Video restoration method and apparatus, and device, storage medium and program product

Abstract

The present disclosure is applied to the technical field of audio and video. Provided are a video restoration method and apparatus, and a device, a storage medium and a program product. The method comprises: acquiring a plurality of original images captured for a target position from a plurality of preset orientations; identifying and analyzing the plurality of original images, and determining a main object in the plurality of original images; constructing a target model on the basis of image information of the plurality of original images, the main object and a plurality of original models in a historical database; and using a virtual camera to perform frame supplementation on the target model from the plurality of preset orientations, respectively, so as to restore a first surround video of the target position from the plurality of preset orientations.

Description

Video restoration method, device, equipment, storage medium and program product

[0001] Cross-reference to related applications

[0002] The present disclosure claims priority to Chinese Patent Application No. 202410480338.5, filed on April 19, 2024 in China, the entire contents of which are incorporated herein by reference. TECHNICAL FIELD

[0003] The present disclosure relates to the field of audio and video technology, and particularly relates to a video restoration method, device, equipment, storage medium and program product. BACKGROUND

[0004] With the development of photography technology, volume photography technology is continuously applied in the fields of music video (MV) shooting, sports event live broadcast, three-dimensional (3D) medical simulation, etc. Unlike panoramic photography, this technology uses a large number of cameras to capture target tasks and scenes in a studio from multiple angles, and can generate and output dynamic 3D images and models. However, the related art video restoration of volume shooting mostly adopts a tolerance splicing method. The synthesized video data is processed by a push stream server and video encoding and decoding, transmitted to a broadcast end for full decoding, and then presented to a user. This method has a complex physical deployment link in the front-end acquisition, large data transmission volume, and high cost of physical acquisition equipment, so that the video restoration process is difficult. SUMMARY

[0005] Embodiments of the present disclosure provide a video restoration method, device, equipment, storage medium and program product to solve the problem of difficult video restoration process in the related art video restoration method.

[0006] To solve the above technical problems, the present disclosure is implemented as follows:

[0007] In a first aspect, the embodiments of the present disclosure provide a video restoration method, which comprises:

[0008] Obtaining a plurality of original images photographed from a plurality of preset directions to a target position, wherein each preset direction photographs at least one original image;

[0009] Identifying and analyzing the plurality of original images to determine a subject object in the plurality of original images;

[0010] Constructing a target model based on image information of the plurality of original images, the subject object and a plurality of original models in a historical database;

[0011] The virtual camera is used to add pictures to the target model from the plurality of preset orientations respectively, so as to restore the first surround video of the target position at the plurality of preset orientations.

[0012] Optionally, the identifying and analyzing the plurality of original images to determine the subject object in the plurality of original images comprises:

[0013] extracting edge information, color histogram and gradient histogram of each original image in the plurality of original images;

[0014] obtaining a feature vector of each original image according to the edge information, color histogram and gradient histogram of the original image;

[0015] obtaining edge contour information of each original image according to the feature vector of the original image;

[0016] calculating a maximum closed contour area of each original image according to the edge contour information of the original image;

[0017] performing feature matching on objects corresponding to the maximum closed contour areas of the original images, and determining the objects corresponding to the maximum closed contour areas of the N original images as the subject object in a case where the objects corresponding to the maximum closed contour areas of the N original images are the same, N being an integer greater than 1.

[0018] Optionally, the plurality of original models comprises a scene model and a plurality of subject models, and the constructing a target model based on the image information of the plurality of original images, the subject object and the plurality of original models in the historical database comprises:

[0019] determining a target subject model matched with the subject object from the plurality of subject models;

[0020] performing vector alignment on the target subject model and the scene model based on the image information of the plurality of original images, superimposing the target subject model into the scene model to obtain a target model.

[0021] Optionally, the determining a target subject model matched with the subject object from the plurality of subject models comprises:

[0022] calculating a plurality of first Euclidean distances between a feature vector of the subject object and feature vectors of the plurality of subject models respectively, wherein each first Euclidean distance is a Euclidean distance between the feature vector of the subject object and a feature vector of each subject model;

[0023] determine a target subject model matching the subject object from the plurality of subject models based on the plurality of first Euclidean distances, wherein a first Euclidean distance between the feature vector of the target subject model and the feature vector of the subject object is the minimum Euclidean distance in the plurality of first Euclidean distances.

[0024] Optionally, in a case where the subject object is an object, the calculating the plurality of first Euclidean distances between the feature vector of the subject object and the feature vectors of the plurality of subject models respectively comprises:

[0025] calculating a plurality of second Euclidean distances based on a plurality of first parameter pairs respectively, wherein the first parameter pair comprises a first parameter and a second parameter corresponding to the first parameter, the first parameter is a feature vector of the object at a preset orientation, and the second parameter is a feature vector of a first subject model at a corresponding preset orientation;

[0026] calculating a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between the feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0027] Optionally, in a case where the subject object is a person, the calculating the plurality of first Euclidean distances between the feature vector of the subject object and the feature vectors of the plurality of subject models respectively comprises:

[0028] calculating a plurality of third Euclidean distances based on a plurality of second parameter pairs respectively, wherein each second parameter pair comprises a third parameter and a fourth parameter corresponding to the third parameter, the third parameter being used to represent a skeletal joint of the person, and the fourth parameter being used to represent a corresponding skeletal joint of a first subject model;

[0029] calculating a sum of the plurality of third Euclidean distances to obtain a first Euclidean distance between the feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0030] Optionally, the picture supplementing the target model from the plurality of preset orientations by using the virtual camera respectively to restore a first surround video of the target position at the plurality of preset orientations comprises:

[0031] obtaining actual shooting angles of the target position at the plurality of preset orientations;

[0032] based on the actual shooting angles of the target position at the plurality of preset orientations, picture supplementing the target model from the plurality of preset orientations by using the virtual camera respectively to obtain an initial surround video of the target position at the plurality of preset orientations.

[0033] extracting key information of the plurality of original images;

[0034] rendering the initial surround video of the target position at the plurality of preset orientations by using the key information to obtain a first surround video of the target position at the plurality of preset orientations.

