Digital anchor virtual image generation method and system based on big data artificial intelligence

Abstract

The invention relates to a digital anchor virtual image generation method and system based on big data artificial intelligence, and relates to the field of computer technology.The method comprises the steps that broadcast characters and broadcast videos of news are obtained, word segmentation processing is carried out on the broadcast characters to obtain multiple first word combinations of the broadcast characters, and the multiple first word combinations of the broadcast characters are obtained; performing character recognition and feature recognition on the broadcast video to obtain a second word combination and features in the broadcast video, determining the news field to which the news belongs based on the first word combination, the second word combination and the features, determining the wearing of the reference digital anchor based on the news field to which the news belongs, and mapping the wearing to the reference digital anchor. And obtaining a target digital anchor, determining the position of the target digital anchor based on the broadcast video, and mapping the target digital anchor to the position in the broadcast video. The method and the device have the effect of reducing information display missing caused by shielding of the digital anchor.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method and system for generating virtual images of digital anchors based on big data artificial intelligence. Background Art

[0002] With the development of technologies such as artificial intelligence and AI creation, the emergence of news broadcasts through virtual images, digital anchors, etc. in the digital media field has made it no longer necessary for real people to prepare scripts and appear on camera for news broadcasts, reducing labor costs and broadcast errors.

[0003] However, when digital anchors conduct news broadcasts, since the position of the digital anchor in the news video is usually set by relevant personnel, the image of the digital anchor will block a part of the news video, resulting in the lack of information display in the news video. Therefore, how to reduce the lack of information display caused by the occlusion of the digital anchor has become a problem. Summary of the Invention

[0004] In order to reduce the lack of information display caused by the occlusion of the digital anchor, this application provides a method and system for generating virtual images of digital anchors based on big data artificial intelligence.

[0005] In a first aspect, this application provides a method for generating a virtual image of a digital anchor based on big data artificial intelligence, adopting the following technical solution: A method for generating a virtual image of a digital anchor based on big data artificial intelligence includes: Obtain the broadcast text and broadcast video of the news; Perform word segmentation on the broadcast text to obtain multiple first word combinations of the broadcast text; Perform text recognition and feature recognition on the broadcast video to obtain second word combinations and features in the broadcast video; Determine the news field to which the news belongs based on the first word combination, second word combination, and features; Determine the clothing of the reference digital anchor based on the news field and map the clothing to the reference digital anchor to obtain the target digital anchor; Determine the position of the target digital anchor based on the broadcast video and map the target digital anchor to the position in the broadcast video.

[0006] By adopting the above technical solutions, obtaining the broadcast text and broadcast video facilitates subsequent analysis. The broadcast text is segmented to obtain multiple first word combinations that make up the broadcast text. The broadcast video is subjected to text recognition and feature recognition to obtain second word combinations in the broadcast video. The first word combinations, second word combinations, and features in the broadcast video are all key factors reflecting the news field to which the news belongs. Therefore, according to the first word combinations, second word combinations, and features, the news field to which the news belongs can be accurately determined. After determining the required news field, the clothing on the reference digital anchor is determined, so that the image of the digital anchor is more in line with the news field. Then, the clothing is mapped onto the reference digital anchor to obtain the target digital anchor. Since the picture changes at each position in the broadcast video are different, the position of the target digital anchor on the broadcast video is determined according to the broadcast video, that is, the position with the least information loss caused by the news broadcast, and the target digital anchor is mapped to the determined position, finally achieving the effect of reducing the information display loss caused by the occlusion of the digital anchor.

[0007] In another possible implementation manner, each news field corresponds to a preset word library and a preset feature library. The determining the news field to which the news belongs based on the first word combinations, second word combinations, and features includes: Determining the first quantity of the multiple first word combinations and the second word combinations in each preset word library; Determining the second quantity of the multiple features in each preset feature library; Determining multiple target data sets, where each target data set is each second word combination and the features in the picture of the broadcast video at the time point when each second word combination appears; Determining the news field hit by each target data set and the third quantity of the number of times each hit news field appears, and determining the preferred field of each target data set as the news field with the most third quantity; Determining the news field to which the news belongs based on the first quantity, second quantity, and preferred field.

[0008] In another possible implementation manner, the determining the news field to which the news belongs based on the first quantity, second quantity, and preferred field includes: Sorting the first quantity to obtain the target preset word library with the most first quantity; Sorting the second quantity to obtain the target preset feature library with the most second quantity; Summarizing the preferred fields of all target data sets to obtain the target preferred field with the most occurrences; If there are at least two identical ones among the news fields of the target preset word library, the news fields of the target preset feature library, and the target preference field, then determine the at least two identical fields as the news field to which the news belongs.

