image is calculated according to the pixel value of each pixel position in the target image, wherein the pixel visual saliency is used to characterize the visual saliency of the pixel;

[0085] Calculate the mean value of the pixel visual saliency corresponding to the same pixel position in each target image after denoising;

[0086] Based on the mean value of the visual saliency of each pixel, a high-resolution light field image corresponding to the target scene is established.

[0087] It can be understood that visual saliency indicates the degree to which image information receives visual attention. For example, if a pixel has a higher grayscale in an image, the degree to which the pixel is valued by the human eye is higher than that of other pixels. That is, the image information of the pixel is more easily noticed (receives a higher degree of visual attention), and the image information of a pixel with a lower degree of visual attention is more easily ignored by the human eye.

[0088] Exemplarily, the high-resolution light field image is determined by the following formula:

[0089]

[0090] Where LH represents the high-resolution light field image, Pv i represents the pixel visual saliency corresponding to the i-th denoised target image, and N represents the number of denoised target images.

[0091] Optionally, in one embodiment, the above-mentioned denoising process is performed on each target image to obtain each target image after denoising, and the implementation process includes:

[0092] For each target image, the fusion operation window is used to divide the target image into multiple pixel blocks, and each pixel block is subjected to bilateral filtering to obtain the denoised target image.

[0093] As a possible implementation, the target image is used as an input source for reconstructing a high-resolution light field image, and the size of the fusion operation window is set to the sub-aperture size of the liquid crystal microlens array, where the sub-aperture size refers to the side length of the largest circle inscribed in the circular hole array in the liquid crystal microlens array.

[0094] Optionally, in one embodiment, for each pixel position in the target image after denoising, the pixel visual saliency corresponding to the pixel position is determined by the following formula:

[0095] Pv i (x0,y0)=|Pd i (x0,y0)-mean i (i,j)|

[0096] Among them, (x0, y0) represents the pixel position, Pv i (x0, y0) represents the pixel visual saliency corresponding to the pixel position, Pd i (x0, y0) represents the pixel value at the pixel point, mean i (i, j) represents the pixel average value of the pixel block corresponding to the pixel position, and (i, j) represents the size of the pixel block.

[0097] The high-resolution light field image reconstruction method provided in the above embodiment utilizes the electrically controlled focusing characteristics of the liquid crystal microlens array to realize the acquisition of different light field images. The acquisition process does not require frequent movement of the liquid crystal microlens array, which can reduce the light field image acquisition time and the required manpower and material resources, thereby improving the efficiency of the above high-resolution light field image reconstruction method. In addition, light field images at different focusing positions are used to reconstruct high-resolution light field images based on a signal recovery method, so that the reconstruction process of the high-resolution light field image does not need to introduce a high-resolution network, thereby making the above high-resolution light field image reconstruction method easier to implement.

[0098] like Figure 3 As shown, an embodiment of the present invention provides a high-resolution light field image reconstruction device, comprising:

[0099] An acquisition module 20 is used to acquire light field images at different focus positions of a target scene using a liquid crystal microlens array;

[0100] A processing module 30 is used to perform signal recovery on each light field image to obtain a target image corresponding to the light field image;

[0101] The reconstruction module 40 is used to establish a high-resolution light field image corresponding to the target scene based on the target image corresponding to each light field image.

[0102] Optionally, the reconstruction module 40 is specifically used to perform denoising on each target image to obtain each target image after denoising; for each target image after denoising, calculate the pixel visual saliency corresponding to each pixel position in the target image according to the pixel value of each pixel position in the target image, wherein the pixel visual saliency is used to characterize the visual saliency of the pixel; calculate the average of the pixel visual saliency corresponding to the same pixel position in each target image after denoising; and establish a high-resolution light field image corresponding to the target scene based on the average of the visual saliency of each pixel.

[0103] Optionally, the reconstruction module 40 is further configured to divide each target image into a plurality of pixel blocks using a fusion operation window, and perform bilateral filtering on each pixel block to obtain a denoised target image.

[0104] Optionally, the processing module 30 is specifically used to perform signal recovery on the light field image using a pre-established measurement matrix for each light field image to obtain a target image corresponding to the light field image, wherein the measurement matrix is ​​used to characterize the conversion relationship between the light field image and the target image corresponding to the light field image; wherein, for each light field image, the target image corresponding to the light field image is determined by the following formula:

[0105] P(i)=ΦX(i)

[0106] Among them, P(i) represents the light field image, X(i) represents the target image, and Φ represents the measurement matrix.

[0107] An embodiment of the present invention provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal device, the terminal device executes the steps of the high-resolution light field image reconstruction method in any of the above embodiments.

[0108] like Figure 4 As shown, an electronic device 500 provided by an embodiment of the present invention includes a memory 510, a processor 520, and a program 530 stored in the memory 510 and running on the processor 520. When the processor 520 executes the program 530, the steps of the high-resolution light field image reconstruction method of any of the above embodiments are implemented.

[0109] Among them, the electronic device 500 can be a computer, a mobile phone, etc. Correspondingly, its program 530 is a computer software or a mobile phone App, etc., and the above-mentioned parameters and steps in an electronic device 500 of the present invention can refer to the parameters and steps in the embodiment of the high-resolution light field image reconstruction method in the above text, and will not be repeated here.