[0035] Optionally, after the first surround video of the target position at the plurality of preset orientations is restored by using the virtual camera to respectively perform picture supplementation on the target model from the plurality of preset orientations, the method further comprises:

[0036] aligning the playing visual angle of the player with the actual shooting visual angle, and then sending the first surround video of the target position at the plurality of preset orientations to the player.

[0037] Optionally, after the first surround video of the target position at the plurality of preset orientations is restored by using the virtual camera to respectively perform picture supplementation on the target model from the plurality of preset orientations, the method further comprises:

[0038] receiving the correction information sent by the player;

[0039] generating a second surround video based on the correction information and the first surround video of the target position at the plurality of preset orientations;

[0040] aligning the playing visual angle of the player with the actual shooting visual angle, and then sending the second surround video of the target position at the plurality of preset orientations to the player.

[0041] In a second aspect, the embodiments of the present disclosure further provide a video restoration device, which comprises:

[0042] a first acquisition module configured to acquire a plurality of original images obtained by shooting a target position from a plurality of preset orientations, wherein at least one original image is shot from each preset orientation;

[0043] a first determination module configured to perform identification analysis on the plurality of original images to determine a subject object in the plurality of original images;

[0044] a first construction module configured to construct a target model based on image information of the plurality of original images, the subject object and a plurality of original models in a historical database;

[0045] a first restoration module configured to use a virtual camera to respectively perform picture supplementation on the target model from the plurality of preset orientations to restore a first surround video of the target position at the plurality of preset orientations.

[0046] Optionally, the first determining module comprises:

[0047] a first extraction unit, configured to extract edge information, a color histogram and a gradient histogram of each of the plurality of original images;

[0048] a first processing unit, configured to obtain a feature vector of each of the plurality of original images according to the edge information, the color histogram and the gradient histogram of each of the plurality of original images;

[0049] a second processing unit, configured to obtain edge contour information of each of the plurality of original images according to the feature vector of each of the plurality of original images;

[0050] a first calculation unit, configured to calculate a maximum closed contour area of each of the plurality of original images according to the edge contour information of each of the plurality of original images;

[0051] a first determining unit, configured to perform feature matching on objects corresponding to the maximum closed contour areas of the plurality of original images, and determine that objects corresponding to maximum closed contour areas of N original images are main objects in a case where the objects corresponding to the maximum closed contour areas of the N original images are the same, wherein N is an integer greater than 1.

[0052] Optionally, the plurality of original models comprise a scene model and a plurality of main models, and the first constructing module comprises:

[0053] a second determining unit, configured to determine a target main model matching the main object from the plurality of main models;

[0054] a third processing unit, configured to perform vector alignment on the target main model and the scene model based on image information of the plurality of original images, and superimpose the target main model into the scene model to obtain a target model.

[0055] Optionally, the second determining unit comprises:

[0056] a first calculation sub-unit, configured to calculate a plurality of first Euclidean distances between a feature vector of the main object and feature vectors of the plurality of main models respectively, wherein each of the first Euclidean distances is a Euclidean distance between the feature vector of the main object and a feature vector of each of the main models;

[0057] a first determining sub-unit, configured to determine a target main model matching the main object from the plurality of main models based on the plurality of first Euclidean distances, wherein a first Euclidean distance between the feature vector of the target main model and the feature vector of the main object is a minimum Euclidean distance in the plurality of first Euclidean distances.

[0058] Optionally, in the case that the subject object is an object, the first calculation subunit is specifically configured to:

[0059] calculate a plurality of second Euclidean distances respectively based on a plurality of sets of first parameter pairs, wherein each set of the first parameter pairs comprises a first parameter and a second parameter corresponding to the first parameter, the first parameter being a feature vector of the object at a preset orientation, and the second parameter being a feature vector of a first subject model at a corresponding preset orientation;

[0060] calculate a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between a feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0061] Optionally, in the case that the subject object is a person, the first calculation subunit is specifically configured to:

[0062] calculate a plurality of third Euclidean distances respectively based on a plurality of sets of second parameter pairs, wherein each set of the second parameter pairs comprises a third parameter and a fourth parameter corresponding to the third parameter, the third parameter being used to represent a skeletal joint of the person, and the fourth parameter being used to represent a corresponding skeletal joint of a first subject model;

[0063] calculate a sum of the plurality of third Euclidean distances to obtain a first Euclidean distance between a feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0064] Optionally, the first restoration module comprises:

[0065] a first acquisition unit configured to acquire actual shooting angles of the target position at the plurality of preset orientations;

[0066] a fourth processing unit configured to perform picture supplementation on the target model from the plurality of preset orientations based on the actual shooting angles of the target position at the plurality of preset orientations by using the virtual camera to obtain initial surround videos of the target position at the plurality of preset orientations;

[0067] a first extraction unit configured to extract key information of the plurality of original images;

[0068] a fifth processing unit configured to perform rendering on the initial surround videos of the target position at the plurality of preset orientations by using the key information to obtain first surround videos of the target position at the plurality of preset orientations.

[0069] Optionally, the apparatus further comprises:

[0070] The first sending module is configured to send, to the player, the first surround video of the target position at the plurality of preset orientations after aligning the playing perspective of the player with the actual shooting perspective.

[0071] Optionally, the apparatus further comprises:

[0072] The first receiving module is configured to receive the correction information sent by the player.

[0073] The first generating module is configured to generate a second surround video based on the correction information and the first surround video of the target position at the plurality of preset orientations.

[0074] The second sending module is configured to send, to the player, the second surround video of the target position at the plurality of preset orientations after aligning the playing perspective of the player with the actual shooting perspective.

[0075] In a third aspect, an electronic device is provided, which includes a transceiver and a processor, the transceiver being configured to:

[0076] obtain a plurality of original images of a target position shot from a plurality of preset orientations, wherein at least one original image is shot from each preset orientation;

[0077] The processor is configured to:

[0078] perform recognition analysis on the plurality of original images to determine a subject object in the plurality of original images.

[0079] construct a target model based on image information of the plurality of original images, the subject object, and a plurality of original models in a historical database.

[0080] perform picture supplementation on the target model from the plurality of preset orientations by using a virtual camera to restore a first surround video of the target position at the plurality of preset orientations.

[0081] Optionally, the processor is specifically configured to:

[0082] extract edge information, a color histogram, and a gradient histogram of each original image in the plurality of original images.

