Vehicle preview image generation method and its device, equipment and medium

By acquiring customized information of the display scene and recognizing the rendering effects of vehicle images in similar scenes, adjusting the angle and lighting effects of the vehicle's 3D model, and generating a vehicle product preview image, the shortcomings of vehicle display in different scene environments in online car viewing services are solved, and a naturally integrated vehicle product preview is achieved.

CN117078835BActive Publication Date: 2024-11-19GUANGZHOU DIGITAL COLLECTION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310887162.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-07-18
Publication Date
2024-11-19
Estimated Expiration
2043-07-18

AI Technical Summary

Technical Problem

Existing online car viewing services and live streaming platforms do not adequately showcase the appearance of vehicles in different scenarios, failing to meet the needs of users and streamers.

Method used

By acquiring custom information about the display scene, identifying the rendering effects of vehicle images in similar scenes, adjusting the angle and lighting effects of the vehicle's 3D model, generating a vehicle product preview image, and enabling the vehicle to be displayed in a custom scene.

Benefits of technology

It provides a natural integration of vehicle products into user-defined scene environments, reducing the visual incongruity after replacement and meeting the display needs of users and broadcasters.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117078835B_ABST
    Figure CN117078835B_ABST
Patent Text Reader

Abstract

The present application relates to a method for generating a vehicle preview image and its device, equipment, and medium, the method comprising: obtaining a similar scene vehicle image corresponding to customized information of a display scene; performing vehicle rendering effect recognition on a vehicle region image in the similar scene vehicle image, and obtaining vehicle rendering information corresponding to the vehicle region image; obtaining a vehicle stereo model of a vehicle product, adjusting the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle region image according to the vehicle rendering information, and performing stereo model rendering on the vehicle stereo model to generate a vehicle model image; replacing the vehicle region image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image. The present application can provide users with a vehicle preview image generation service that can customize the environment of a car display scene, and use the rendering effect of the vehicle in the similar scene image to perform a preview of the vehicle stereo model rendering, thereby improving the display effect of the vehicle in the preview image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of image rendering generation, and in particular to a vehicle preview image generation method and its corresponding device, a vehicle preview image generation device, and a computer-readable storage medium. Background Art

[0002] With the development of the Internet, many Internet platforms provide online car viewing services so that users can browse the vehicle products sold on the platform through the Internet. The existing online car viewing services generally only provide conventional three-view photos or stereo models of vehicle products for users to browse, and users cannot understand the appearance of the vehicle in different scenarios. In addition, in the existing online live broadcast business, the anchor user can also use the online car viewing service to display the vehicle products for sale in the live broadcast room. However, for the online live broadcast business, in order to better introduce the vehicle products they sell to the audience users, the anchor user needs to reflect the product advantages of the vehicle products in different scenarios. The vehicle photos or stereo models provided only through the online car viewing service cannot meet the needs of the anchor users well. Therefore, regarding the problem of how to display the appearance of the vehicle in different scenarios online, the applicant has made corresponding explorations in order to solve this problem. Summary of the invention

[0003] The purpose of the present application is to solve the above-mentioned problems and to provide a vehicle preview image generation method and its corresponding device, vehicle preview image generation equipment, computer-readable storage medium, and computer program product.

[0004] According to one aspect of the present application, a method for generating a vehicle preview image is provided, comprising the following steps:

[0005] Obtaining customized information of a display scene, and obtaining a vehicle image of a similar scene corresponding to the customized information of the display scene;

[0006] Performing vehicle rendering effect recognition on the vehicle region image in the similar scene vehicle image, and obtaining vehicle rendering information corresponding to the vehicle region image;

[0007] Acquire a vehicle stereo model of a target vehicle product, adjust a display angle of the vehicle stereo model to a vehicle display angle represented by the vehicle region image according to the vehicle rendering information, and perform stereo model rendering on the vehicle stereo model to generate a vehicle model image;

[0008] The vehicle area image in the similar scene vehicle image is replaced with the vehicle model image to generate a vehicle product preview image acting on a target vehicle product.

[0009] In a further embodiment, the steps of obtaining the display scene custom information and obtaining the similar scene vehicle image corresponding to the display scene custom information include the following steps:

[0010] Acting on the display scene customization event of the target vehicle product, recognizing the display scene customization information from the user's voice;

[0011] In response to an editing completion event of a display scene custom text editing control in a graphical user interface, a custom text corresponding to the display scene custom text editing control is obtained, and the custom text is used as display scene custom information acting on a target vehicle product.

[0012] In a preferred embodiment, the steps of obtaining the display scene custom information and obtaining the similar scene vehicle image corresponding to the display scene custom information include the following steps:

[0013] Performing text preprocessing on the custom text of the display scene custom information, and extracting the semantic feature vector of the custom text after the text preprocessing;

[0014] Determining the semantic feature similarity between the semantic feature vector and the semantic feature vectors of each scene vehicle image in the scene vehicle image library;

[0015] One or more scene vehicle images whose semantic feature similarity exceeds a preset threshold are used as similar scene vehicle images.

[0016] In a further embodiment, the step of performing vehicle rendering effect recognition on the vehicle region image in the similar scene vehicle image and obtaining vehicle rendering information corresponding to the vehicle region image includes the following steps:

[0017] Identifying a vehicle region image in the similar vehicle scene image;

[0018] Calling a vehicle angle recognition model trained to convergence, and identifying vehicle display angle data of a vehicle object in the vehicle area image based on an image feature vector corresponding to the vehicle area image;

[0019] Calling a vehicle light and shadow recognition model that has been trained to convergence, and identifying light and shadow rendering data acting on a vehicle object in the vehicle area image based on an image feature vector corresponding to the vehicle area image;

[0020] Generate vehicle rendering information including the vehicle display angle data and the light and shadow rendering data.

[0021] In a further embodiment, the step of obtaining a vehicle stereo model of a target vehicle product, adjusting the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and performing stereo model rendering on the vehicle stereo model to generate a vehicle model image includes the following steps:

[0022] According to the vehicle identification of the target vehicle product, obtaining the target vehicle three-dimensional model corresponding to the vehicle identification from the vehicle three-dimensional model library;

[0023] According to the vehicle display angle data included in the vehicle rendering information, adjusting the vehicle display angle of the target vehicle stereoscopic model by an equal amount;

[0024] According to the light and shadow rendering data included in the vehicle rendering information, the model light and shadow effects of the target vehicle stereoscopic model whose display angle has been adjusted are rendered to generate a vehicle model image having the target vehicle stereoscopic model.

[0025] In a further embodiment, the step of replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on the target vehicle product includes the following steps:

[0026] Crop and remove the vehicle region image in the similar scene vehicle image, and use the cropped and removed region in the similar scene vehicle image as the vehicle synthesis region;

[0027] Generating a vehicle model image corresponding to the vehicle stereo model for which the stereo model rendering has been completed, and adjusting the image size of the vehicle model image according to the area size of the vehicle synthesis area;

[0028] The vehicle model image that has completed image resizing is synthesized into the vehicle synthesis area in the similar scene vehicle image to generate a vehicle product preview image.

