Imaging device and vehicle
By splitting the light into visible and invisible beams using an optical waveguide, the problem of reduced light intensity caused by beam splitting is solved, achieving high-quality imaging and environmental information acquisition, and improving the user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG GEELY HLDG GRP CO LTD
- Filing Date
- 2026-03-12
- Publication Date
- 2026-05-08
AI Technical Summary
Existing optical waveguide-based imaging devices suffer from reduced light intensity and impaired imaging performance due to the splitting of the light beam into a transmission beam and a transmission beam.
An optical waveguide is used to split the incident light into two beams, visible light and invisible light. The beams are transmitted through the optical waveguide to the target location and the image acquisition module for imaging processing. The image output module projects the target image onto the optical waveguide and transmits it to the target location.
It improves imaging performance, provides high-quality image display, and enhances the acquisition of environmental information and user experience.
Smart Images
Figure CN121995640A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of image technology, and in particular relates to an imaging device and a vehicle. Background Technology
[0002] Existing imaging devices based on optical waveguides mainly couple light rays through a coupling device, splitting the light into a first beam that is transmitted and a second beam that is transmitted through the optical waveguide. Since the two beams are splits of the same ray, the intensity of both the first beam (i.e., the projected beam) and the second beam (i.e., the imaging beam) will be reduced, affecting the imaging effect. Summary of the Invention
[0003] This application provides an imaging device and a vehicle that can improve imaging results.
[0004] In a first aspect, embodiments of this application provide an imaging device, the device comprising: an optical waveguide, an image acquisition module, and an image output module, wherein the image acquisition module and the image output module are communicatively connected; The optical waveguide is used to split the incident light into a first beam and a second beam, and to transmit the first beam to the target position and the second beam to the image acquisition module. The first beam is visible light and the second beam is invisible light. The image acquisition module is used to image and process the second beam to obtain a target image; The image output module is used to project the target image onto the optical waveguide, and the optical waveguide transmits the target image to the target location.
[0005] Secondly, embodiments of this application also provide a vehicle including the imaging device described in the first aspect.
[0006] The imaging device provided in this embodiment has two advantages. First, the first beam obtained by optical waveguide splitting is visible light, and the second beam is invisible light. After such splitting, the light intensity of the first beam is still relatively high. After being transmitted to the target position through the optical waveguide, it will not affect the visual effect when viewed by the human eye. Second, the image acquisition module can image and process the second beam to obtain the target image. The target image can be projected back onto the optical waveguide and transmitted to the target position by the optical waveguide. The target image can include more environmental information, which can improve the imaging effect and provide users with high-quality image display effect, which is beneficial to improving the user experience. Attached Figure Description
[0007] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0008] Figure 1 This is a schematic diagram of the imaging device provided in the embodiments of this application.
[0009] Figure label: 1-Optical waveguide; 11-Transparent substrate; 12-Color filter; 13-First optical coupling module; 14-Second optical coupling module; 15-Third optical coupling module; 21-Imaging submodule; 22-Image enhancement submodule; 212-Focusing unit; 213-Photoelectric sensor; 3-Image output module; 31-Display; 32-Collimation device; 4-Incident light; 41-First beam; 42-Second beam. Detailed Implementation
[0010] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.
[0011] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.
[0012] In all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. Additionally, when embodiments of this application require access to sensitive personal information, separate permission or consent from the user is obtained through pop-ups or redirects to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments obtained.
[0013] To address the problems of the prior art, embodiments of this application provide an imaging device, such as... Figure 1 As shown, the imaging device provided in this application embodiment includes: an optical waveguide 1, an image acquisition module (including an imaging submodule 21 and an image enhancement submodule 22) and an image output module 3, wherein the image acquisition module and the image output module 3 are communicatively connected; The optical waveguide 1 is used to split the incident light 4 to obtain a first beam 41 and a second beam 42, and to transmit the first beam 41 to the target position and the second beam 42 to the image acquisition module. The first beam 41 is visible light and the second beam 42 is invisible light. The image acquisition module is used to image and process the second beam 42 to obtain a target image; The image output module 3 is used to project the target image onto the optical waveguide 1, and the optical waveguide 1 transmits the target image to the target location.