[0009] In another possible implementation manner, determining the position of the target digital anchor based on the broadcast video includes: Performing edge detection on the target digital anchor to obtain the contour of the target digital anchor; Map the contour to the starting position in the lower left corner of the broadcast video, and horizontally translate the contour at a preset step length to obtain a plurality of candidate regions, where the candidate regions include the region where the contour is located when it is at the starting position; Segment the broadcast video according to each candidate region to obtain the region video of each candidate region; Determine each frame of the region video, calculate the first information entropy of each frame and the similarity between adjacent frames; Draw a target region centered on the contour and with a preset width, and segment each frame of the broadcast video according to the target region to obtain the target region images of each candidate region; Determine the pixel difference at the same position between two adjacent frames to obtain an absolute difference map, and calculate the differential entropy of each absolute difference map; Determine the first importance of each candidate region based on the first information entropy, differential entropy, and similarity; Calculate the second information entropy of each target region image of each region video and determine the gray values of the pixels on both sides of the contour; Determine the second importance of each candidate region based on the second information entropy and gray values; Determine the total importance based on the first importance and the second importance, and determine the candidate region with the lowest total importance as the position of the target digital anchor.

[0010] In another possible implementation manner, determining the first importance of each candidate region based on the first information entropy, differential entropy, and similarity includes: Calculate the first average value of the information entropy of each region video and the average value of the differential entropy of each absolute difference map; Calculate the second average value of the similarity between adjacent frames of each region video and the similarity variance; Determine the first importance based on the first average value, differential entropy average value, second average value, similarity variance, and their respective corresponding coefficients.

[0011] In another possible implementation manner, determining the second importance degree of each candidate region based on the second information entropy and the grayscale value includes: Dividing each target region screen of each region video into multiple sub-region screens; Determining the third average value of the grayscale values corresponding to both sides of the contour of each sub-region screen and determining the difference between the third average values; Summing the differences of each sub-region screen to obtain the total value of each target region screen with respect to the difference; Based on the total value of each sub-region screen, determining the fourth average value of each candidate region with respect to the total value, based on the second information entropy of each sub-region screen, determining the fifth average value of each candidate region with respect to the second information entropy, and determining the second importance degree of each candidate region based on the fourth average value, the fifth average value, and their respective corresponding coefficients.

[0012] In another possible implementation manner, each news field belongs to multiple preset dressing combinations, and each preset dressing combination corresponds to multiple tags. Determining the dressing of the reference digital anchor based on the news field to which it belongs includes: Determining the fourth quantity of the first word combination, the second word combination, and the tags that feature hits each preset dressing combination; Determining the preset dressing combination with the largest fourth quantity as the dressing of the reference digital anchor.

[0013] In a second aspect, the present application provides a digital anchor virtual image generation system based on big data artificial intelligence, adopting the following technical solution: A digital anchor virtual image generation system based on big data artificial intelligence includes: An acquisition module, configured to acquire the broadcast text and broadcast video of the news; A first processing module, configured to perform word segmentation processing on the broadcast text to obtain multiple first word combinations of the broadcast text; A second processing module, configured to perform text recognition and feature recognition on the broadcast video to obtain the second word combination and features in the broadcast video; A field determination module, configured to determine the news field to which the news belongs based on the first word combination, the second word combination, and the features; A dressing mapping module, configured to determine the dressing of the reference digital anchor based on the news field to which it belongs and map the dressing to the reference digital anchor to obtain a target digital anchor; A position mapping module, configured to determine the position of the target digital anchor based on the broadcast video and map the target digital anchor to the position in the broadcast video.

[0014] By adopting the above technical solution, the acquisition module acquires the broadcast text and the broadcast video for subsequent analysis. The first processing module performs word segmentation on the broadcast text to obtain multiple first word combinations that make up the broadcast text. The second processing module performs text recognition and feature recognition on the broadcast video to obtain second word combinations in the broadcast video. The first word combinations, the second word combinations, and the features in the broadcast video are all key factors reflecting the news field to which the news belongs. Therefore, the field determination module can accurately determine the news field to which the news belongs based on the first word combinations, the second word combinations, and the features. After determining the required news field, the dressing mapping module determines the dressing on the reference digital anchor, so that the image of the digital anchor better fits the news field, and then maps the dressing to the reference digital anchor to obtain the target digital anchor. Since the picture changes at each position in the broadcast video are different, the position mapping module determines the position of the target digital anchor on the broadcast video according to the broadcast video, that is, the position with the least information loss for the news broadcast, and maps the target digital anchor to the determined position, finally achieving the effect of reducing the information display loss caused by the occlusion of the digital anchor.