[0029] In a further embodiment, after the step of replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on the target vehicle product, the following steps are included:

[0030] Outputting the vehicle product preview image to a preview image control of a graphical user interface for display;

[0031] In response to a three-dimensional model display instruction acting on the preview image control, obtaining a vehicle three-dimensional model corresponding to the vehicle product preview image;

[0032] A three-dimensional model interaction window is displayed in a graphical user interface, and the vehicle three-dimensional model is output to the three-dimensional model interaction window.

[0033] According to another aspect of the present application, a vehicle preview image generating device is provided, comprising:

[0034] A similar image acquisition module is used to acquire the display scene custom information and acquire the similar scene vehicle image corresponding to the display scene custom information;

[0035] A rendering information acquisition module, used to identify the vehicle rendering effect of the vehicle area image in the similar scene vehicle image, and obtain the vehicle rendering information corresponding to the vehicle area image;

[0036] A model image generation module is used to obtain a vehicle stereo model of a target vehicle product, adjust the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and perform stereo model rendering on the vehicle stereo model to generate a vehicle model image;

[0037] The vehicle preview image generating module is used to replace the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on a target vehicle product.

[0038] According to another aspect of the present application, a vehicle preview image generating device is provided, comprising a central processing unit and a memory, wherein the central processing unit is used to call and run a computer program stored in the memory to execute the steps of the vehicle preview image generating method described in the present application.

[0039] According to another aspect of the present application, a computer-readable storage medium is provided, which stores a computer program implemented according to the vehicle preview image generating method in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the method are executed.

[0040] According to another aspect of the present application, a computer program product is provided, including a computer program / instruction, which implements the steps of the method described in any one of the embodiments of the present application when executed by a processor.

[0041] The present application provides users with a vehicle preview image generation service. Users can customize the scene environment for displaying vehicle products in the vehicle product preview image, so that the vehicle products in the vehicle product preview image can be displayed in the scene environment customized by the user. After the user completes editing the display scene customization information that acts on the scene environment displayed by the vehicle product, a similar scene vehicle image corresponding to the display scene customization information will be determined. The scene environment displayed in the similar scene vehicle image corresponds to the scene environment customized and edited by the user, and then, with reference to the rendering effect of the vehicle object displayed in the vehicle area image in the similar scene vehicle image, the same rendering is performed on the vehicle three-dimensional model specified by the user to display the vehicle product, including adjusting the display angle of the vehicle three-dimensional model to the display angle of the vehicle object in the similar scene vehicle image, and performing light and shadow rendering on the vehicle three-dimensional model according to the light and shadow effects of the vehicle object, thereby generating a vehicle model image with the rendered vehicle three-dimensional model to render the similar scene vehicle image in the vehicle area image. The vehicle area image displaying the vehicle object is replaced with the vehicle model image, that is, the vehicle object originally displayed in the similar scene vehicle image is replaced with the vehicle product, and the vehicle three-dimensional model of the vehicle product is rendered identically based on the rendering effect of the vehicle object originally displayed in the similar scene vehicle image, so that the vehicle product displayed in the similar scene vehicle image after replacement can be naturally integrated into the displayed scene environment, thereby alleviating the visual incongruity caused by replacing the vehicle image, so that the similar scene vehicle image replaced with the vehicle product can be used as a vehicle product preview image for displaying the vehicle product in the user-defined scene environment, so that users can display the vehicles they want to display in different scene environments when viewing cars online, and browse the vehicle appearance in different scene environments. As for the anchor users conducting live broadcasts on the Internet, by displaying the vehicle products they sell in the expected scene environment, the audience users with purchasing intentions in the live broadcast room can understand the driving experience of the imagined vehicle products in different scene environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0043] Figure 1 A schematic diagram of a typical network deployment architecture related to implementing the technical solution of this application;

[0044] Figure 2 A schematic diagram of a typical embodiment of a method for generating a vehicle preview image of the present application;

[0045] Figure 3 A flowchart of a specific embodiment of obtaining customized information of a display scene in this application;

[0046] Figure 4A flowchart of a specific embodiment of obtaining a similar scene vehicle image corresponding to the display scene custom information in the embodiment of the present application;

[0047] Figure 5 A flow chart of a specific embodiment of the present application for identifying a vehicle region image in a vehicle image of a similar scene, and performing vehicle rendering effect recognition on the vehicle region image at the identified location;

[0048] Figure 6 A flow chart of a flow chart of a specific embodiment of the present application for rendering a vehicle stereo model according to vehicle rendering information to generate a vehicle model image;

[0049] Figure 7 A schematic diagram of a flow chart formed by a specific embodiment of replacing a vehicle region image in a similar scene vehicle image with a vehicle model image to generate a vehicle product preview image in the present application;

[0050] Figure 8 A schematic diagram of a flow chart formed by a specific embodiment of outputting a preview image of a vehicle product to a graphical user interface for control interaction in this application;

[0051] Fig. 9 This is a principle block diagram of a typical embodiment of a vehicle preview image generating device of the present application;

[0052] Fig.10 This is a basic structural block diagram of a vehicle preview image generating device according to an embodiment of the present application. DETAILED DESCRIPTION

[0053] The "server" referred to in this application can also be extended to the case of service clusters. According to the network deployment principle understood by those skilled in the art, the servers should be logically divided. In physical space, these servers can be independent of each other but can be called through interfaces, or integrated into a physical computer or a set of computer clusters. Those skilled in the art should understand this flexibility, and should not use it to restrict the implementation of the network deployment method of this application.

[0054] Those skilled in the art should be aware that, although the various methods of the present application are described based on the same concept and thus present commonality to each other, unless otherwise specified, these methods can be independently executed. Similarly, for each embodiment disclosed in the present application, they are all proposed based on the same inventive concept, therefore, concepts with the same expression, and concepts that are appropriately changed for convenience despite different expressions, should be understood as equivalent.

[0055] Unless the mutually exclusive relationship between the embodiments to be disclosed in this application is explicitly stated, the relevant technical features involved in each embodiment can be cross-combined to flexibly construct a new embodiment, as long as such combination does not deviate from the creative spirit of this application and can meet the needs of the prior art or solve certain deficiencies in the prior art. Those skilled in the art should be aware of this flexibility.

[0056] See also Figure 1 , the hardware foundation required for the implementation of the relevant technical solutions of this application can be deployed according to the architecture shown in the figure. The server 80 referred to in this application is deployed in the cloud. As an online server, it can be responsible for further connecting relevant data servers and other servers that provide relevant support, etc., so as to form a logically related service cluster to provide services for related terminal devices such as the smartphone 81 and personal computer 82 shown in the figure or a third-party server (not shown). The smartphone and personal computer can access the Internet through a well-known network access method and establish a data communication link with the server 80 in the cloud to run terminal applications related to the services provided by the server.

[0057] For the server, the application is usually constructed as a service process, opening the corresponding program interface for remote calls by applications running on various terminal devices. The relevant technical solutions suitable for running on the server in this application can be implemented in the server in this way.

[0058] The application described herein refers to an application running on a server or terminal device. This application implements the relevant technical solutions of the present application in a programming manner. Its program code can be stored in a non-volatile storage medium recognizable by a computer in the form of computer executable instructions, and can be loaded into the memory by the central processing unit for execution. The relevant device of the present application is constructed by the operation of the application on the computer.