[0014] Specifically, the optical waveguide 1 is made of transparent material. The optical waveguide 1 splits the incident light 4 to obtain a first beam 41 and a second beam 42. For example, the first beam 41 is visible light and the second beam 42 is infrared polarized light.
[0015] The first beam 41 and the second beam 42 are transmitted in the optical waveguide 1, which is made of a transparent material. The optical waveguide 1 controls the transmission of the first beam 41 to the target position and also controls the transmission of the second beam 42 to the image acquisition module. The image acquisition module acquires the second beam 42 and performs imaging, and processes the image, such as image reconstruction, to improve image quality and obtain the target image. The target image is projected onto the optical waveguide 1 by the image output module 3, and then transmitted to the target position, which can be the position of the human eye.
[0016] The incident light 4 can be a beam of light containing different information to be displayed. The information to be displayed can be set according to actual usage requirements. It can be virtual information or real-world information. This application does not limit this.
[0017] The imaging device provided in this application embodiment can be applied to virtual reality scenarios.
[0018] In this embodiment, since the first beam 41 obtained by the optical waveguide 1 is visible light and the second beam 42 is invisible light, on the one hand, after such beam splitting, the light intensity of the first beam 41 is still relatively high. After being transmitted to the target position through the optical waveguide 1, it will not affect the visual effect when viewed by the human eye. On the other hand, the image acquisition module can image and process the second beam 42 to obtain the target image. The target image can be projected back onto the optical waveguide 1, and the optical waveguide 1 will transmit the target image to the target position. The target image can include more environmental information, which can improve the imaging effect and thus provide users with a high-quality image display effect, which is beneficial to improving the user experience.
[0019] In one embodiment of this application, the optical waveguide 1 includes a transparent substrate 11, and a dichroic filter 12 and a first optical coupling module 13 respectively disposed on opposite sides of the transparent substrate 11; The dichroic filter 12 is used to split the incident light 4 to obtain a first beam 41 and a second beam 42. The transparent substrate 11 is used to transmit the first light beam 41 to the first optical coupling module 13 and the second light beam 42 to the image acquisition module; The first optical coupling module 13 is used to couple the first beam 41 and the target image to the target position.
[0020] In the above, the incident light 4 is split by the dichroic filter 12 to obtain a first beam 41 and a second beam 42. For example, the first beam 41 is visible light and the second beam 42 is infrared polarized light. When imaging is performed based on the second beam 42, a deeper environmental perception can be achieved, providing more environmental information.
[0021] The first beam 41 and the second beam 42 are transmitted in a transparent substrate 11, which is made of a transparent material. The transparent substrate 11 can control the transmission of the first beam 41 to the first optical coupling module 13, and also controls the transmission of the second beam 42 to the image acquisition module. The image acquisition module acquires the second beam 42 and performs imaging, and processes the image, for example, image reconstruction processing, to improve image quality, and then obtains the target image. The target image is projected onto the transparent substrate 11 by the image output module 3, and the transparent substrate 11 transmits the target image to the first optical coupling module 13. The first optical coupling module 13 couples the first beam 41 and the target image to (i.e., couples them out to) the target position, which can be the position of the human eye.
[0022] The dichroic filter 12 is a wavelength-selective optical element. Its design is based on "thin-film interference" and "differences in material optical properties". By depositing multiple dielectric films on the transparent substrate 11, wavelength selection is achieved by utilizing the "thin-film interference effect": when visible light is incident, the multilayer film produces "constructive interference", allowing visible light to pass through with almost no loss; when infrared light is incident, the multilayer film produces "destructive interference", blocking the transmission of infrared light and instead reflecting it to a preset angle.
[0023] Specifically, the dichroic filter 12 can be formed by depositing multiple dielectric films on the transparent substrate 11. For example, after depositing at least one of optical borosilicate glass, quartz glass, sapphire, and optical plastic on the optical waveguide 1, at least one of the following films is then deposited: high refractive index titanium dioxide, tantalum pentoxide, zirconium oxide, low refractive index silicon dioxide, magnesium fluoride, and aluminum fluoride, thereby obtaining the dichroic filter 12.