[0083] obtain a feature vector of each original image according to the edge information, the color histogram, and the gradient histogram of the original image.

[0084] obtain edge contour information of each original image according to the feature vector of the original image.

[0085] calculate a maximum closed contour area of each original image according to the edge contour information of the original image.

[0086] The objects corresponding to the maximum closed contour areas of each original image are feature-matched with each other, and in a case where the objects corresponding to the maximum closed contour areas of N original images are the same, the objects corresponding to the maximum closed contour areas of the N original images are determined as the main object, N being an integer greater than 1.

[0087] Optionally, the processor is specifically configured to:

[0088] determine a target main object model matching the main object from the plurality of main object models;

[0089] perform vector alignment between the target main object model and the scene model based on image information of the plurality of original images, superimpose the target main object model into the scene model to obtain a target model.

[0090] Optionally, the processor is specifically configured to:

[0091] calculate a plurality of first Euclidean distances between the feature vector of the main object and the feature vectors of the plurality of main object models respectively, wherein each first Euclidean distance is a Euclidean distance between the feature vector of the main object and the feature vector of each main object model;

[0092] determine a target main object model matching the main object from the plurality of main object models based on the plurality of first Euclidean distances, wherein a first Euclidean distance between the feature vector of the target main object model and the feature vector of the main object is the smallest Euclidean distance in the plurality of first Euclidean distances.

[0093] Optionally, in a case where the main object is an object, the processor is specifically configured to:

[0094] calculate a plurality of second Euclidean distances based on a plurality of first parameter pairs respectively, the first parameter pair including a first parameter and a second parameter corresponding to the first parameter, the first parameter being a feature vector of the object in a preset orientation, and the second parameter being a feature vector of a first main object model in a corresponding preset orientation;

[0095] calculate a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between the feature vector of the main object and the feature vector of the first main object model, the first main object model being any one of the plurality of main object models.

[0096] Optionally, in a case where the main object is a person, the processor is specifically configured to:

[0097] a plurality of third Euclidean distances are calculated based on a plurality of sets of second parameter pairs respectively, wherein each set of the second parameter pairs comprises a third parameter and a fourth parameter corresponding to the third parameter, the third parameter is used to represent a skeletal joint of the character, and the fourth parameter is used to represent a corresponding skeletal joint of a first subject model;

[0098] a sum of the plurality of third Euclidean distances is calculated to obtain a first Euclidean distance between a feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0099] Optionally, the processor is specifically configured to:

[0100] an actual shooting perspective of the target position at the plurality of preset orientations is obtained;

[0101] based on the actual shooting perspective of the target position at the plurality of preset orientations, the virtual camera is used to respectively perform picture supplementation on the target model from the plurality of preset orientations to obtain an initial surround video of the target position at the plurality of preset orientations;

[0102] key information of the plurality of original images is extracted;

[0103] the initial surround video of the target position at the plurality of preset orientations is rendered by using the key information to obtain a first surround video of the target position at the plurality of preset orientations.

[0104] Optionally, the transceiver is further configured to:

[0105] after aligning a playing perspective of a player with the actual shooting perspective, the first surround video of the target position at the plurality of preset orientations is sent to the player.

[0106] Optionally, the transceiver is further configured to:

[0107] correction information sent by a player is received;

[0108] based on the correction information and the first surround video of the target position at the plurality of preset orientations, a second surround video is generated;

[0109] after aligning a playing perspective of a player with the actual shooting perspective, the second surround video of the target position at the plurality of preset orientations is sent to the player.

[0110] In a fourth aspect, the embodiments of the present disclosure further provide an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor, when the computer program is executed by the processor, the steps of the video restoration method described above are implemented.

[0111] In a fifth aspect, the embodiments of the present disclosure further provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps of the video restoration method described above.

[0112] In a sixth aspect, the embodiments of the present disclosure provide a computer program product, which includes computer instructions. The computer instructions are executed by a processor to implement the steps of the method described in the first aspect.

[0113] The video restoration method of the embodiments of the present disclosure includes: acquiring a plurality of original images of a target position captured from a plurality of preset orientations, wherein at least one original image is captured from each preset orientation; performing identification analysis on the plurality of original images to determine a subject object in the plurality of original images; constructing a target model based on image information of the plurality of original images, the subject object, and a plurality of original models in a historical database; and using a virtual camera to perform picture supplementation on the target model from the plurality of preset orientations respectively to restore a first surround video of the target position at the plurality of preset orientations. The method reduces the deployment of cameras in the front-end physical environment and the amount of data transmission, and only uses a few original pictures in the physical environment as a basis to restore the video of the target position by using the virtual camera in the digital space, thereby reducing the difficulty of the video restoration process. BRIEF DESCRIPTION OF DRAWINGS

[0114] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings needed in the description of the embodiments of the present disclosure will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without creative labor.

[0115] FIG. 1 is a flowchart of a video restoration method provided by the embodiments of the present disclosure;

[0116] FIG. 2 is a schematic diagram of a video restoration method provided by the embodiments of the present disclosure;

[0117] FIG. 3 is a schematic diagram of edge contour information extraction of a person provided by the embodiments of the present disclosure;

[0118] FIG. 4 is a schematic diagram of picture supplementation on an initial surround video provided by the embodiments of the present disclosure;

[0119] FIG. 5 is a structural diagram of a video restoration device provided by an embodiment of the present disclosure;

[0120] FIG. 6 is a structural diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0121] The technical solutions in the embodiments of the present disclosure will be clearly and completely described with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are some, but not all, of the embodiments of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present disclosure.

[0122] The present disclosure provides a video restoration method. Referring to FIG. 1, FIG. 1 is a flowchart of a video restoration method provided by an embodiment of the present disclosure, as shown in FIG. 1, the method comprises the following steps, it should be noted that the following steps can be executed by a server:

[0123] Step 101, obtaining a plurality of original images photographed from a plurality of preset positions to a target position, wherein at least one original image is photographed from each preset position;

[0124] In this step, for example, three cameras can be erected in an actual physical environment to photograph original images of the top (assuming this position as Z-axis position), front (assuming this position as X-axis position) and side (assuming this position as Y-axis position) of the target position.