[0059] See also Figure 2 A vehicle preview image generation method of the present application, in its typical embodiment, comprises the following steps:

[0060] Step S11, obtaining the display scene custom information, and obtaining the similar scene vehicle image corresponding to the display scene custom information:

[0061] This method provides a vehicle preview image generation service for users. When using the vehicle preview image generation service, the user can select through text, voice or tags to customize the scene in which the target vehicle product to be displayed is located, so as to drive the vehicle preview image generation service to generate a vehicle preview image with a scene corresponding to the target vehicle product and the customized edited scene.

[0062] The display scene custom information is generally customized and edited by users who use the vehicle preview image generation service. Users can edit the scene environment for displaying vehicle products through voice, text or tags. After the user customizes the corresponding scene information, the display scene custom information that acts on the target vehicle product that the user wants to preview will be generated. Among them, if the user uses voice to edit the scene environment for displaying vehicle products, the terminal device or server where the user is located will recognize the corresponding custom text of the display scene from the user voice recorded by the user through a microphone or other audio input hardware, and use the custom text as the display scene custom information that the user acts on the vehicle product that he wants to preview; if the user uses text to edit the display scene custom information, the terminal device or server where the user is located will recognize the corresponding custom text of the display scene from the user voice recorded by the user through a microphone or other audio input hardware, and use the custom text as the user's custom information for the display scene that acts on the vehicle product that the user wants to preview; A custom text editing control for the display scene will be displayed in the graphical user interface of the terminal device. After the user enters text for describing the scene environment in the custom text editing control and completes the text editing, the custom text in the custom text editing control for the display scene will be obtained, and the custom text will be used as the custom information of the display scene acting on the vehicle product; if the user edits the custom information of the display scene by selecting a scene label for describing the environmental scene, the scene label may include location types such as desert, city or forest, weather types such as rainy, sunny or snowy, or time types such as morning, noon or evening. After the user selects the corresponding scene label, the custom information of the display scene including one or more scene identifiers selected by the user will be generated.

[0063] After obtaining the display scene customization information customized by the user to describe the scene environment for displaying vehicle products, one or more similar scene vehicle images similar to the scene environment represented by the display scene customization information will be obtained. The vehicle preview image generation service has multiple scene vehicle images of displaying vehicles in different scene environments. These scene vehicle images are generally stored in a scene vehicle image library for centralized management.

[0064] Regarding how to determine scene vehicle images that are similar to the customized information of the display scene as similar scene vehicle images, generally, text matching can be performed through preset similar scene vehicle image matching rules, or semantic feature similarity calculation can be performed through a neural network model trained to convergence.

[0065] Regarding the preset similar scene vehicle image matching rules, each scene vehicle image in the scene vehicle image library is mapped and stored with one or more corresponding scene labels, that is, each scene vehicle image in the scene vehicle image library is classified by different scene labels. For example, the scene label may include location types such as desert, city or forest, weather types such as rainy day, sunny day or snowy day, or time types such as morning, noon or evening, and the description of the scene vehicle image will be mapped and stored with various scene labels describing its scene environment. Therefore, the custom text contained in the display scene custom information can be mapped with each scene label in the scene vehicle image library through the preset regular expression. The scene vehicle image is mapped to the stored scene tags for keyword matching, or by extracting the text semantic feature vector of the custom text, calculating the text semantic similarity between the text semantic feature vector and the text semantic feature vector of the text represented by each scene tag, determining the target scene tag whose text semantic feature vector exceeds the similarity threshold, and determining the scene vehicle image with a preset number of target scene tags as the similar scene vehicle image, wherein the similarity threshold can be set within the range of 70% to 90%, and the preset number can be set to a number equal to or greater than 2. Of course, those skilled in the art can design flexibly and will not elaborate on it.

[0066] Regarding determining similar scene vehicle images corresponding to the display scene custom information through a neural network model trained to convergence, generally, text preprocessing is performed on the custom text of the display scene custom information, and a semantic feature vector of the custom text that has completed the text preprocessing is extracted, and then the semantic feature similarity between the semantic feature vector and the semantic feature vector of each scene vehicle image in the scene vehicle image library is determined, so as to use one or more scene vehicle images whose semantic feature similarity exceeds a preset threshold as similar scene vehicle images.

[0067] If there are multiple similar scene vehicle images determined based on the user-defined information of the display scene, these similar scene vehicle images can be output to the similar scene vehicle image selection control of the graphical interface for display, so that the user can select one or more similar vehicle scene images from the similar scene vehicle images in the similar scene vehicle image selection control, and then generate a vehicle product preview image of the target vehicle product that the user needs to display based on the selected similar scene vehicle images.

[0068] Step S12, performing vehicle rendering effect recognition on the vehicle region image in the similar scene vehicle image, and obtaining vehicle rendering information corresponding to the vehicle region image:

[0069] After obtaining the similar scene vehicle image corresponding to the custom information of the display scene customized by the user, the vehicle rendering effect of the vehicle area image in the scene vehicle image will be identified, that is, the rendering effect of the vehicle object in the similar scene vehicle image will be identified, so as to subsequently render the vehicle three-dimensional model of the target vehicle product that the user needs to display based on the rendering effect of the vehicle object in the similar scene vehicle image, so that when the vehicle model image generated after the vehicle three-dimensional model rendering replaces the vehicle area image in the similar scene vehicle image, the vehicle model image can be combined with the image effect in the similar scene vehicle image, thereby reducing the incongruous visual effect caused by replacing the vehicle image, and making the vehicle model image have a natural display effect in the similar scene vehicle image.

[0070] After obtaining similar scene vehicle images, it is first necessary to identify the vehicle area images with vehicle object images in the similar scene vehicle images, that is, the image areas with vehicle image pixels in the similar scene vehicle images. Generally, the vehicle area images in the similar scene vehicle images are identified by a vehicle target recognition model that is trained to convergence. The vehicle target recognition model can be constructed based on other neural network algorithms for target detection such as YOLO or RetinaNet.

[0071] After identifying the vehicle area image in the similar scene vehicle image, the vehicle rendering effect will be identified for the vehicle area image, that is, the rendering effect acting on the vehicle object in the vehicle area image will be identified. Because the vehicle product image displayed in the vehicle product preview image is drawn based on its vehicle stereo model, it is necessary to identify the vehicle display angle of the vehicle object displayed in the vehicle area image in the similar scene vehicle image, so as to adjust the vehicle stereo model of the vehicle product to the vehicle display angle of the vehicle object displayed in the similar vehicle image, so that the display angle of the displayed vehicle product in the finally generated vehicle model image is consistent with the display angle of the vehicle object in the vehicle area image to be replaced in the similar scene vehicle image, so as to reduce the sense of disobedience of the replaced vehicle model image in the similar scene vehicle image. Generally, a vehicle angle recognition model trained to convergence is used to identify vehicle display angle data of vehicle objects displayed in vehicle area images in similar scene vehicle images, wherein the vehicle angle recognition model is a model constructed based on a neural network algorithm. During the training process, a large number of images showing vehicles at different angles are input for training, so that the image feature extraction layer of the vehicle angle recognition model, the display angle recognition layer and the display angle classification layer based on the convolution algorithm, the full connection algorithm and the vector pooling algorithm are trained to convergence, and then the vehicle angle recognition model is trained to convergence, so as to identify the vehicle display angle data of vehicle objects displayed in vehicle area images in similar scene vehicle images, and the identified vehicle display angle data is used as one of the data included in the vehicle rendering information.