[0024] In another embodiment of this application, the optical waveguide 1 further includes a second optical coupling module 14 disposed on the transparent substrate 11; the second optical coupling module 14 is used to couple a second light beam 42 transmitted through the transparent substrate 11 to the image acquisition module. The second light beam 42 is coupled out to the image acquisition module through the second optical coupling module 14. The image acquisition module includes an imaging submodule 21 and an image enhancement submodule 22; The imaging submodule 21 is used to process the received second beam 42 to obtain an initial image; The image enhancement submodule 22 is used to perform image enhancement processing on the initial image to obtain the target image.
[0025] Specifically, the imaging submodule 21 includes a focusing unit 212 and a photoelectric sensor 213 arranged at relative intervals. The focusing unit 212 is used to focus the received second beam 42 to obtain a third beam. The photoelectric sensor 213 is used to convert the third beam into an electrical digital signal and obtain an initial image based on the electrical digital signal. The second beam 42 is infrared polarized light, which includes depth information in the environment. When using the second beam 42 for infrared imaging, a deeper environmental perception can be achieved, allowing the initial image to include more environmental information.
[0026] During the transmission of the second beam 42 through the transparent substrate 11, and during the coupling process from the second optical coupling module 14 to the image acquisition module, the light signal of the second beam 42 may be affected, resulting in stray light in the initial image. For example, the initial image may have problems such as patchy light spots, low contrast, and high noise substrate, which affect the imaging quality and make it impossible to obtain clear environmental information.
[0027] To improve image quality, an image enhancement submodule 22 is used to enhance the initial image to obtain the target image. Specifically, the image enhancement submodule 22 is used for: The initial image is enhanced by a target generator to obtain the target image. The target generator has the ability to remove stray light from the initial image. The target generator is obtained through adversarial training.
[0028] In this embodiment, a target generator is used to remove stray light from the initial image, thereby improving the quality of the target image.
[0029] In another embodiment of this application, the training process of the target generator includes: Obtain a first sample set, which includes multiple first images and a reference image corresponding to each first image. The first image includes stray light, and the reference image of the first image does not include stray light. The initial generator is obtained by training the base model using the first sample set; Obtain a second sample set, which includes multiple second images and a label corresponding to each second image. The multiple second images include images captured by a camera and images generated by the initial generator. The label is used to identify whether the second image is captured by the camera or generated by the initial generator. The initial discriminator is trained using the second sample set to obtain the discrimination result for each of the second images; Based on the discrimination result of each second image and the label of the second image, the parameters of the initial generator are adjusted to obtain the target generator.
[0030] In the above, the first sample set can be obtained by imaging the imaging device of this application and an independent imaging device in the same scene or under the same brightness conditions. The first image is obtained by the photoelectric sensor 213 of the imaging device, and the second image is obtained by the independent imaging device. The first image includes stray light, and the second image does not include stray light. In multiple scenes or under multiple brightness conditions, multiple first images and corresponding reference images can be acquired.
[0031] The base model is trained in a supervised manner using the first sample set to obtain an initial generator, which has a certain ability to eliminate stray light. The base model can refer to an untrained generator.
[0032] The basic model can include an encoder and a decoder. The encoder consists of four convolutional layers (Conv2D + ReLU), each followed by MaxPooling to progressively extract image features (such as the distribution characteristics of stray light regions). An attention mechanism (such as CBAM) is added to the encoder: channel attention and spatial attention are added to the high-dimensional feature layers, allowing the model to automatically focus on fixed regions where stray light is concentrated. A bottleneck layer is then added: one or two residual blocks are added at the end of the encoder to enhance feature extraction capabilities (alleviating gradient vanishing).
[0033] The decoder, symmetrical to the encoder, upsamples each layer using transposed convolutions (Conv2DTranspose) and combines this with skip connections in the corresponding encoder layers to recover image details (such as edges and textures). A contrast enhancement module is added at the end of the decoder, adjusting the brightness and contrast of the output image through 1×1 convolutions (e.g., learning a gain factor and bias to linearly adjust the output pixels). Finally, the output layer is connected, activated with Conv2D(3, kernel_size=1) + Tanh, outputting an image of the same size as the input.