[0015] In another possible implementation manner, each news field corresponds to a preset word library and a preset feature library. When the field determination module determines the news field to which the news belongs based on the word combinations and features, it specifically is used for: Determine the first quantity of the multiple first word combinations and the second word combinations in each preset word library; Determine the second quantity of the multiple features in each preset feature library; Determine multiple target data sets, where each target data set is each second word combination and the features in the picture of the broadcast video at the time point when each second word combination appears; Determine the news field hit by each target data set and the third quantity of the number of times each hit news field appears, and determine the preferred field of each target data set as the news field with the most third quantity; Determine the news field to which the news belongs based on the first quantity, the second quantity, and the preferred field.

[0016] In another possible implementation manner, when the field determination module determines the news field to which the news belongs based on the first quantity, the second quantity, and the preferred field, it specifically is used for: Sort the first quantity to obtain the target preset word library with the most first quantity; Sort the second quantity to obtain the target preset feature library with the most second quantity; Summarize the preferred fields of all target data sets to obtain the target preferred field with the most occurrences; If there are at least two identical ones among the news fields of the target preset word library, the news fields of the target preset feature library, and the target preference fields, then determine the at least two identical fields as the news field to which the news belongs.

[0017] In another possible implementation manner, when the position mapping module determines the position of the target digital anchor based on the broadcast video, it specifically is used for: Perform edge detection on the target digital anchor to obtain the contour of the target digital anchor; Map the contour to the starting position at the lower left corner of the broadcast video, and horizontally translate the contour at a preset step length to obtain a plurality of candidate regions, where the candidate regions include the region where the contour is located when it is at the starting position; Segment the broadcast video according to each candidate region to obtain the region video of each candidate region; Determine each frame of the region video, calculate the first information entropy of each frame and the similarity between adjacent frames; Draw a target region centered on the contour and with a preset width, and segment each frame of the broadcast video according to the target region to obtain the target region images of each candidate region; Determine the pixel difference at the same position between two adjacent frames to obtain an absolute difference map, and calculate the difference entropy of each absolute difference map; Determine the first importance degree of each candidate region based on the first information entropy, difference entropy, and similarity; Calculate the second information entropy of each target region image of each region video and determine the gray values of the pixels on both sides of the contour; Determine the second importance degree of each candidate region based on the second information entropy and gray values; Determine the total importance degree based on the first importance degree and the second importance degree, and determine the candidate region with the lowest total importance degree as the position of the target digital anchor.

[0018] In another possible implementation manner, when the position mapping module determines the first importance degree of each candidate region based on the first information entropy, difference entropy, and similarity, it specifically is used for: Calculate the first average value of the information entropy of each region video and the average value of the difference entropy of each absolute difference map; Calculate the second average value of the similarity between adjacent frames of each region video and the similarity variance; Determine the first importance degree based on the first average value, difference entropy average value, second average value, similarity variance, and their respective corresponding coefficients.

[0019] In another possible implementation, when the position mapping module determines the second importance of each candidate area based on the second information entropy and the grayscale value, it specifically is used for: Dividing each target area screen of each area video into multiple sub - area screens; Determining the third average value of the grayscale values corresponding to both sides of the contour of each sub - area screen and determining the difference of the third average values; Summing the differences of each sub - area screen to obtain the total value of each target area screen with respect to the difference; Based on the total value of each sub - area screen, determining the fourth average value of each candidate area with respect to the total value, based on the second information entropy of each sub - area screen, determining the fifth average value of each candidate area with respect to the second information entropy, and determining the second importance of each candidate area based on the fourth average value, the fifth average value, and their respective corresponding coefficients.

[0020] In another possible implementation, each news field to which it belongs corresponds to multiple preset dressing combinations, and each preset dressing combination corresponds to multiple tags. When the dressing mapping module determines the dressing of the reference digital anchor based on the news field to which it belongs, it specifically is used for: Determining the fourth quantity of the first word combination, the second word combination, and the tags that feature hits each preset dressing combination; Determining the preset dressing combination with the largest fourth quantity as the dressing of the reference digital anchor.

[0021] In a third aspect, the present application provides an electronic device, adopting the following technical solution: An electronic device, the electronic device includes: At least one processor; A memory; At least one application program, where at least one application program is stored in the memory and is configured to be executed by at least one processor, and at least one configuration is for: executing the method for generating a digital anchor virtual image based on big data artificial intelligence shown in any possible implementation manner of the first aspect.

[0022] In a fourth aspect, the present application provides a computer - readable storage medium, adopting the following technical solution: A computer - readable storage medium, when the computer program is executed in a computer, causing the computer to execute the method for generating a digital anchor virtual image based on big data artificial intelligence described in any item of the first aspect.