[0072] In addition to identifying the vehicle display angle of the vehicle object in the vehicle area image in the vehicle image of the similar scene, the light and shadow rendering effect acting on the displayed vehicle object in the vehicle area image will also be identified to identify the light and shadow rendering data acting on the vehicle object in the vehicle area image, and then use the light and shadow rendering data to perform three-dimensional model rendering on the vehicle three-dimensional model of the target vehicle product. Three-dimensional model rendering refers to adding light and shadow effects to the vehicle three-dimensional model through computer simulation of light propagation and reflection to increase the sense of reality and artistic effect. Therefore, light and shadow rendering data generally includes brightness data, shadow data, light reflection data, light scattering data, color data, depth of field effect data, etc. The light and shadow rendering data of the light and shadow rendering effect of the vehicle three-dimensional model, and each light and shadow data acts on different planes of the vehicle three-dimensional model in the display direction after the vehicle display angle is adjusted. The existing three-dimensional model is generally a three-dimensional structure composed of multiple triangular planes. Therefore, it can also be understood that each light and shadow rendering data acts on different triangular planes of the vehicle three-dimensional model. Therefore, when the light and shadow rendering data acting on the vehicle object in the vehicle area image is identified, the vehicle object in the vehicle area image will also be identified by light and shadow rendering in pixel units to improve the recognition accuracy, so that the finally identified light and shadow rendering data can be applied to different planes of the vehicle three-dimensional model, thereby improving the light and shadow rendering effect of the vehicle three-dimensional model.

[0073] When performing light and shadow rendering recognition on the vehicle area image in the vehicle image of the similar scene, a vehicle light and shadow recognition model trained to convergence will be used for recognition. The vehicle light and shadow recognition model is generally a model constructed based on a neural network algorithm, which has an image feature extraction layer, a light and shadow rendering recognition layer and a light and shadow rendering classification layer. By performing image feature extraction on the vehicle area image, the image feature vector corresponding to the vehicle area image is obtained, and then the image feature vector is input into the light and shadow rendering recognition layer for convolution calculation, full connection calculation and pooling calculation. Finally, the light and shadow rendering data of each image pixel of the vehicle area image is determined through the light and shadow rendering classification layer, and the light and shadow rendering data is used as one of the data included in the vehicle rendering information.

[0074] Step S13, obtaining a vehicle stereo model of the target vehicle product, adjusting the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and performing stereo model rendering on the vehicle stereo model to generate a vehicle model image:

[0075] When the vehicle rendering effect is identified for the vehicle area image in the similar scene vehicle image, the vehicle three-dimensional model corresponding to the target vehicle product for which the user needs to generate a vehicle product preview image will be obtained. The vehicle preview image generation service has a vehicle model library that stores vehicle three-dimensional models corresponding to multiple vehicle products. These vehicle three-dimensional models are stored in multiple categories according to the brand name, vehicle model and vehicle type of their corresponding vehicle products, and different vehicle three-dimensional models have corresponding vehicle identification mappings stored in the vehicle model library. Through the target vehicle identification corresponding to the target vehicle product, the vehicle three-dimensional model stored and mapped with the target vehicle identification is queried from the vehicle model library to generate a vehicle model image showing the vehicle three-dimensional model.

[0076] After obtaining the vehicle three-dimensional model of the target vehicle product and identifying the vehicle rendering information corresponding to the vehicle area image in the similar scene vehicle image, the vehicle display angle of the vehicle three-dimensional model is adjusted according to the vehicle rendering information, and model light and shadow rendering is performed on the vehicle three-dimensional model after the vehicle display angle is adjusted to render the model light and shadow effects of the vehicle three-dimensional model, thereby generating a vehicle model image of the vehicle three-dimensional model having a vehicle display angle that is the same as the vehicle display angle of the vehicle object displayed by the vehicle area image in the similar scene vehicle image, and a model light and shadow effect that is similar to the light and shadow effect of the vehicle object.

[0077] Adjustment of the vehicle display angle refers to the angle of the vehicle three-dimensional model facing the display plane. The display plane takes a flat image display as an example. That is, by adjusting the three-dimensional coordinate axes of the vehicle three-dimensional model, the angle of the vehicle three-dimensional model facing the flat display is adjusted to the vehicle display angle data contained in the vehicle rendering information, so that the angle of the vehicle three-dimensional model facing the display plane is the same as the angle of the vehicle object in the vehicle area image in the vehicle image of the similar scene facing the display plane.

[0078] After adjusting the vehicle display angle of the vehicle three-dimensional model, the model light and shadow effect of the vehicle three-dimensional model will be rendered according to the light and shadow rendering data contained in the vehicle rendering information. According to the light and shadow data in the light and shadow rendering data (such as brightness data, shadow data, light reflection data, light scattering data, color data or depth of field effect data) acting on different planes of the vehicle three-dimensional model, the corresponding light and shadow data are rendered on different planes of the vehicle three-dimensional model to complete the rendering of the model light and shadow effect of the vehicle three-dimensional model, and then generate a vehicle model image of the vehicle three-dimensional model with the same vehicle display angle and similar light and shadow effects as the vehicle object displayed in the vehicle area image in the vehicle image of the similar scene.

[0079] Step S14, replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on the target vehicle product:

[0080] After generating a vehicle model image corresponding to the target vehicle product, the vehicle region image in the similar scene vehicle image is replaced with the vehicle model image, so that the vehicle object displayed in the similar scene vehicle image is the target vehicle product.

[0081] Since the vehicle area image in the similar scene vehicle image has been identified before, the vehicle area image in the similar scene image can be directly cropped and removed, and the image area in the similar scene vehicle image where the vehicle area image is cropped and removed is used as the vehicle synthesis area for synthesizing the vehicle model image, and the image size of the vehicle model image is adjusted to match the image size of the vehicle synthesis area, and then the vehicle model image with the image size adjustment is synthesized into the vehicle synthesis area of ​​the similar scene vehicle image, to generate a vehicle product preview image in which the vehicle object displayed in the similar scene vehicle image is the target vehicle product.

[0082] In order to improve the appearance of the target vehicle product in the vehicle product preview image, after the vehicle area image in the similar scene vehicle image is replaced with the vehicle model image, a scene filter image that conforms to the scene environment displayed by the similar scene vehicle image can be synthesized in the area of ​​the vehicle model image in the similar scene vehicle image. The vehicle preview image generation service has scene filter images corresponding to each scene vehicle image. For example, a scene filter image displaying a sand and dust effect can correspond to various scene vehicle images displaying desert scenes, and a scene filter image displaying a rain effect can correspond to various scene vehicle images displaying rainy scenes, etc. When the vehicle area image in a similar scene vehicle image is replaced with the vehicle model image, the scene filter image corresponding to the similar scene vehicle image is obtained, and the scene filter image is synthesized to cover the vehicle model image in the similar scene vehicle image, or the scene filter image is synthesized into the similar scene vehicle image in a full coverage manner.