[0110] Those skilled in the art know that the present invention can be implemented as a system, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: it can be complete hardware, it can be complete software (including firmware, resident software, microcode, etc.), and it can also be a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" herein. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable medium contains computer-readable program code.

[0111] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0112] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A high-resolution light field image reconstruction method, characterized in that: include: The liquid crystal microlens array is used to obtain light field images at different focus positions of the target scene; For each of the light field images, performing signal recovery on the light field image to obtain a target image corresponding to the light field image; Based on the target images corresponding to the light field images, a high-resolution light field image corresponding to the target scene is established; The step of establishing a high-resolution light field image corresponding to the target scene based on the target image corresponding to each of the light field images comprises: Performing denoising processing on each of the target images to obtain each of the target images after denoising processing; For each of the target images after denoising, calculating the pixel visual saliency corresponding to each pixel position in the target image according to the pixel value of each pixel position in the target image, wherein the pixel visual saliency is used to characterize the visual saliency of the pixel; Calculating the mean value of the pixel visual saliency corresponding to the same pixel position in each of the target images after denoising; Based on the average of the visual saliency of each pixel, a high-resolution light field image corresponding to the target scene is established; The step of performing denoising on each of the target images to obtain each of the target images after denoising includes: For each of the target images, the target image is divided into a plurality of pixel blocks using a fusion operation window, and each of the pixel blocks is subjected to a bilateral filtering process to obtain the target image after denoising; For each of the light field images, performing signal recovery on the light field image to obtain a target image corresponding to the light field image includes: For each of the light field images, performing signal recovery on the light field image using a pre-established measurement matrix to obtain a target image corresponding to the light field image, wherein the measurement matrix is ​​used to characterize a conversion relationship between the light field image and the target image corresponding to the light field image; Wherein, for each of the light field images, the target image corresponding to the light field image is determined by the following formula: P(i)=ΦX(i) Among them, P(i) represents the light field image, X(i) represents the target image, and Φ represents the measurement matrix.

2. The method according to claim 1, characterized in that For each pixel position in the target image after denoising, the pixel visual saliency corresponding to the pixel position is determined by the following formula: Pv i (x0,y0)=|Pd i (x0,y0)-mean i (i,j)| Among them, (x0, y0) represents the pixel position, Pv i (x0, y0) represents the pixel visual saliency corresponding to the pixel position, Pd i (x0, y0) represents the pixel value at the pixel point, mean i (i, j) represents the pixel average value of the pixel block corresponding to the pixel position, and (i, j) represents the size of the pixel block.

3. The method according to any one of claims 1 to 2, characterized in that The liquid crystal microlens array comprises a substrate conductive layer for aligning liquid crystal in a predetermined direction, wherein the substrate conductive layer comprises a zinc oxide film with a grid structure.

4. The method according to claim 3, characterized in that The substrate conductive layer further comprises a glass substrate, and the substrate conductive layer is prepared by the following method: preparing a zinc oxide thin film on the glass substrate by using a pulsed laser deposition technique; The zinc oxide film on the glass substrate is etched into a grid structure by using photolithography and etching technology to obtain the substrate conductive layer.

5. A high-resolution light field image reconstruction device, characterized in that: include: An acquisition module, used to acquire light field images at different focus positions of a target scene using a liquid crystal microlens array; A processing module, configured to perform signal recovery on each of the light field images to obtain a target image corresponding to the light field image; A reconstruction module, used to establish a high-resolution light field image corresponding to the target scene based on the target images corresponding to the light field images; The reconstruction module is specifically used to: perform denoising on each of the target images to obtain each of the target images after denoising; For each of the target images after denoising, calculating the pixel visual saliency corresponding to each pixel position in the target image according to the pixel value of each pixel position in the target image, wherein the pixel visual saliency is used to characterize the visual saliency of the pixel; Calculating the mean value of the pixel visual saliency corresponding to the same pixel position in each of the target images after denoising; Based on the average of the visual saliency of each pixel, a high-resolution light field image corresponding to the target scene is established; The reconstruction module is specifically used to: perform denoising on each of the target images to obtain each of the target images after denoising, including: For each of the target images, the target image is divided into a plurality of pixel blocks using a fusion operation window, and each of the pixel blocks is subjected to a bilateral filtering process to obtain the target image after denoising; The processing module is specifically used to: for each of the light field images, perform signal recovery on the light field image to obtain a target image corresponding to the light field image, including: For each of the light field images, performing signal recovery on the light field image using a pre-established measurement matrix to obtain a target image corresponding to the light field image, wherein the measurement matrix is ​​used to characterize a conversion relationship between the light field image and the target image corresponding to the light field image; Wherein, for each of the light field images, the target image corresponding to the light field image is determined by the following formula: P(i)=ΦX(i) Among them, P(i) represents the light field image, X(i) represents the target image, and Φ represents the measurement matrix.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device executes the steps of the high-resolution light field image reconstruction method according to any one of claims 1 to 4.

7. An electronic device comprising a memory, a processor, and a program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the steps of the high-resolution light field image reconstruction method according to any one of claims 1 to 4 are implemented.

Images