[0023] In summary, the present application includes at least one of the following beneficial technical effects: Obtaining the broadcast text and broadcast video facilitates subsequent analysis. The broadcast text is segmented to obtain multiple first word combinations that make up the broadcast text. The broadcast video is subjected to text recognition and feature recognition to obtain second word combinations in the broadcast video. The first word combinations, second word combinations, and features in the broadcast video are all key factors reflecting the news field to which the news belongs. Therefore, according to the first word combinations, second word combinations, and features, the news field to which the news belongs can be accurately determined. After determining the required news field, the clothing on the benchmark digital anchor is determined, so that the image of the digital anchor is more in line with the news field. Then, the clothing is mapped onto the benchmark digital anchor to obtain the target digital anchor. Since the picture changes at different positions in the broadcast video are different, the position of the target digital anchor on the broadcast video is determined according to the broadcast video, that is, the position with the least information loss for the news broadcast, and the target digital anchor is mapped to the determined position, finally achieving the effect of reducing the information display loss caused by the occlusion of the digital anchor. Brief Description of the Drawings

[0024] Figure 1 It is a schematic flowchart of a method for generating a virtual image of a digital anchor based on big data artificial intelligence according to an embodiment of the present application.

[0025] Figure 2 It is a schematic structural diagram of a system for generating a virtual image of a digital anchor based on big data artificial intelligence according to an embodiment of the present application.

[0026] Figure 3 It is a schematic structural diagram of an electronic device according to an embodiment of the present application. Detailed Description of the Embodiment

[0027] The following further elaborates on the present application in conjunction with the accompanying drawings.

[0028] After reading this specification, those skilled in the art can make modifications to this embodiment without creative contributions as needed, but as long as they are within the scope of the claims of the present application, they are protected by the patent law.

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present application.

[0030] In addition, the term "and / or" in this text is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this text generally represents an "or" relationship between the associated objects before and after, unless otherwise specified.

[0031] The embodiments of the present application will be further described in detail below with reference to the accompanying drawings of the specification.

[0032] The embodiments of the present application provide a method for generating a digital anchor virtual image based on big data artificial intelligence, which is executed by an electronic device. The electronic device can be a server or a terminal device. Among them, the server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc., but is not limited thereto. The terminal device and the server can be directly or indirectly connected through wired or wireless communication methods, and the embodiments of the present application do not make limitations here. For example Figure 1 As shown, the method includes step S101, step S102, step S103, step S104, step S105 and step S106, where S101, obtain the broadcast text and broadcast video of the news.

[0033] For the embodiments of the present application, both the broadcast text and the broadcast video can be pre-edited and produced by staff and then input into the electronic device so that the electronic device obtains the broadcast text and the broadcast video.

[0034] S102, perform word segmentation on the broadcast text to obtain multiple first word combinations of the broadcast text.

[0035] For the embodiments of the present application, the electronic device performs word segmentation according to the hidden Markov model, conditional random field, deep learning model, etc. to obtain multiple first word combinations, or can also perform word segmentation on the broadcast text through plug-in tools such as jieba and HanLP to obtain multiple first word combinations.

[0036] S103, perform text recognition and feature recognition on the broadcast video to obtain the second word combination and features in the broadcast video.

[0037] For the embodiments of the present application, since there is audio in the broadcast video, the electronic device parses the broadcast video to obtain the audio file in the broadcast video, then inputs the audio file into a network model for audio-to-text conversion for text recognition to obtain the text information corresponding to the audio in the broadcast video, and then obtains the second word combination in the broadcast video by the method described in step S102. The electronic device inputs each frame of the broadcast video into the trained network model for feature recognition, so as to obtain the features appearing in the broadcast video.

[0038] S104. Determine the news field to which the news belongs based on the first word combination, the second word combination, and the features.

[0039] For the embodiments of the present application, the first word combination, the second word combination, and the features appearing in the broadcast video are all key factors for describing the news field required for the news. Therefore, the electronic device can accurately determine the news field required for the news by combining the first word combination, the second word combination, and the features. The news fields include international, sports, entertainment, automobiles, movies, etc.

[0040] S105. Determine the clothing of the reference digital anchor based on the news field to which it belongs and map the clothing onto the reference digital anchor to obtain the target digital anchor.

[0041] For the embodiments of the present application, in order to make the image of the digital anchor more conform to the required news field, the electronic device determines the clothing of the reference digital anchor according to the required news field. The reference digital anchor is only a human body model and the basic clothing attached to the human body model, and the basic clothing does not belong to any news field. After the electronic device determines the clothing of the reference digital anchor, mapping the clothing onto the reference digital anchor can obtain the target digital anchor, and the target digital anchor is more in line with the news field, thereby improving the viewing experience and effect.