[0083] After generating a vehicle product preview image corresponding to the target vehicle product, the vehicle product preview image is output to the graphical user interface of the user's terminal device for display. If the user is an anchor user, the vehicle product preview image will be broadcast to the live broadcast room through the server of the network live broadcast platform, so that the graphical user interface of the terminal device of the anchor user and the audience user can display the vehicle product preview image.

[0084] In addition, the vehicle product preview image can be output to the preview image control in the graphical user interface for display. The preview image control can be used to jump to the vehicle product details introduction page corresponding to the vehicle product, so that the user can further understand the vehicle product displayed in the vehicle product preview image. The user can touch and click the preview image control to drive the terminal device to obtain the vehicle three-dimensional model corresponding to the vehicle product preview image and output it to the preview image control for display, so that the user can interact with the vehicle three-dimensional model of the previewed vehicle product in the preview image control, preview the appearance and interior of the vehicle product from various angles, so that the user can better understand the vehicle product.

[0085] The above typical embodiments and their variations fully disclose typical implementation schemes of the vehicle preview image generation method of the present application. However, various variations of the method can still be derived by changing and amplifying some technical means. Other embodiments are briefly described as follows:

[0086] According to another embodiment of the present application, please refer to Figure 3 , obtaining the display scene custom information, and obtaining the similar scene vehicle image corresponding to the display scene custom information, including the following steps:

[0087] Step S111, acting on the display scene customization event of the target vehicle product, and identifying the display scene customization information from the user's voice:

[0088] After the user uses the vehicle preview image generation service to select the target vehicle product for which a vehicle product preview image needs to be generated, the user can customize and edit the display scene customization information of the scene environment that displays the target vehicle product in the vehicle product preview image. The vehicle preview image generation service can provide users with the service of customizing the scene through voice, and display a control for recording the scene customization information in the graphical user interface of the terminal device where the user is located. The user uses the control to activate the audio input device of the terminal device (such as the microphone device of the mobile phone terminal itself) to record the user's voice, and then identify the corresponding custom text for customizing the scene from the recorded user voice as the display scene customization information.

[0089] Step S112, in response to an edit completion event of a display scene custom text edit control in a graphical user interface, obtain custom text corresponding to the display scene custom text edit control, and use the custom text as display scene custom information acting on a target vehicle product:

[0090] In addition to the above-mentioned display custom information generated by voice, the vehicle preview image generation service also provides users with traditional scene custom editing services through text editing. The display scene custom text editing control is displayed in the graphical user interface of the user's terminal device, and the user enters the corresponding text through the display scene custom text editing control. After the editing is completed, the terminal device responds to the editing completion event of the display scene custom text editing control, and obtains the custom text entered by the user in the display scene custom text editing control, so as to use the custom text as the display scene custom information acting on the target vehicle product.

[0091] In the embodiments disclosed above, the vehicle preview image generation service provides users with the ability to edit and display scene customization information via voice and text, allowing users to freely choose a corresponding method to customize and edit the scene environment of the vehicle product in the vehicle product preview image.

[0092] According to another embodiment of the present application, please refer to Figure 4 , obtaining the display scene custom information, and obtaining the similar scene vehicle image corresponding to the display scene custom information, including the following steps:

[0093] Step S111', preprocessing the custom text of the display scene custom information, and extracting the semantic feature vector of the custom text after the text preprocessing:

[0094] Obtain the custom text contained in the custom information of the display scene, and perform text preprocessing on the custom text, for example, perform word segmentation, sentence segmentation, word normalization or stop word removal on the custom text. After completing the text preprocessing of the custom text, the semantic feature vector of the custom text will be extracted through the semantic feature extraction layer.

[0095] Step S112', determining the semantic feature similarity between the semantic feature vector and the semantic feature vectors of each scene vehicle image in the scene vehicle image library:

[0096] After the semantic feature vector of the custom text is extracted, the semantic feature similarity between the semantic feature vector of the custom text and the semantic feature vector of each scene vehicle image in the vehicle scene library is calculated. After the semantic feature vector of the custom text is calculated, semantic feature extraction is performed on each scene vehicle image to obtain the corresponding semantic feature vector. The semantic feature vector may also be pre-extracted for each scene vehicle image. If it is a pre-extracted semantic feature vector, each semantic feature vector will be mapped with the scene vehicle image to which it belongs and stored in the scene vehicle image library.

[0097] By mapping the semantic feature vector of the custom text and the semantic feature vector of the scene vehicle image currently undergoing semantic feature similarity calculation to the same semantic space, the distance between the two semantic feature vectors is calculated as the semantic feature similarity between the custom text and the scene vehicle image. The semantic feature similarity calculation can be implemented using any of the large-scale vector retrieval engines such as Faiss, ElasticSearch, Milvus, etc., or can be calculated using any of the existing algorithms such as cosine similarity, inner product, Manhattan distance, Euclidean distance, etc.

[0098] Step S113', taking one or more scene vehicle images whose semantic feature similarity exceeds a preset threshold as similar scene vehicle images:

[0099] When calculating the semantic feature similarity between the semantic feature vector of the custom text and the semantic feature vector of each scene vehicle image, it will be judged that the semantic feature similarity of both parties exceeds the preset threshold. The scene vehicle image whose semantic feature similarity exceeds the preset threshold will be determined as the display scene vehicle image.

[0100] In the embodiment disclosed above, by extracting the semantic feature vector of the custom text edited by the user for displaying the scene environment of the vehicle product in the customized vehicle product picture, and calculating the semantic feature similarity between the semantic feature vector of the custom text and the semantic feature vector of each scene vehicle picture, the scene vehicle image whose semantic feature similarity with the custom text exceeds a threshold is used as a similar scene vehicle image, and a neural network is used to filter out similar scene vehicle images that conform to the scene environment customized by the user, so that the filtered similar scene vehicle images conform to the scene environment customized by the user, and the scene environment displaying the vehicle product in the finally generated vehicle preview picture can meet the user's needs for using the vehicle preview picture generation service.

[0101] According to another embodiment of the present application, please refer to Figure 5 , the step of performing vehicle rendering effect recognition on the vehicle area image in the vehicle image of the similar scene and obtaining vehicle rendering information corresponding to the vehicle area image comprises the following steps:

[0102] Step S121, identifying the vehicle area image in the similar vehicle scene image:

[0103] After obtaining similar scene vehicle images, a vehicle target recognition model trained to convergence is used to identify vehicle area images in similar scene vehicle images. The vehicle target recognition model is generally a neural network target detection model constructed based on other neural network algorithms for target detection such as YOLO or RetinaNet, wherein the target of the vehicle target recognition model is the vehicle object displayed in the image, and the image feature vectors of similar scene vehicle images are extracted to identify the vehicle objects displayed in the similar scene vehicle images, and the area image with the vehicle object is used as the vehicle area image in the display scene vehicle image.