[0034] In one embodiment of this application, a base model is trained using a first sample set to obtain an initial generator, including: The image enhancement process is performed on each of the first images using the base model to obtain the enhanced image corresponding to each of the first images; The average absolute error loss is calculated based on the enhanced image of each of the first images and the reference image corresponding to the first image to obtain the first sub-loss value; The structural similarity loss is calculated based on the enhanced image of each of the first images and the reference image corresponding to the first image to obtain the second sub-loss value; The second sub-loss value is multiplied by a preset coefficient and then added to the first sub-loss value to obtain the second loss value; The parameters of the base model are adjusted based on the second loss value to obtain the initial generator.
[0035] Specifically, the Mean Absolute Error (MAE) loss preserves edges and reduces blur, while the Structural Similarity (SSIM) loss enhances structural similarity and avoids contrast loss. Combining the first and second sub-loss values yields the second loss value. Adjusting the parameters of the base model using this second loss value allows the image generated by the initial generator to preserve edges, reduce blur, enhance structural similarity, avoid contrast loss, and improve image quality.
[0036] In the above, the preset coefficient can be used to adjust the proportion of the first sub-loss value in the second loss value. The larger the proportion, the more the image generated by the initial generator focuses on enhancing structural similarity. For example, the preset coefficient can be 0.1, or other values between 0 and 1 can be used.
[0037] A second sample set is constructed using an initial generator. For example, images are generated using the initial generator and added to the second sample set. Images captured by a camera are also added to the second sample set. Furthermore, the second images in the second sample set are tagged, with the tags indicating whether the second image was captured by a camera or generated by the initial generator.
[0038] The initial discriminator can adopt a PatchGAN structure (outputting an N×N probability map, where each position represents the true probability of the corresponding image patch), rather than global binary classification, which focuses more on the realism of local details. Architecture details: Use 4 convolutional layers (Conv2D + LeakyReLU), with a stride of 2 for each layer (to achieve downsampling), gradually compressing the spatial dimensions. Add BatchNorm after each convolutional layer (except the first layer) for stable training. The output layer uses Conv2D (1, kernel_size=1) + Sigmoid activation, outputting a 32×32 or 16×16 probability map. Other structures can also be used for the initial discriminator; no restrictions are imposed here.
[0039] During the adversarial training phase, the parameters of the initial generator's feature extraction part are fixed, and an initial discriminator is introduced for adversarial training. Through the game between the generator and the discriminator, the generator's ability to reproduce details (such as edges and textures) is optimized, while retaining the global stray light elimination ability learned in the self-supervised training phase.
[0040] The second sample set is used to train the initial discriminator, that is, the second image is input into the initial discriminator, and the initial discriminator judges the second image to obtain the judgment result of the second image. The judgment result includes whether the second image determined by the initial discriminator is generated by the generator or captured by the camera.
[0041] In another embodiment of this application, the parameters of the initial generator are adjusted based on the discrimination result of each second image and the label of the second image to obtain a target generator, including: Based on the discrimination result of each second image and the label of the second image, a first loss value is determined; If the change between the first loss value obtained in the current iteration and the first loss value obtained in the previous iteration is greater than a preset threshold, then the parameters of the initial generator and the initial discriminator are adjusted according to the first loss value to obtain a temporary generator and a temporary discriminator. Using the temporary generator as the initial generator and the temporary discriminator as the initial discriminator, proceed to the step of obtaining the second sample set. If the change between the first loss value obtained in the current iteration and the first loss value obtained in the previous iteration is less than or equal to a preset threshold, then the temporary generator obtained in the last iteration is used as the target generator.
[0042] Specifically, the first loss value is calculated using the adversarial loss function. The adversarial loss function is the core loss function in Generative Adversarial Network (GAN) training; it quantifies the generator's ability to deceive the discriminator. Optimizing the adversarial loss function achieves two goals simultaneously: enabling the generator to produce more realistic images and allowing the discriminator to better distinguish between real and generated images. During GAN training, the generator and discriminator learn adversarially by minimizing and maximizing the same loss function.
[0043] By continuously optimizing the process in a loop, the initial discriminator becomes unable to distinguish between the image generated by the initial generator and the real image (i.e., the image captured by the camera).