[0042] S106. Determine the position of the target digital anchor based on the broadcast video and map the target digital anchor to the position in the broadcast video.

[0043] For the embodiments of the present application, since the amount of information and the picture changes at each place in the broadcast video are different, the electronic device determines the position of the target digital anchor in the broadcast video according to the broadcast video. This position is the position where the overall broadcast video information display is least missing and the broadcast impact is least when the target digital anchor blocks the broadcast video. Then mapping the target digital anchor to this position can reduce the information display loss caused by the occlusion of the digital anchor.

[0044] In a possible implementation manner of the embodiments of the present application, each news field corresponds to a preset word library and a preset feature library. Determining the news field to which the news belongs based on the first word combination, the second word combination, and the feature in step S104 specifically includes step S1041 (not shown in the figure), step S1042 (not shown in the figure), step S1043 (not shown in the figure), step S1044 (not shown in the figure), step S1045 (not shown in the figure), where S1041, determine the first quantity of multiple first word combinations and the second word combination in each preset word library.

[0045] For the embodiments of the present application, both the first word combination and the second word combination are descriptions of the news in terms of text. The words in the preset word library corresponding to each news field are classified by staff and used to describe the news field. Therefore, the electronic device fuses the first word combination and the second word combination, and then matches and counts all the word combinations with the preset word library of each news field respectively, so as to determine the first quantity of all the word combinations in each preset word library. The more the first quantity corresponding to a certain preset word library, the greater the possibility that the news belongs to the news field of the preset word library.

[0046] S1042, determine the second quantity of multiple features in each preset feature library.

[0047] For the embodiments of the present application, the features in the preset feature library corresponding to each news field are also classified by staff and used to describe the news field. The electronic device can determine the second quantity of multiple features in each preset feature library in the manner described in step S1041. The more the second quantity corresponding to a certain preset feature library, the greater the possibility that the news belongs to the news field of the preset feature library.

[0048] S1043, determine multiple target data sets.

[0049] Wherein, each target data set is each second word combination and the features in the picture of the broadcast video at the time point when each second word combination appears.

[0050] For the embodiments of the present application, the electronic device determines the time point when the second word combination appears in the broadcast video and the corresponding picture at that time point. The features in the picture at the same time point have a strong correlation with the second word combination at that time point. Therefore, the electronic device determines such multiple target data sets.

[0051] S1044, determine the news field hit by each target data set and the third quantity of the number of times each news field is hit, and determine the preferred field of each target data set as the news field with the largest third quantity.

[0052] For the embodiments of the present application, if the second word combinations and the news fields hit by the features in a certain target data set are the same, that is, the number of news fields hit by the target data set is 1, it indicates that the target data set of this picture is more likely to reflect that the news belongs to the hit news field. If the number of hit news fields is greater than 1, it indicates that the target data set of this picture may reflect that the news belongs to different news fields. Therefore, determine the number of times each news field hit by the second word combination and the feature in the target data set, that is, the third quantity. The news field with the largest third quantity is the news field to which the target data set of this picture is most likely to reflect that the news belongs, that is, the preferred field. If there are at least two news fields with the same third quantity among the multiple hit news fields, the electronic device determines the news fields with the same third quantity as the preferred field.

[0053] S1045, determine the news field to which the news belongs based on the first quantity, the second quantity, and the preferred field.

[0054] For the embodiments of the present application, in summary, the first quantity, the second quantity, and the preferred field are all key factors characterizing the news field to which the news truly belongs. Therefore, the electronic device can accurately determine the required news field of the news by combining the first quantity, the second quantity, and the preferred field.

[0055] A possible implementation manner of the embodiments of the present application is that in step S1045, determining the news field to which the news belongs based on the first quantity, the second quantity, and the preferred field specifically includes step S1 (not shown in the figure), step S2 (not shown in the figure), step S3 (not shown in the figure), and step S4 (not shown in the figure), where S1, sort the first quantity to obtain the target preset word library with the largest first quantity.

[0056] For the embodiments of the present application, the electronic device sorts the first quantity from largest to smallest to obtain the target preset word library that is most matched in terms of the first word combination and the second word combination. The number of the first word combination and the second word combination hit in the target preset word library is the largest, indicating that the news field corresponding to the target preset word library of the first word combination and the second word combination is most in line with the true belonging field of this news.

[0057] S2, sort the second quantity to obtain the target preset feature library with the largest second quantity.