[0104] Step S122, calling the vehicle angle recognition model trained to convergence, and identifying the vehicle display angle data of the vehicle object in the vehicle area image based on the image feature vector corresponding to the vehicle area image:

[0105] After identifying the vehicle area image in the similar scene vehicle image, it is necessary to identify the vehicle display angle of the vehicle object displayed in the vehicle area image in the similar scene vehicle image. Generally, a vehicle angle recognition model trained to convergence is used to identify the vehicle display angle data of the vehicle object displayed in the vehicle area image in the similar scene vehicle image, wherein the vehicle angle recognition model is a model constructed based on a neural network algorithm. During the training process, a large number of images showing vehicles at different angles are input for training to train the image feature extraction layer of the vehicle angle recognition model, the display angle recognition layer and the display angle classification layer based on the convolution algorithm, the fully connected algorithm and the vector pooling algorithm, and then the vehicle angle recognition model is trained to convergence to identify the vehicle display angle data of the vehicle object displayed in the vehicle area image in the similar scene vehicle image, and the identified vehicle display angle data is used as one of the data included in the vehicle rendering information.

[0106] Step S123, calling the vehicle light and shadow recognition model trained to convergence, and identifying the light and shadow rendering data acting on the vehicle object in the vehicle area image based on the image feature vector corresponding to the vehicle area image:

[0107] When the vehicle display angle of the vehicle object in the vehicle area image in the similar scene vehicle image is identified, the light and shadow rendering effect of the vehicle object on display in the vehicle area image will also be identified to identify the light and shadow rendering data acting on the vehicle object in the vehicle area image. When the light and shadow rendering of the vehicle area image in the similar scene vehicle image is identified, a vehicle light and shadow recognition model trained to convergence will be used for identification. The vehicle light and shadow recognition model is generally a model constructed based on a neural network algorithm, which has an image feature extraction layer, a light and shadow rendering recognition layer and a light and shadow rendering classification layer. By extracting image features from the vehicle area image, an image feature vector corresponding to the vehicle area image is obtained, and then the image feature vector is input into the light and shadow rendering recognition layer for convolution calculation, full connection calculation and pooling calculation. Finally, the light and shadow rendering data of each image pixel of the vehicle area image is determined through the light and shadow rendering classification layer, and the light and shadow rendering data is used as one of the data included in the vehicle rendering information.

[0108] Step S124, generating vehicle rendering information including the vehicle display angle data and the light and shadow rendering data:

[0109] After identifying the vehicle display angle data and light and shadow rendering data of the vehicle object in the vehicle area image in the similar scene vehicle image, the vehicle display angle data and the light and shadow rendering data are encapsulated to generate vehicle rendering information including the vehicle display angle data and the light and shadow rendering data.

[0110] In the embodiments disclosed above, vehicle display angle recognition and light and shadow rendering recognition are performed on the vehicle object identified in the similar scene vehicle image, and the vehicle stereo model of the vehicle product is adjusted by an equal amount using the identified display angle data, so that the display angle of the replaced vehicle product in the similar scene vehicle image is the same as that of the original vehicle object, thereby reducing the sense of incongruity of the replaced vehicle model image in the similar scene vehicle image, and the light and shadow rendering data is used to perform stereo model rendering on the vehicle stereo model of the target vehicle product, so that the replaced vehicle product in the similar scene vehicle image can be naturally integrated into the scene environment.

[0111] According to another embodiment of the present application, please refer to Figure 6 , obtaining a vehicle stereo model of a target vehicle product, adjusting the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and performing stereo model rendering on the vehicle stereo model, and generating a vehicle model image, including the following steps:

[0112] Step S131, according to the vehicle identification of the target vehicle product, obtaining the target vehicle stereo model corresponding to the vehicle identification from the vehicle stereo model library:

[0113] When identifying vehicle rendering effects in vehicle images of similar scenes, a vehicle three-dimensional model of the target vehicle product for which the user needs to generate a vehicle product preview image will also be obtained. Generally, a query will be made from a vehicle three-dimensional model library that maps and stores vehicle identifications and vehicle three-dimensional models. Specifically, by obtaining the vehicle identification of the target vehicle product, the target vehicle three-dimensional model mapped and stored with the vehicle identification is obtained from the vehicle three-dimensional model library.

[0114] Step S132, adjusting the vehicle display angle of the target vehicle stereoscopic model by an equal amount according to the vehicle display angle data included in the vehicle rendering information:

[0115] After obtaining the vehicle display angle data determined by vehicle display angle recognition of the vehicle area image in the similar scene vehicle image, the vehicle display angle of the target vehicle three-dimensional model facing the display plane is adjusted by an equal amount, that is, the vehicle display angle of the target vehicle three-dimensional model facing the display plane (such as the display plane) is adjusted to be the same as the vehicle display angle of the vehicle object displayed in the similar scene vehicle image.

[0116] Step S133, rendering the model light and shadow effect of the target vehicle stereoscopic model after display angle adjustment according to the light and shadow rendering data included in the vehicle rendering information, and generating a vehicle model image having the target vehicle stereoscopic model:

[0117] After completing the equal adjustment of the vehicle display angle of the vehicle three-dimensional model, the model light and shadow effect of the vehicle three-dimensional model will be rendered according to the light and shadow rendering data obtained by performing light and shadow rendering recognition on the vehicle area image in the vehicle image of the similar scene, and the model light and shadow effect of the vehicle three-dimensional model will be rendered according to the different planes of the vehicle three-dimensional model acted on by various light and shadow data in the light and shadow rendering data (such as brightness data, shadow data, light reflection data, light scattering data, color data or depth of field effect data), so as to render the model light and shadow effect of the vehicle three-dimensional model, and then generate a vehicle model image of the vehicle three-dimensional model with the same vehicle display angle and similar light and shadow effect as the vehicle object displayed in the vehicle area image in the vehicle image of the similar scene.

[0118] After the model light and shadow effect rendering of the target vehicle three-dimensional model is completed, a vehicle model image of the target vehicle three-dimensional model after the equal vehicle display angle adjustment and the model light and shadow effect rendering can be generated.

[0119] In the embodiments disclosed above, based on the vehicle display angle data and light and shadow rendering data respectively obtained by performing vehicle display angle recognition and light and shadow rendering recognition on the vehicle objects in the similar scene vehicle images, the vehicle display angle of the vehicle stereo model of the vehicle product is adjusted and the light and shadow model effect rendering of the vehicle stereo model is performed, so that the display angle and light and shadow effect of the generated vehicle model image are the same as the vehicle object displayed in the similar scene vehicle image, so that after the vehicle model image representing the vehicle product replaces the similar vehicle object in the similar scene vehicle image, the vehicle model image can be naturally integrated with the scene environment in the similar scene vehicle image, thereby reducing the image incongruity caused by the replacement of the displayed vehicle.

[0120] According to another embodiment of the present application, please refer to Figure 7 , the step of replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on the target vehicle product includes the following steps:

[0121] Step S141, cropping and removing the vehicle region image in the similar scene vehicle image, and using the cropped and removed region in the similar scene vehicle image as the vehicle synthesis region:

[0122] After generating a vehicle model image of the target vehicle product specified by the user, the vehicle area image in the similar scene vehicle image will be cropped and removed, that is, the vehicle object originally displayed in the similar scene vehicle image will be removed, so that the cropped and removed area in the cropped vehicle image will be displayed as a vehicle synthesis area, which is used to synthesize the vehicle model image at the vehicle synthesis area.