[0125] Step 102, identifying and analyzing the plurality of original images to determine a subject object in the plurality of original images;

[0126] In this step, the server does not splice the plurality of original images obtained from the plurality of preset positions, but performs image recognition analysis, determines the subject object of each original image according to the proportion of each object occupying the picture in each original image, and if the subject objects of multiple original images are the same, takes the subject object as the basis for subsequent analysis.

[0127] Step 103, constructing a target model based on image information of the plurality of original images, the subject object and a plurality of original models in a historical database;

[0128] In actual application, for example, some large-scale performances, the arrangement environment of the venue and the performance workers are known information before the actual performance live broadcast, which are stored in the historical database and form original models one by one. For example, different subject models can be formed according to the characteristic information of different performance workers, and scene models can be formed according to the arrangement environment information of the venue.

[0129] According to the subject object identified from the plurality of original images, a target subject model matched with the subject object is determined from a plurality of subject models stored in the historical database. Then, the target subject model obtained from the image information of the plurality of original images is bound with the scene model to obtain the target model.

[0130] Step 104, using virtual cameras to respectively add pictures to the target model from the plurality of preset orientations to restore the first surround video of the target position in the plurality of preset orientations.

[0131] In this step, the aforementioned target model is constructed in a digital space, and the essential picture information is less. A virtual camera can be obtained by 3D rendering, and then the virtual camera is used to respectively add pictures to the target model from a plurality of preset orientations to restore the first surround video of the target position in the plurality of preset orientations. Illustratively, a plurality of pictures are taken in the X orientation 360 degrees, and then a set of first surround videos based on the X orientation is generated by using a stitching synthesis algorithm, and the first surround videos of other orientations are generated in the same way.

[0132] The related volume photography technology mainly captures original images by deploying a plurality of cameras in a physical environment (for example, 24-120 cameras are needed to deploy for a free surround view), stitches all the original images by an algorithm, processes by a server, and then displays the picture by playing the video according to the viewing angle selected by the user. In one embodiment, a small number of cameras are deployed in the actual physical space, a plurality of original images of the target position taken from a plurality of preset orientations are obtained by the small number of cameras, the plurality of original images are not stitched but analyzed to determine the main object of the plurality of original images, and then a target model related to the main object is obtained from a historical database to restore the video picture of the target position in the digital space based on the target model. Taking FIG. 2 as an example, three original images are taken from the front, side and top of the cup by using cameras, the three original images are analyzed to determine that the main object of the three original images is a cup, and then a cup model matched with the cup in the historical database and the prompt information (such as a fluorescent lamp, a wooden table, etc.) extracted in the process of analyzing the three original images are used to restore the image picture of the cup in the digital space based on the cup model, and the restored image picture is stitched and synthesized by an algorithm. This embodiment reduces the deployment of cameras in the front-end physical environment and the amount of data transmission, and only uses a few original pictures in the physical environment to restore the video of the target position by using a virtual camera in the digital space, thereby reducing the difficulty of the video restoration process.

[0133] Optionally, the analyzing the plurality of original images to determine the main object in the plurality of original images comprises:

[0134] extracting edge information, color histogram and gradient histogram of each original image in the plurality of original images;

[0135] According to the edge information, the color histogram and the gradient histogram of each original image, a feature vector of each original image is obtained;

[0136] According to the feature vector of each original image, edge contour information of each original image is obtained;

[0137] According to the edge contour information of each original image, a maximum closed contour area of each original image is calculated;

[0138] The objects corresponding to the maximum closed contour areas of each original image are matched with each other, and in a case where the objects corresponding to the maximum closed contour areas of N original images are the same, the objects corresponding to the maximum closed contour areas of the N original images are determined as the main objects, and N is an integer greater than 1.

[0139] In an embodiment, the edge information, the color histogram and the gradient histogram of each original image are extracted, and the three are combined to form a complete feature vector. The extraction process of the feature vector can refer to the following formula: i = σ j (I i );

[0140] Wherein, F represents the feature vector, σ represents the feature extraction method, I represents the input original image, j represents the jth feature extraction method, and i represents the ith original image.

[0141] Then, the edge contour information of each original image is obtained according to the feature vector of each original image, and the maximum closed contour area of each original image is calculated through the Green formula. In the specific implementation process, in order to reduce the extraction time of the edge contour information, referring to FIG. 3, when it is determined that the object is a person, the OpenPose algorithm can be used to quickly extract the skeleton posture of the person in the picture. When the picture recognition main body is an object, the Contour Detection algorithm can be used to quickly extract the object contour line.

[0142] Then, whether the objects corresponding to the maximum closed contour areas in each original image are the same is determined through the feature matching method. If the objects corresponding to the maximum closed contour areas in multiple original images are the same, it is determined that the object is the main object. In this embodiment, by recognizing and analyzing multiple original images, the main object in the multiple original images is obtained, which is beneficial to subsequent restoration of the target position video based on the main object.

[0143] Optionally, the plurality of original models includes a scene model and a plurality of main body models, and the constructing a target model based on the image information of the plurality of original images, the main object and the plurality of original models in the historical database includes:

[0144] determine a target subject model matching the subject object from the plurality of subject models;

[0145] superimpose the target subject model into the scene model to obtain a target model after vector alignment of the target subject model and the scene model based on image information of the plurality of original images.

[0146] In actual applications, for example, some large performances, the arrangement environment of the venue and the performance workers are known information before the actual performance live broadcast. These information are stored in a historical database and form original models one by one. Exemplarily, different subject models can be formed according to the characteristic information of different performance workers, and a scene model can be formed according to the arrangement environment information of the venue.

[0147] In an implementation, the subject object determined above is matched with a plurality of subject models in a database. Specifically, the Euclidean distance between the characteristic vector of the subject object and the characteristic vector of each subject model can be calculated, and the subject model corresponding to the minimum Euclidean distance calculated is determined as the subject model most matching the subject object, that is, the target subject model. The target subject model determined is bound with the scene model. In the binding process, the image information of the plurality of original images is used as a basis. The plurality of image information comes from different preset orientations, for example, X-axis orientation, Y-axis orientation and Z-axis orientation. Therefore, the target subject model and the scene model are vector aligned in each orientation by referring to the real image information of each orientation, so that the target model obtained by superimposing the target subject model and the scene model can more truly restore the actual situation of the target position.