[0058] For the embodiments of the present application, the electronic device sorts the second quantity from largest to smallest to obtain the target preset feature library that is most matched in terms of features. The number of features hit in the target preset feature library is the largest, indicating that the news field corresponding to the target preset word library of the features in the broadcast video is most in line with the true belonging field of this news.

[0059] S3. Induce the preference fields of all target data sets to obtain the target preference field with the most occurrences.

[0060] For the embodiments of the present application, the electronic device induces and counts the preference fields of all target data sets to determine the number of occurrences of each preference field, and then sorts the number of occurrences to obtain the target preference field with the most occurrences. The target preference field is the true field to which the news most conforms in terms of the target data set (strong correlation between the second word combination and the feature within the same moment screen).

[0061] S4. If there are at least two identical news fields among the news field of the target preset word library, the news field of the target preset feature library, and the target preference field, determine the at least two identical fields as the news field to which the news belongs.

[0062] For the embodiments of the present application, if there are at least two identical fields, it indicates that a certain news field accounts for the majority of cases. Therefore, the news field to which the news belongs is determined as the news field with the most identical fields. If the above three fields are all different, determine the news field corresponding to the highest priority among the above target preset word library, target preset feature library, and preference field as the news field to which the news belongs. Among them, the priorities of the above three can be set by the staff according to requirements or actual situations. The news field determined through the above operations using the first word combination, the second word combination, and the feature is more accurate.

[0063] A possible implementation manner of the embodiments of the present application. Determining the position of the target digital anchor based on the broadcast video in step S106 includes step S1061 (not shown in the figure), step S1062 (not shown in the figure), step S1063 (not shown in the figure), step S1064 (not shown in the figure), step S1065 (not shown in the figure), step S1066 (not shown in the figure), step S1067 (not shown in the figure), step S1068 (not shown in the figure), step S1069 (not shown in the figure), and step S10610 (not shown in the figure), where S1061. Perform edge detection on the target digital anchor to obtain the contour of the target digital anchor.

[0064] For the embodiments of the present application, the electronic device acquires the image of the target digital anchor, and then uses an edge detection algorithm, such as Canny edge detection, to determine the edge by finding the points with drastic changes in gray intensity in the image.

[0065] S1062. Map the contour to the starting position in the lower left corner of the broadcast video, and horizontally translate the contour at a preset step size to obtain multiple candidate regions.

[0066] The area to be selected includes the area where the contour is located when it is at the starting position.

[0067] For the embodiments of the present application, the digital anchor is usually placed in the lower area of ​​the broadcast video, so the electronic device maps the outline of the digital anchor to the starting position of the lower left corner of the broadcast video. The preset step size can be a specified number of pixels or distance. Then the electronic device controls the outline to move horizontally to the right in the broadcast video according to the preset step size, thereby determining multiple candidate areas of the outline of the digital anchor on the broadcast video.

[0068] S1063, segmenting the broadcast video according to each area to be selected to obtain a regional video of each area to be selected.

[0069] According to the embodiment of the present application, after the electronic device determines the areas to be selected, the broadcast video is segmented according to the areas to be selected, thereby obtaining regional videos of each area to be selected.

[0070] S1064, determining each frame of the video in each region, and calculating the first information entropy of each frame and the similarity between adjacent frames.

[0071] For the embodiment of the present application, the electronic device can read the video frame by frame through the library function to determine each frame of the video in each area. Then the electronic device uses the information entropy calculation formula to calculate the first information entropy of each frame. Specifically, the electronic device traverses all pixels of each frame, counts the number of times each pixel value appears, and calculates its probability. According to the probability of each pixel value, the information entropy is calculated using the formula H(X)=-sum(p(x)×log2(p(x))), where: H(X) is the information entropy; p(x) is the probability of the pixel value. The electronic device calculates the information entropy of each color channel of each frame separately, and then adds them together to obtain the first information entropy of each frame. The larger the first information entropy in the picture, the more information and uncertainty it contains, which means that the larger the amount of information, the more serious the information loss is when the digital anchor blocks the area.

[0072] The electronic device can input adjacent frames of the regional video into the trained network model for similarity calculation, represent the adjacent frames in vector form, and then calculate the cosine distance between the vectors to determine the similarity between adjacent frames. The higher the similarity, the less drastic the picture changes in the regional video, the more stable the information changes, and the less information is lost when the digital anchor blocks the area.

[0073] S1065, drawing a target area with the outline as the center and according to a preset width, and dividing each frame of the broadcast video according to the target area to obtain a target area image of each to-be-selected area.

[0074] For the embodiments of the present application, the preset width may be a specified number of pixels. Taking the center line of a contour of an electronic device as the reference, it extends on both sides of the contour according to the number of pixels of the preset width, so that the number of pixels on both sides of the contour is the same, that is, the widths on both sides of the contour are the same. The electronic device segments the broadcast video in this way to obtain the target area, and then obtains each frame of the target area, that is, the target area picture.