[0123] Step S142, generating a vehicle model image corresponding to the vehicle stereo model for which the stereo model rendering has been completed, and adjusting the image size of the vehicle model image according to the area size of the vehicle synthesis area:

[0124] The image size of the vehicle model image is adjusted to the area size of the vehicle synthesis area in the similar scene vehicle image, so that the vehicle model image can be completely synthesized into the vehicle synthesis area without blocking other image areas in the similar scene vehicle image.

[0125] Step S143, synthesizing the vehicle model image that has completed image resizing into the vehicle synthesis area in the similar scene vehicle image to generate a vehicle product preview image:

[0126] After adjusting the image size of the vehicle model image to the area size of the vehicle synthesis area in the similar scene vehicle image, the vehicle model image can be synthesized into the vehicle synthesis area in the similar scene vehicle image. The similar scene vehicle image after completing the vehicle model image synthesis is the vehicle product preview image acting on the target vehicle product.

[0127] In the embodiments disclosed above, the vehicle area image in the similar scene vehicle image is cropped and removed, and a vehicle model image of the vehicle product is synthesized in the cropped and removed area. The vehicle model image is resized before synthesis so that the vehicle model image can be merged with the synthesized area. The similar scene vehicle image after synthesizing the vehicle model image is the vehicle product preview image of the vehicle product. The vehicle product preview image shows that the vehicle product is in a scene environment similar to the user's customized editing to meet the user's vehicle product preview needs.

[0128] According to another embodiment of the present application, please refer to Figure 8 , after the step of replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on the target vehicle product, the following steps are included:

[0129] Step S15, outputting the vehicle product preview image to the preview image control of the graphical user interface for display:

[0130] After generating a vehicle product preview image corresponding to the vehicle product specified by the user, the vehicle product preview image is output to the graphical user interface of the terminal device where the user is located. Generally, the vehicle product preview image is output to the preview image control in the graphical user interface for display. The preview image control has a control moving function and a control hiding function, so that the user can move and hide the displayed vehicle product preview image in the graphical user interface.

[0131] Step S16, in response to the three-dimensional model display instruction acting on the preview image control, obtaining the vehicle three-dimensional model corresponding to the vehicle product preview image:

[0132] The preview image control generally has an interactive function of displaying a vehicle three-dimensional model of the vehicle product corresponding to the vehicle product preview image. The user can touch and click the three-dimensional model display control in the preview image control to trigger the terminal device to respond to the three-dimensional model display instruction, and then obtain the vehicle three-dimensional model corresponding to the vehicle product preview image displayed in the preview image control for output display.

[0133] Step S17, displaying a three-dimensional model interaction window in a graphical user interface, and outputting the vehicle three-dimensional model to the three-dimensional model interaction window:

[0134] After obtaining the vehicle stereoscopic model corresponding to the vehicle preview image, a stereoscopic model interaction window will be displayed in the graphical user interface, and the vehicle stereoscopic model will be output to the stereoscopic model interaction window for display. The stereoscopic model interaction window generally has a model interaction function, allowing the user to adjust the angle of the vehicle stereoscopic model and interact with the movable vehicle level in the vehicle stereoscopic model, for example, turning on the lights of the vehicle stereoscopic model, adjusting the door opening angle of the vehicle stereoscopic model, and turning the vehicle steering wheel of the vehicle stereoscopic model.

[0135] In the embodiments disclosed above, a vehicle product preview image of a vehicle product will be generated and displayed in a graphical user interface, and the user will be provided with the function of interacting with a vehicle stereo model of the vehicle product. In addition to being able to preview the appearance display effect of the vehicle product in a customized similar scene environment through the vehicle product preview image, the user can also further browse the vehicle product by interacting with the vehicle stereo model of the vehicle product, thereby enhancing the user's online car viewing experience.

[0136] See also Fig. 9 , a vehicle preview image generating device provided to meet one of the purposes of the present application: a similar picture acquisition module 11, used to acquire display scene custom information, and acquire a similar scene vehicle image corresponding to the display scene custom information; a rendering information acquisition module 12, used to identify the vehicle rendering effect of the vehicle area image in the similar scene vehicle image, and acquire the vehicle rendering information corresponding to the vehicle area image; a model image generation module 13, used to acquire a vehicle stereo model of a target vehicle product, adjust the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and perform stereo model rendering on the vehicle stereo model to generate a vehicle model image; a vehicle preview image generation module 14, used to replace the vehicle area image in the similar scene vehicle image with the vehicle model image, so as to generate a vehicle product preview image acting on the target vehicle product.

[0137] In one embodiment, the similar image acquisition module 11 includes: acting on a display scene customization event of a target vehicle product, identifying display scene customization information from a user voice; responding to an editing completion event of a display scene customization text editing control in a graphical user interface, acquiring a custom text corresponding to the display scene customization text editing control, and using the custom text as the display scene customization information acting on the target vehicle product.

[0138] In another embodiment, the similar picture acquisition module 11 also includes: a semantic feature extraction submodule, which is used to perform text preprocessing on the custom text that displays the scene custom information, and extract the semantic feature vector of the custom text that has completed the text preprocessing; a feature similarity calculation submodule, which is used to determine the semantic feature similarity between the semantic feature vector and the semantic feature vectors of each scene vehicle image in the scene vehicle image library; and a similar image determination submodule, which is used to take one or more scene vehicle images whose semantic feature similarity exceeds a preset threshold as similar scene vehicle images.

[0139] In one embodiment, the rendering information acquisition module 12 includes: a vehicle area recognition submodule, used to recognize the vehicle area image in the similar vehicle scene picture; a display angle recognition submodule, used to call a vehicle angle recognition model trained to convergence, and based on the image feature vector corresponding to the vehicle area image, identify the vehicle display angle data of the vehicle object in the vehicle area image; a light and shadow rendering recognition submodule, used to call a vehicle light and shadow recognition model trained to convergence, and based on the image feature vector corresponding to the vehicle area image, identify the light and shadow rendering data acting on the vehicle object in the vehicle area image; a rendering information generation submodule, used to generate vehicle rendering information including the vehicle display angle data and the light and shadow rendering data.

[0140] In one embodiment, the model image generation module 13 includes: a target vehicle model acquisition submodule, which is used to obtain a target vehicle three-dimensional model corresponding to the vehicle identification from a vehicle three-dimensional model library according to the vehicle identification of the target vehicle product; a vehicle model angle adjustment submodule, which is used to adjust the vehicle display angle of the target vehicle three-dimensional model by an equal amount according to the vehicle display angle data included in the vehicle rendering information; and a vehicle model light and shadow rendering submodule, which is used to render the model light and shadow effect of the target vehicle three-dimensional model whose display angle has been adjusted according to the light and shadow rendering data included in the vehicle rendering information, and generate a vehicle model image having the target vehicle three-dimensional model.

[0141] In one embodiment, the vehicle preview image generation module 14 includes: a regional image cropping and removal submodule, which is used to crop and remove the vehicle regional image in the similar scene vehicle image, and use the cropped and removed area in the similar scene vehicle image as the vehicle synthesis area; a vehicle image size adjustment submodule, which is used to generate a vehicle model image corresponding to the vehicle stereo model of the vehicle that has completed stereo model rendering, and adjust the image size of the vehicle model image according to the area size of the vehicle synthesis area; a vehicle image synthesis submodule, which is used to synthesize the vehicle model image that has completed image size adjustment into the vehicle synthesis area in the similar scene vehicle image to generate a vehicle product preview image.