[0148] Optionally, the determining a target subject model matching the subject object from the plurality of subject models comprises:

[0149] a plurality of first Euclidean distances between the characteristic vector of the subject object and the characteristic vectors of the plurality of subject models are respectively calculated, wherein each first Euclidean distance is the Euclidean distance between the characteristic vector of the subject object and the characteristic vector of each subject model;

[0150] a target subject model matching the subject object is determined from the plurality of subject models based on the plurality of first Euclidean distances, wherein the first Euclidean distance between the characteristic vector of the target subject model and the characteristic vector of the subject object is the minimum Euclidean distance in the plurality of first Euclidean distances.

[0151] In an embodiment, the target subject model can be determined from the plurality of subject models by calculating a first Euclidean distance between the feature vector of the subject object and the feature vector of each subject model, respectively. The first Euclidean distance can be calculated according to the following formula:

[0152] wherein d represents the first Euclidean distance, a1 represents the feature vector of the subject object, and a2 represents the feature vector of the subject model.

[0153] The first Euclidean distance between the feature vector of the target subject model and the subject object is the smallest among the plurality of calculated first Euclidean distances, indicating that the target subject model has the greatest correlation with the subject object, and is the subject model that best matches the subject object among the plurality of subject models.

[0154] In this embodiment, by calculating the first Euclidean distance between the feature vector of the subject object and the feature vector of each subject model in the plurality of subject models, the target subject model that matches the subject object can be quickly determined from the plurality of subject models.

[0155] Optionally, in the case where the subject object is an object, the calculating of the plurality of first Euclidean distances between the feature vector of the subject object and the feature vectors of the plurality of subject models comprises:

[0156] calculating a plurality of second Euclidean distances based on a plurality of sets of first parameter pairs, wherein the first parameter pair comprises a first parameter and a second parameter corresponding to the first parameter, the first parameter is a feature vector of the object at a preset orientation, and the second parameter is a feature vector of a first subject model at a corresponding preset orientation;

[0157] calculating a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between the feature vector of the subject object and the feature vector of the first subject model, wherein the first subject model is any one of the plurality of subject models.

[0158] In an embodiment, in the case where the subject object is an object, the second Euclidean distance between the feature vector of the object at each preset orientation and the feature vector of each subject model at a corresponding preset orientation can be calculated first, and then the second Euclidean distances calculated at each preset orientation are added to obtain the first Euclidean distance. For example, in the case where the plurality of preset orientations are X-axis orientation, Y-axis orientation, and Z-axis orientation, the first Euclidean distance in this embodiment can be calculated according to the following formula:

[0159] Wherein, d represents the first Euclidean distance, x1 represents the feature vector of the object in the x-axis direction, x2 represents the feature vector of the subject model in the x-axis direction, y1 represents the feature vector of the object in the y-axis direction, y2 represents the feature vector of the subject model in the y-axis direction, z1 represents the feature vector of the object in the z-axis direction, and z2 represents the feature vector of the subject model in the z-axis direction.

[0160] In this implementation, when the subject object is an object, by first calculating the second Euclidean distance between the feature vector of the object in each preset direction and the feature vector of the subject model in each preset direction, and then obtaining the first Euclidean distance based on the second Euclidean distance, the accuracy of the calculated first Euclidean distance can be improved according to the type of the subject object.

[0161] Optionally, when the subject object is a person, the method further includes:

[0162] Calculating a plurality of third Euclidean distances based on a plurality of groups of second parameter pairs, wherein each group of second parameter pairs includes a third parameter and a fourth parameter corresponding to the third parameter, the third parameter being used to represent a skeletal joint of the person, and the fourth parameter being used to represent a corresponding skeletal joint of the first subject model;

[0163] Calculating a sum of the plurality of third Euclidean distances to obtain the first Euclidean distance between the feature vector of the subject object and the feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0164] In one implementation, when the subject object is a person, the first Euclidean distance can be calculated based on a plurality of skeletal joints on the skeletal line of the person. Specifically, the third Euclidean distance between each skeletal joint of the person and the corresponding skeletal joint of the subject model can be calculated first, and a plurality of third Euclidean distances can be calculated by a plurality of pairs of skeletal joints, and then the plurality of third Euclidean distances are added to obtain the first Euclidean distance. The calculation of the first Euclidean distance in this implementation can be referred to the following formula: i i

[0165] Wherein, d represents the first Euclidean distance, a represents the skeletal joint of the person, b represents the skeletal joint of the subject model corresponding to the skeletal joint of the person, and i represents the i-th skeletal joint.

[0166] ​​In the embodiment, when the subject object is a person, the third Euclidean distance between each skeletal joint on the skeleton line of the person and the corresponding skeletal joint of the subject model is calculated first, and then the first Euclidean distance is obtained based on the third Euclidean distance, so as to improve the accuracy of the calculated first Euclidean distance according to the type of the subject object.

[0167] Optionally, the picture supplementing of the target model from the plurality of preset orientations by using the virtual camera to restore the first surround video of the target position at the plurality of preset orientations comprises:

[0168] obtaining actual shooting angles of the target position at the plurality of preset orientations;

[0169] based on the actual shooting angles of the target position at the plurality of preset orientations, picture supplementing of the target model from the plurality of preset orientations by using the virtual camera to obtain initial surround videos of the target position at the plurality of preset orientations;

[0170] extracting key information of the plurality of original images;

[0171] rendering the initial surround videos of the target position at the plurality of preset orientations by using the key information to obtain the first surround videos of the target position at the plurality of preset orientations.

[0172] In an embodiment, the preset orientations include X-axis orientations, Y-axis orientations and Z-axis orientations. The actual shooting angle of the target position at the X-axis orientations is θ. A virtual camera is constructed around θ in the digital space by using 3D rendering, a plurality of pictures of F(x)=θ(0°-360°) are shot by using the virtual camera, and a set of 360° initial surround videos based on the X-axis orientations is generated by using a splicing synthesis algorithm on the plurality of pictures shot around θ. Similarly, 360° initial surround videos of the Y-axis orientations and 360° initial surround videos of the Z-axis orientations are obtained.