[0044] During adversarial training, the initial generator is first fixed, and the initial discriminator is optimized to maximize the adversarial loss function, thereby improving the discriminative ability of the initial discriminator; then the initial discriminator is fixed, and the initial generator is optimized to minimize the adversarial loss function, thereby improving the generation ability of the initial generator.
[0045] The target generator obtained through the above adversarial training process not only has the ability to remove stray light from the image, but also makes the generated image close to the real image, which greatly improves the image enhancement capability of the target generator and makes the image generated by the target generator of high quality.
[0046] In addition to removing stray light from the initial image using the methods described above, the image enhancement submodule 22 can also use the following methods: Based on the imaging characteristics of optical waveguide 1, an imaging model corresponding to optical waveguide 1 is constructed; an optimization function is determined; the target image is obtained by optimizing and solving the optimization function, or the target image is obtained based on algorithms such as deep learning.
[0047] or, The imaging device and an independent imaging device of the same specifications perform multiple synchronous imaging operations on the same specific scene with different exposure times and ambient brightness. Two images under the same conditions are registered in position. After feature registration, the brightness of the two images is registered so that the high-brightness and low-brightness areas of the two images are approximately at the same scale. Differential calculations are performed on the images after the two registrations to calculate the influence area of stray light, the average brightness, and the maximum brightness of the stray light in both images. The stray light intensity under different brightness and exposure time conditions is fitted to obtain the response curve of stray light to brightness information and exposure time. Depending on the different application scenarios of the imaging device, the stray light in the initial image can be corrected pixel-by-pixel using the response curve of brightness and exposure time to obtain the target image.
[0048] or, The imaging device is systematically modeled, and then ray tracing is performed using optical simulation software. The model meticulously considers factors such as the surface roughness, reflectivity, absorptivity of optical elements, and the scattering characteristics within the system to simulate the propagation and scattering process of stray light within the optical system. Through extensive ray tracing calculations, the energy distribution of stray light reaching the detector surface is statistically analyzed to establish a correction matrix for stray light. The original measurement signal is multiplied by the spectral stray light correction matrix using simple matrix multiplication to obtain the spectral stray light correction signal. Alternatively, a mathematical model of stray light can be obtained using a polynomial curve fitting method based on least squares. This mathematical model is then used for automatic region division and polynomial fitting to effectively correct stray light in the initial image, resulting in the target image.
[0049] In another embodiment of this application, the image output module 3 includes a display 31 and a collimation device 32; The display 31 is used to project the target image; The collimation device 32 is used to collimate the light beam projected by the display 31 to obtain a collimated light beam, and to project the collimated light beam onto the transparent substrate 11.
[0050] Specifically, the display 31 emits a light beam carrying the target image, and the collimating device 32 collimates the light beam projected by the display 31 to obtain a collimated beam. This ensures that the collimated beam projected onto the transparent substrate 11 is parallel, improving the image display effect after the collimated beam reaches the target position. The specific form of the collimating device 32 can be found in related technologies, such as lenses, lens groups, mirrors, and diffractive optical elements; this application does not limit its application to these methods.
[0051] In the above, the optical waveguide 1 further includes a third optical coupling module 15 disposed on the transparent substrate 11; the third optical coupling module 15 is used to couple the collimated beam into the transparent substrate 11.
[0052] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described as examples. However, the method process of this application is not limited to the specific steps described. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.
[0053] It should be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.
[0054] This application also provides a vehicle that includes the imaging device described above.
[0055] The foregoing flowcharts and / or block diagrams of methods, apparatus (systems) according to embodiments of the present disclosure have described various aspects of the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to create a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block of the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.
[0056] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.
Claims
1. An imaging device, characterized in that, The device includes: an optical waveguide, an image acquisition module, and an image output module, wherein the image acquisition module and the image output module are communicatively connected. The optical waveguide is used to split the incident light into a first beam and a second beam, and to transmit the first beam to the target position and the second beam to the image acquisition module. The first beam is visible light and the second beam is invisible light. The image acquisition module is used to image and process the second beam to obtain a target image; The image output module is used to project the target image onto the optical waveguide, and the optical waveguide transmits the target image to the target location.