[0142] Another embodiment of the present application also provides a vehicle preview image generating device. Fig.10 As shown, the vehicle preview image generating device can be implemented by a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The database may store a control information sequence. When the computer-readable instructions are executed by the processor, the processor can implement a vehicle preview image generating method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the vehicle preview image generating method of the present application. The network interface of the computer device is used to connect and communicate with a terminal. Those skilled in the art can understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0143] In this embodiment, the processor is used to execute Fig.10 The memory stores the program code and various data required to execute the above modules or submodules. The network interface is used to transmit data between user terminals or servers. The memory in this embodiment stores the program code and data required to execute all modules in the vehicle preview image generation device of this application, and the server can call the program code and data of the server to execute the functions of all modules.

[0144] The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the vehicle preview image generation method of any embodiment of the present application.

[0145] The present application also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any embodiment of the present application when executed by one or more processors.

[0146] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments of the present application can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0147] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

[0148] In summary, the present application provides users with a vehicle preview image generation service, and users can customize the scene environment for displaying vehicle products in the vehicle product preview image, so that the vehicle products in the vehicle product preview image can be displayed in the scene environment customized by the user.

Claims

1. A method for generating a vehicle preview image, characterized in that: The steps include: Obtaining display scene custom information, and obtaining similar scene vehicle images corresponding to the display scene custom information, including: performing text preprocessing on the custom text of the display scene custom information, and extracting the semantic feature vector of the custom text after the text preprocessing; determining the semantic feature similarity between the semantic feature vector and the semantic feature vectors of each scene vehicle image in the scene vehicle image library; and using one or more scene vehicle images whose semantic feature similarity exceeds a preset threshold as similar scene vehicle images; Performing vehicle rendering effect recognition on the vehicle area image in the vehicle image of the similar scene, and obtaining vehicle rendering information corresponding to the vehicle area image, including: recognizing the vehicle area image in the vehicle image of the similar scene; calling a vehicle angle recognition model trained to convergence, and based on the image feature vector corresponding to the vehicle area image, identifying the vehicle display angle data of the vehicle object in the vehicle area image; calling a vehicle light and shadow recognition model trained to convergence, and based on the image feature vector corresponding to the vehicle area image, identifying the light and shadow rendering data acting on the vehicle object in the vehicle area image; generating vehicle rendering information including the vehicle display angle data and the light and shadow rendering data; Acquire a vehicle stereo model of a target vehicle product, adjust the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and perform stereo model rendering on the vehicle stereo model to generate a vehicle model image, including: acquiring a target vehicle stereo model corresponding to the vehicle identification from a vehicle stereo model library according to the vehicle identification of the target vehicle product; adjusting the vehicle display angle of the target vehicle stereo model by an equal amount according to the vehicle display angle data included in the vehicle rendering information; rendering the model light and shadow effects of the target vehicle stereo model whose display angle has been adjusted according to the light and shadow rendering data included in the vehicle rendering information, and generating a vehicle model image having the target vehicle stereo model; The vehicle area image in the similar scene vehicle image is replaced with the vehicle model image to generate a vehicle product preview image acting on a target vehicle product.

2. The method according to claim 1, characterized in that The steps of obtaining the display scene custom information and obtaining the similar scene vehicle image corresponding to the display scene custom information include the following steps: In response to a display scene customization event acting on a target vehicle product, the display scene customization information is recognized from a user's voice; In response to an editing completion event of a display scene custom text editing control in a graphical user interface, a custom text corresponding to the display scene custom text editing control is obtained, and the custom text is used as display scene custom information acting on a target vehicle product.

3. The method according to claim 1, characterized in that The step of replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on a target vehicle product comprises the following steps: Crop and remove the vehicle region image in the similar scene vehicle image, and use the cropped and removed region in the similar scene vehicle image as the vehicle synthesis region; Generating a vehicle model image corresponding to the vehicle stereo model for which the stereo model rendering has been completed, and adjusting the image size of the vehicle model image according to the area size of the vehicle synthesis area; The vehicle model image that has completed image resizing is synthesized into the vehicle synthesis area in the similar scene vehicle image to generate a vehicle product preview image.

4. The method according to claim 1, characterized in that: After the step of replacing the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on the target vehicle product, the following steps are included: Outputting the vehicle product preview image to a preview image control of a graphical user interface for display; In response to a three-dimensional model display instruction acting on the preview image control, obtaining a vehicle three-dimensional model corresponding to the vehicle product preview image; A three-dimensional model interaction window is displayed in a graphical user interface, and the vehicle three-dimensional model is output to the three-dimensional model interaction window.

5. A vehicle preview image generating device, characterized in that: include: A similar image acquisition module is used to acquire the display scene custom information and acquire the similar scene vehicle image corresponding to the display scene custom information, including: performing text preprocessing on the custom text of the display scene custom information and extracting the semantic feature vector of the custom text after the text preprocessing; determining the semantic feature similarity between the semantic feature vector and the semantic feature vectors of each scene vehicle image in the scene vehicle image library; and taking one or more scene vehicle images whose semantic feature similarity exceeds a preset threshold as similar scene vehicle images; A rendering information acquisition module is used to identify the vehicle rendering effect of the vehicle area image in the vehicle image of the similar scene, and obtain the vehicle rendering information corresponding to the vehicle area image, including: identifying the vehicle area image in the vehicle image of the similar scene; calling the vehicle angle recognition model trained to convergence, and based on the image feature vector corresponding to the vehicle area image, identifying the vehicle display angle data of the vehicle object in the vehicle area image; calling the vehicle light and shadow recognition model trained to convergence, and based on the image feature vector corresponding to the vehicle area image, identifying the light and shadow rendering data acting on the vehicle object in the vehicle area image; generating vehicle rendering information including the vehicle display angle data and the light and shadow rendering data; A model image generation module is used to obtain a vehicle stereo model of a target vehicle product, adjust the display angle of the vehicle stereo model to the vehicle display angle represented by the vehicle area image according to the vehicle rendering information, and perform stereo model rendering on the vehicle stereo model to generate a vehicle model image, including: according to the vehicle identification of the target vehicle product, obtaining a target vehicle stereo model corresponding to the vehicle identification from a vehicle stereo model library; according to the vehicle display angle data included in the vehicle rendering information, adjusting the vehicle display angle of the target vehicle stereo model by an equal amount; according to the light and shadow rendering data included in the vehicle rendering information, rendering the model light and shadow effect of the target vehicle stereo model whose display angle has been adjusted, and generating a vehicle model image having the target vehicle stereo model; The vehicle preview image generating module is used to replace the vehicle area image in the similar scene vehicle image with the vehicle model image to generate a vehicle product preview image acting on a target vehicle product.

6. A vehicle preview image generating device, comprising a central processing unit and a memory, characterized in that: The central processing unit is used to call and run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: It stores a computer program implemented according to the method described in any one of claims 1 to 4 in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the corresponding method are executed.