[0173] Referring to FIG. 4, since the initial surround videos obtained have less substantial picture information, key information can be extracted from the original images to form a “key information controller”, and the initial surround videos are rendered by using a control network (C-Net) algorithm. The key information can include color depth, color accuracy, color gamut, light, color tone and picture style of the original images, so as to obtain the first surround videos after the initial videos are rendered.

[0174] In the embodiment, based on the actual shooting angle of the target position, the target model is supplemented with a virtual camera, and then the obtained picture is rendered again with the original image, which is beneficial to improve the restoration degree of the obtained video of the target position.

[0175] Optionally, after the target model is supplemented with a virtual camera from the plurality of preset directions to restore the first surround video of the target position in the plurality of preset directions, the method further comprises:

[0176] After aligning the playing angle of the player with the actual shooting angle, the first surround video of the target position in the plurality of preset directions is sent to the player.

[0177] With reference to the examples of the foregoing embodiments, assuming that the actual shooting angle of the target position in the X-axis direction is θ, the playing angle of the player in the X-axis direction is θ 0 Then, after the player obtains the first surround video of the target position in the X-axis direction sent by the server, the user can watch the first surround video by sliding the dial on the player, and the angle adjustment and video restoration of other preset directions are the same.

[0178] In the embodiment, aligning the playing angle of the player with the actual shooting angle is beneficial to synchronize the actual picture of the target position when the user watches the video on the player, and improve the restoration degree of the obtained video of the target position.

[0179] Optionally, after the target model is supplemented with a virtual camera from the plurality of preset directions to restore the first surround video of the target position in the plurality of preset directions, the method further comprises:

[0180] Receiving the correction information sent by the player;

[0181] Based on the correction information and the first surround video of the target position in the plurality of preset directions, a second surround video is generated;

[0182] After aligning the playing angle of the player with the actual shooting angle, the second surround video of the target position in the plurality of preset directions is sent to the player.

[0183] In an implementation, the user can also modify the first surround video according to his / her needs and preferences. For example, the user wants to change the dancers in the performance into rabbits and change the style of the performance into a forest style. The modification information can include the two variables of rabbits and forest style. Then, the first surround video is modified according to the modification information to obtain the second surround video that meets the needs and preferences of the user. This implementation is advantageous to increase the diversity of the video restored to the target position and improve the interest of the user in watching the video on the player.

[0184] Referring to FIG. 5, FIG. 5 is a structural diagram of a video restoration apparatus according to an embodiment of the present disclosure. As shown in FIG. 6, the video restoration apparatus 500 includes:

[0185] A first acquisition module 501 is configured to acquire a plurality of original images captured from a plurality of preset positions of a target position, wherein each preset position captures at least one original image.

[0186] A first determination module 502 is configured to perform recognition analysis on the plurality of original images to determine a subject object in the plurality of original images.

[0187] A first construction module 503 is configured to construct a target model based on image information of the plurality of original images, the subject object, and a plurality of original models in a historical database.

[0188] A first restoration module 504 is configured to use a virtual camera to perform picture supplementation on the target model from the plurality of preset positions to restore a first surround video of the target position at the plurality of preset positions.

[0189] Optionally, the first determination module includes:

[0190] A first extraction unit is configured to extract edge information, a color histogram, and a gradient histogram of each original image in the plurality of original images.

[0191] A first processing unit is configured to obtain a feature vector of each original image according to the edge information, the color histogram, and the gradient histogram of the original image.

[0192] A second processing unit is configured to obtain edge contour information of each original image according to the feature vector of the original image.

[0193] A first calculation unit is configured to calculate a maximum closed contour area of each original image according to the edge contour information of the original image.

[0194] The first determining unit is configured to perform feature matching on objects corresponding to the maximum closed contour areas of the original images, and determine the objects corresponding to the maximum closed contour areas of the N original images as the main objects when the objects corresponding to the maximum closed contour areas of the N original images are the same, where N is an integer greater than 1.

[0195] Optionally, the plurality of original models include a scene model and a plurality of main object models, and the first constructing module includes:

[0196] The second determining unit is configured to determine a target main object model matching the main object from the plurality of main object models.

[0197] The third processing unit is configured to perform vector alignment on the target main object model and the scene model based on image information of the plurality of original images, superimpose the target main object model into the scene model, and obtain a target model.

[0198] Optionally, the second determining unit includes:

[0199] The first calculating sub-unit is configured to calculate a plurality of first Euclidean distances between a feature vector of the main object and feature vectors of the plurality of main object models respectively, where each first Euclidean distance is a Euclidean distance between the feature vector of the main object and a feature vector of each main object model.

[0200] The first determining sub-unit is configured to determine a target main object model matching the main object from the plurality of main object models based on the plurality of first Euclidean distances, where a first Euclidean distance between the feature vector of the target main object model and the feature vector of the main object is the smallest Euclidean distance in the plurality of first Euclidean distances.

[0201] Optionally, in a case where the main object is an object, the first calculating sub-unit is specifically configured to:

[0202] calculate a plurality of second Euclidean distances based on a plurality of first parameter pairs respectively, where the first parameter pair includes a first parameter and a second parameter corresponding to the first parameter, the first parameter is a feature vector of the object in a preset orientation, and the second parameter is a feature vector of a first main object model in a corresponding preset orientation;

[0203] calculate a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between the feature vector of the main object and a feature vector of the first main object model, where the first main object model is any one of the plurality of main object models.

[0204] Optionally, in a case where the main object is a person, the first calculating sub-unit is specifically configured to:

[0205] calculate a plurality of third Euclidean distances respectively based on a plurality of sets of second parameter pairs, wherein each set of the second parameter pairs comprises a third parameter and a fourth parameter corresponding to the third parameter, the third parameter being used to represent a skeletal joint of the character, and the fourth parameter being used to represent a corresponding skeletal joint of a first subject model;

[0206] calculate a sum of the plurality of third Euclidean distances to obtain a first Euclidean distance between a feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0207] Optionally, the first restoring module comprises:

[0208] a first obtaining unit, configured to obtain actual shooting angles of the target position at the plurality of preset orientations;

[0209] a fourth processing unit, configured to perform picture supplementation on the target model from the plurality of preset orientations respectively based on the actual shooting angles of the target position at the plurality of preset orientations by using the virtual camera, to obtain initial surround videos of the target position at the plurality of preset orientations.