[0075] S1066. Determine the pixel difference at the same position between two adjacent frames of the picture to obtain an absolute difference map, and calculate the differential entropy of each absolute difference map.

[0076] For the embodiments of the present application, the electronic device determines adjacent pictures of the regional video and calculates the difference between the pixel values at the same position of the adjacent pictures to obtain the pixel difference, thereby obtaining the absolute difference map. A new differential image, that is, the absolute difference map, is generated by calculating the absolute difference between the corresponding pixels of the two images. In the differential image, the area with a higher pixel value indicates a greater difference between the original images. Then, the electronic device can use the information entropy calculation formula to determine the differential entropy of each absolute difference map. The greater the differential entropy, the greater the change between adjacent pictures, and thus the more unstable the information change, and the more serious the missing situation of information display caused by the occlusion of the digital anchor.

[0077] S1067. Determine the first importance of each candidate area based on the first information entropy, differential entropy, and similarity.

[0078] For the embodiments of the present application, in summary, the first information entropy, differential entropy, and similarity are all key factors characterizing the amount of information and the degree of information change in each regional video. Therefore, the electronic device comprehensively determines the more accurate first importance of each candidate area according to these three factors, namely the first information entropy, etc.

[0079] S1068. Calculate the second information entropy of each target area picture of each candidate area and determine the gray values of the pixels on both sides of the contour.

[0080] For the embodiments of the present application, the second information entropy of the target area picture is also calculated using the information entropy calculation formula. The greater the second information entropy, the greater the amount of information in the broadcast video at the edge of the digital anchor, and the more important the information. The electronic device performs gray-scale transformation on each target area picture to obtain the gray values of the pixels on both sides of the contour. The greater the difference in gray values, the richer the content of the target area picture, and thus the greater the change of information at the contour.

[0081] S1069. Determine the second importance of each candidate area based on the second information entropy and the gray value.

[0082] For the embodiment of the present application, the second information entropy characterizes the amount of information of the video picture at the contour and the uncertainty randomness of the pixels, and the grayscale value characterizes the richness of the picture at the contour and the change of the pixels. Therefore, the second information entropy and the grayscale value are both key factors affecting the amount of video information at the contour and whether the information is important. Therefore, the electronic device comprehensively determines the second importance of the video in each area based on the second information entropy and the grayscale value.

[0083] S10610: Determine a total importance based on the first importance and the second importance, and determine the candidate area with the lowest total importance as the location of the target digital anchor.

[0084] For the embodiment of the present application, the first importance represents the overall information change and information volume of the regional video (selected area), and the second importance represents the information change and information volume of the video at the contour. Therefore, the electronic device can determine the total importance of each selected area based on the first importance and the second importance. Specifically, the electronic device can sum the first importance and the second importance to obtain the total importance. The higher the total importance, the more important the information, the more drastic the information change, and the larger the amount of information. Therefore, the electronic device can determine the selected area with the lowest total importance as the location of the target digital anchor. By determining the first importance and the second importance, it is more accurate to finally filter out the location of the target digital anchor from the selected area.

[0085] In a possible implementation of the embodiment of the present application, in step S1067, the first importance of each candidate region is determined based on the first information entropy, the difference entropy and the similarity, specifically including step Sa (not shown in the figure), step Sb (not shown in the figure) and step Sc (not shown in the figure), wherein: Sa, calculate the first average of the information entropy of each region video and the average of the difference entropy of each absolute difference map.

[0086] For the embodiment of the present application, the information entropy of each region includes the information entropy of each frame, so the electronic device determines the first average value through the average value calculation formula. The larger the first average value, the greater the amount of information in the regional video as a whole, the greater the randomness of the pixels, and the more important the regional video. The average value of the difference entropy of the absolute difference graph is determined through the average value calculation formula. The larger the average value of the difference entropy, the greater the pixel change in the regional video, the more important the information, and the more important the regional video.

[0087] Sb, calculates the second average value and similarity variance of the similarities of adjacent frames of each region video.

[0088] For the embodiment of the present application, the electronic device calculates the second average value of the similarity by the average value calculation formula, and determines the variance of the similarity by using the variance calculation formula. The larger the second average value, the closer the adjacent pictures are, the smaller the degree of change of the pictures over time, the smaller the information change, and the less important the regional video is. The larger the variance, the more drastic the picture change, the greater the information change, and the more important the regional video is.

[0089] Sc, determining the first importance based on the first average value, the difference entropy average value, the second average value, the similarity variance, and the respective corresponding coefficients.