2. The apparatus according to claim 1, characterized in that, The optical waveguide includes a transparent substrate, and dichroic filters and a first optical coupling module respectively disposed on opposite sides of the transparent substrate; The dichroic filter is used to split the incident light to obtain a first beam and a second beam. The transparent substrate is used to transmit the first light beam to the first optical coupling module and the second light beam to the image acquisition module; The first optical coupling module is used to couple the first light beam and the target image to the target position.
3. The apparatus according to claim 2, characterized in that, The optical waveguide also includes a second optical coupling module disposed on the transparent substrate; The second optical coupling module is used to couple the second light beam transmitted through the transparent substrate to the image acquisition module.
4. The apparatus according to claim 1, characterized in that, The image acquisition module includes an imaging submodule and an image enhancement submodule; The imaging submodule is used to process the received second beam to obtain an initial image; The image enhancement submodule is used to perform image enhancement processing on the initial image to obtain the target image.
5. The apparatus according to claim 4, characterized in that, The imaging submodule includes focusing units and photoelectric sensors arranged at relatively intervals; The focusing unit is used to focus the received second beam to obtain a third beam; The photoelectric sensor is used to convert the third light beam into an electrical digital signal and obtain an initial image based on the electrical digital signal.
6. The apparatus according to claim 2, characterized in that, The image output module includes a projection element and a collimation device; The projection element is used to project the target image; The collimation device is used to collimate the light beam projected by the projection element to obtain a collimated light beam, and to project the collimated light beam onto the transparent substrate.
7. The apparatus according to claim 6, characterized in that, The optical waveguide also includes a third optical coupling module disposed on the transparent substrate; The third optical coupling module is used to couple the collimated beam into the transparent substrate.
8. The apparatus according to any one of claims 2-7, characterized in that, The color separation filter is formed by depositing multiple layers of dielectric film on the transparent substrate.
9. The apparatus according to any one of claims 2-7, characterized in that, The second beam is infrared polarized light.
10. The apparatus according to claim 4, characterized in that, The image enhancement submodule is used for: The initial image is enhanced by a target generator to obtain the target image. The target generator has the ability to remove stray light from the initial image. The target generator is obtained through adversarial training.
11. The apparatus according to claim 10, characterized in that, The training process of the target generator includes: Obtain a first sample set, which includes multiple first images and a reference image corresponding to each first image. The first image includes stray light, and the reference image of the first image does not include stray light. The initial generator is obtained by training the base model using the first sample set; Obtain a second sample set, which includes multiple second images and a label corresponding to each second image. The multiple second images include images captured by a camera and images generated by the initial generator. The label is used to identify whether the second image is captured by the camera or generated by the initial generator. The initial discriminator is trained using the second sample set to obtain the discrimination result for each of the second images; Based on the discrimination result of each second image and the label of the second image, the parameters of the initial generator are adjusted to obtain the target generator.
12. The apparatus according to claim 11, characterized in that, Based on the discrimination result and label of each of the second images, the parameters of the initial generator are adjusted to obtain the target generator, including: Based on the discrimination result of each second image and the label of the second image, a first loss value is determined; If the change between the first loss value obtained in the current iteration and the first loss value obtained in the previous iteration is greater than a preset threshold, then the parameters of the initial generator and the initial discriminator are adjusted according to the first loss value to obtain a temporary generator and a temporary discriminator. Using the temporary generator as the initial generator and the temporary discriminator as the initial discriminator, proceed to the step of obtaining the second sample set. If the change between the first loss value obtained in the current iteration and the first loss value obtained in the previous iteration is less than or equal to a preset threshold, then the temporary generator obtained in the last iteration is used as the target generator.
13. The apparatus according to claim 11, characterized in that, The initial generator is obtained by training the base model using the first sample set, including: The image enhancement process is performed on each of the first images using the base model to obtain the enhanced image corresponding to each of the first images; The average absolute error loss is calculated based on the enhanced image of each of the first images and the reference image corresponding to the first image to obtain the first sub-loss value; The structural similarity loss is calculated based on the enhanced image of each of the first images and the reference image corresponding to the first image to obtain the second sub-loss value; The second sub-loss value is multiplied by a preset coefficient and then added to the first sub-loss value to obtain the second loss value; The parameters of the base model are adjusted based on the second loss value to obtain the initial generator.
14. A vehicle, characterized in that, The vehicle includes an imaging device as described in any one of claims 1-13.