[0210] a first extracting unit, configured to extract key information of the plurality of original images;

[0211] a fifth processing unit, configured to perform rendering on the initial surround videos of the target position at the plurality of preset orientations by using the key information, to obtain first surround videos of the target position at the plurality of preset orientations.

[0212] Optionally, the apparatus further comprises:

[0213] a first sending module, configured to send the first surround videos of the target position at the plurality of preset orientations to a player after aligning a playing angle of the player with the actual shooting angles.

[0214] Optionally, the apparatus further comprises:

[0215] a first receiving module, configured to receive correction information sent by a player;

[0216] a first generating module, configured to generate second surround videos based on the correction information and the first surround videos of the target position at the plurality of preset orientations;

[0217] a second sending module, configured to send the second surround videos of the target position at the plurality of preset orientations to the player after aligning the playing angle of the player with the actual shooting angles.

[0218] The video restoration apparatus 500 can implement each process of the method embodiment shown in FIG. 1 and achieve the same beneficial effects. To avoid repetition, details are not described herein.

[0219] The embodiments of the present disclosure further provide an electronic device, comprising a processor, a memory, and a program stored in the memory and executable on the processor. The program, when executed by the processor, implements each process of the video restoration method embodiments applied to the electronic device and achieves the same technical effects. To avoid repetition, details are not described herein.

[0220] Specifically, referring to FIG. 6, the embodiments of the present disclosure further provide an electronic device, comprising a bus 601, a transceiver 602, an antenna 603, a bus interface 604, a processor 605, and a memory 606.

[0221] The transceiver 602 is configured to:

[0222] obtain a plurality of original images of a target position captured from a plurality of preset orientations, wherein at least one original image is captured from each preset orientation;

[0223] The processor 605 is configured to:

[0224] perform recognition analysis on the plurality of original images to determine a subject object in the plurality of original images;

[0225] construct a target model based on image information of the plurality of original images, the subject object, and a plurality of original models in a historical database;

[0226] perform picture supplementation on the target model from the plurality of preset orientations by using a virtual camera to restore a first surround video of the target position at the plurality of preset orientations.

[0227] Optionally, the processor 605 is specifically configured to:

[0228] extract edge information, a color histogram, and a gradient histogram of each original image in the plurality of original images;

[0229] obtain a feature vector of each original image according to the edge information, the color histogram, and the gradient histogram of the original image;

[0230] obtain edge contour information of each original image according to the feature vector of the original image;

[0231] calculate a maximum closed contour area of each original image according to the edge contour information of the original image;

[0232] The objects corresponding to the maximum closed contour areas of each original image are matched with each other in features, and in a case that the objects corresponding to the maximum closed contour areas of N original images are the same, the objects corresponding to the maximum closed contour areas of the N original images are determined as the main object, N being an integer greater than 1.

[0233] Optionally, the processor 605 is specifically configured to:

[0234] determine a target main object model matched with the main object from the plurality of main object models;

[0235] superimpose the target main object model into the scene model after vector alignment of the target main object model and the scene model based on image information of the plurality of original images, to obtain a target model.

[0236] Optionally, the processor 605 is specifically configured to:

[0237] calculate a plurality of first Euclidean distances between the feature vector of the main object and the feature vectors of the plurality of main object models respectively, wherein each first Euclidean distance is a Euclidean distance between the feature vector of the main object and the feature vector of each main object model;

[0238] determine a target main object model matched with the main object from the plurality of main object models based on the plurality of first Euclidean distances, wherein a first Euclidean distance between the feature vector of the target main object model and the feature vector of the main object is the minimum Euclidean distance in the plurality of first Euclidean distances.

[0239] Optionally, in a case that the main object is an object, the processor 605 is specifically configured to:

[0240] calculate a plurality of second Euclidean distances based on a plurality of groups of first parameter pairs respectively, the first parameter pair including a first parameter and a second parameter corresponding to the first parameter, the first parameter being a feature vector of the object in a preset orientation, and the second parameter being a feature vector of the main object model in a corresponding preset orientation;

[0241] calculate a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between the feature vector of the main object and the feature vector of the first main object model, the first main object model being any one of the plurality of main object models.

[0242] Optionally, in a case that the main object is a person, the processor 605 is specifically configured to:

[0243] Calculate a plurality of third Euclidean distances based on a plurality of sets of second parameter pairs respectively, wherein each set of the second parameter pairs comprises a third parameter and a fourth parameter corresponding to the third parameter, the third parameter being used to represent a skeletal joint of the character, and the fourth parameter being used to represent a corresponding skeletal joint of a first subject model;

[0244] Calculate a sum of the plurality of third Euclidean distances to obtain a first Euclidean distance between a feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

[0245] Optionally, the processor 605 is specifically configured to:

[0246] Obtain actual shooting angles of the target position at the plurality of preset orientations;

[0247] Based on the actual shooting angles of the target position at the plurality of preset orientations, use the virtual camera to respectively perform picture supplementation on the target model from the plurality of preset orientations to obtain initial surround videos of the target position at the plurality of preset orientations.

[0248] Extract key information of the plurality of original images;

[0249] Use the key information to render the initial surround videos of the target position at the plurality of preset orientations to obtain first surround videos of the target position at the plurality of preset orientations.

[0250] Optionally, the transceiver 602 is further configured to:

[0251] After aligning the playing angle of the player with the actual shooting angle, send the first surround videos of the target position at the plurality of preset orientations to the player.

[0252] Optionally, the transceiver 602 is further configured to:

[0253] Receive correction information sent by the player;

[0254] Based on the correction information and the first surround videos of the target position at the plurality of preset orientations, generate second surround videos;

[0255] After aligning the playing angle of the player with the actual shooting angle, send the second surround videos of the target position at the plurality of preset orientations to the player.

[0256] In Figure 6, a bus architecture (represented by bus 601) can include any number of interconnected buses and bridges, the bus 601 linking together various circuits including one or more processors represented by processor 605 and memory represented by memory 606. The bus 601 can also link various other circuits together, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art, and thus, not further described herein. A bus interface 604 provides an interface between the bus 601 and the transceiver 602. The transceiver 602 can be a single element or multiple elements, such as a plurality of receivers and transmitters, that provide a means for communicating with various other apparatus over a transmission medium. Data processed by the processor 605 is transmitted over a wireless medium via the antenna 603, and further, the antenna 603 receives data and communicates the data to the processor 605.

[0257] The processor 605 is responsible for managing the bus 601 and general processing, and can also provide various functions including timing, peripheral interfaces, voltage regulation, power management, and other control functions. The memory 606 can be used to store data used by the processor 605 during execution of operations.

[0258] The embodiments of the present disclosure further provide a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to implement each process of the video restoration method embodiments and achieve the same technical effects. To avoid repetition, details are not described herein. The computer readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0259] The embodiments of the present disclosure further provide a computer program product, which includes computer instructions. The computer instructions are executed by a processor to implement each process of the method embodiments shown in Figure 1 and achieve the same technical effects. To avoid repetition, details are not described herein.

[0260] It should be noted that, in this document, the term "comprising" or "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that processes, methods, articles, or apparatuses including a series of elements not only include those elements, but also include other elements not explicitly listed, or further include elements inherent to such processes, methods, articles, or apparatuses. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the process, method, article, or apparatus including the element.

[0261] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present disclosure can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes a plurality of instructions for making a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) execute the method described in various embodiments of the present disclosure.

[0262] The embodiments of the present disclosure are described above in combination with the drawings, but the present disclosure is not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative, not limiting, and those of ordinary skill in the art can make many forms without departing from the purpose of the present disclosure and the scope protected by the claims under the inspiration of the present disclosure, which all belong to the protection of the present disclosure.

Claims

1. A video restoration method, the method comprising: acquiring a plurality of original images captured from a plurality of preset orientations towards a target location, wherein at least one original image is captured from each preset orientation; performing recognition analysis on the plurality of original images to determine a subject object in the plurality of original images; constructing a target model based on image information of the plurality of original images, the subject object, and a plurality of original models in a historical database; performing picture supplementation on the target model from the plurality of preset orientations respectively by using a virtual camera to restore a first surround video of the target location at the plurality of preset orientations.

2. The video restoration method according to claim 1, wherein: The plurality of original models comprises a scene model and a plurality of subject models, and the constructing a target model based on the image information of the plurality of original images, the subject object, and the plurality of original models in the historical database comprises: determining a target subject model matched with the subject object from the plurality of subject models; performing vector alignment between the target subject model and the scene model based on the image information of the plurality of original images, and then superimposing the target subject model into the scene model to obtain the target model.

3. The video restoration method of claim 2, wherein, The determining a target subject model matched with the subject object from the plurality of subject models comprises: calculating a plurality of first Euclidean distances between a feature vector of the subject object and feature vectors of the plurality of subject models respectively, wherein each first Euclidean distance is a Euclidean distance between the feature vector of the subject object and a feature vector of each subject model; determining the target subject model matched with the subject object from the plurality of subject models based on the plurality of first Euclidean distances, wherein a first Euclidean distance between the feature vector of the target subject model and the feature vector of the subject object is the smallest Euclidean distance in the plurality of first Euclidean distances.

4. The video restoration method of claim 3, wherein, In a case where the subject object is an object, the calculating a plurality of first Euclidean distances between a feature vector of the subject object and feature vectors of the plurality of subject models respectively comprises: calculating a plurality of second Euclidean distances based on a plurality of first parameter pairs respectively, wherein the first parameter pair comprises a first parameter and a second parameter corresponding to the first parameter, the first parameter is a feature vector of the object at a preset orientation, and the second parameter is a feature vector of a first subject model at a corresponding preset orientation; calculating a sum of the plurality of second Euclidean distances to obtain a first Euclidean distance between the feature vector of the subject object and a feature vector of the first subject model, wherein the first subject model is any one of the plurality of subject models.

5. The video restoration method of claim 3, wherein, In a case where the subject object is a person, the calculating a plurality of first Euclidean distances between a feature vector of the subject object and feature vectors of the plurality of subject models respectively comprises: calculating a plurality of third Euclidean distances based on a plurality of second parameter pairs respectively, wherein each second parameter pair comprises a third parameter and a fourth parameter corresponding to the third parameter, the third parameter is used to represent a skeletal joint of the person, and the fourth parameter is used to represent a corresponding skeletal joint of a first subject model; The sum of the plurality of third Euclidean distances is calculated to obtain a first Euclidean distance between a feature vector of the subject object and a feature vector of the first subject model, the first subject model being any one of the plurality of subject models.

6. The video restoration method of claim 1, wherein, The method further comprises the following steps after restoring the first surround video of the target position at the plurality of preset orientations by using the virtual cameras to respectively perform picture supplementation on the target model from the plurality of preset orientations: obtaining actual shooting angles of the target position at the plurality of preset orientations; based on the actual shooting angles of the target position at the plurality of preset orientations, using the virtual cameras to respectively perform picture supplementation on the target model from the plurality of preset orientations to obtain initial surround videos of the target position at the plurality of preset orientations; extracting key information of the plurality of original images; using the key information to render the initial surround videos of the target position at the plurality of preset orientations to obtain the first surround videos of the target position at the plurality of preset orientations.

7. The video restoration method of claim 6, wherein, The method further comprises the following steps after restoring the first surround video of the target position at the plurality of preset orientations by using the virtual cameras to respectively perform picture supplementation on the target model from the plurality of preset orientations: aligning a playing angle of a player with the actual shooting angles, and sending the first surround videos of the target position at the plurality of preset orientations to the player; or, receiving correction information sent by the player; based on the correction information and the first surround videos of the target position at the plurality of preset orientations, generating second surround videos; aligning a playing angle of a player with the actual shooting angles, and sending the second surround videos of the target position at the plurality of preset orientations to the player. 8.An electronic device comprising a transceiver, a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the video restoration method according to any one of claims 1 to 7. 9.A computer readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the video restoration method according to any one of claims 1 to 7. 10.A computer program product stored in a storage medium, wherein the program product is executed by at least one processor to implement the steps of the video restoration method according to any one of claims 1 to 7.

Images