[0090] For the embodiment of the present application, in summary, the first average value, the difference entropy average value, the second average value and the similarity variance are all key factors affecting the importance of the selected area, so the staff sets the corresponding coefficients for the above four factors and stores them in the electronic device, and the electronic device calls the corresponding coefficients and performs weighted calculation on the above four factors to obtain a score, which represents the first importance. The first importance obtained by comprehensive calculation and analysis of these four factors is more accurate.

[0091] In a possible implementation of the embodiment of the present application, in step S1069, the second importance of each candidate region is determined based on the second information entropy and the gray value, which specifically includes step 1, step 2, step 3 and step 4, wherein: Step 1: divide each target area screen of each area video into multiple sub-area screens.

[0092] According to the embodiment of the present application, the electronic device may divide the target area screen into a plurality of sub-area screens along the outline according to a preset length.

[0093] Step 2: determine the third average values ​​of the grayscale values ​​corresponding to both sides of the contour of each sub-region image and determine the difference between the third average values.

[0094] For the embodiment of the present application, the electronic device determines the third average value of the grayscale values ​​corresponding to each sub-area image on both sides of the contour through an average value calculation formula, and then subtracts the two third average values ​​to obtain a difference. The larger the difference, the greater the pixel change on both sides of the contour, the greater the amount of information, and the more important the sub-area image.

[0095] Step three, summing up the difference values ​​of each sub-region picture to obtain the total value of the difference values ​​of each target region picture.

[0096] For the embodiment of the present application, the electronic device sums the difference of each sub-area picture to obtain a total value representing the difference of the entire target area picture, and the total value represents the information change of the target area picture as a whole on both sides of the outline.

[0097] Step 4: Determine the fourth average value of each candidate region with respect to the total value based on the total value of each sub-region screen, determine the fifth average value of each candidate region with respect to the second information entropy based on the second information entropy of each sub-region screen, and determine the second importance degree of each candidate region based on the fourth average value, the fifth average value, and their respective corresponding coefficients.

[0098] For the embodiments of the present application, the electronic device calculates the average value of all the second information entropies, that is, the fifth average value, and uses this average value to characterize the amount of information in the video at the contour of each candidate region and the uncertainty of the pixels. In summary, both the fourth average value and the fifth average value are key factors affecting whether the video at the contour of each candidate region is important. The staff sets their respective coefficients for the fourth average value and the fifth average value and stores them in the electronic device. Then, the electronic device calls their respective corresponding coefficients to perform weighted calculation on the fourth average value and the fifth average value to obtain a score, and uses this score to characterize the second importance degree. The higher the second importance degree, the more important the information at the contour, and the less suitable it is as the position of the target digital anchor. Determining the second importance degree by comprehensively determining the fourth average value and the fifth average value is more accurate.

[0099] In a possible implementation manner of the embodiments of the present application, each news field corresponds to multiple preset dressing combinations, and each preset dressing combination corresponds to multiple labels. Determining the dressing of the reference digital anchor based on the news field in step S105 specifically includes step five and step six, where Step 5: Determine the first word combination, the second word combination, and the fourth quantity of the labels that the features hit each preset dressing combination.

[0100] Step 6: Determine the preset dressing combination with the largest fourth quantity as the dressing of the reference digital anchor.

[0101] For the embodiments of the present application, each news field corresponds to multiple preset dressing combinations, and there are differences in the labels between the preset dressing combinations. Therefore, the electronic device determines the first word combination, the second word combination, and the fourth quantity of the labels that the features hit each preset dressing combination. The larger the fourth quantity of a certain preset dressing combination, the more the preset dressing combination fits the current news, and the higher the degree of fit with the news. Therefore, the electronic device can determine the preset dressing combination with the largest fourth quantity as the dressing of the reference digital anchor. Determining the dressing of the reference digital anchor in the above manner is more accurate and improves the degree of fit between the digital anchor and the news.

[0102] The above embodiments introduce the method for generating the virtual image of the digital anchor based on big data artificial intelligence from the perspective of the method flow. The following embodiments introduce the system for generating the virtual image of the digital anchor based on big data artificial intelligence from the perspective of virtual modules or virtual units. For details, see the following embodiments.

[0103] The embodiment of the present application provides a digital anchor virtual image generation system 20 based on big data artificial intelligence, as Figure 2 shown. The digital anchor virtual image generation system 20 based on big data artificial intelligence may specifically include: An acquisition module 201, configured to acquire the broadcast text and broadcast video of the news; A first processing module 202, configured to perform word segmentation on the broadcast text to obtain a plurality of first word combinations of the broadcast text; A second processing module 203, configured to perform text recognition and feature recognition on the broadcast video to obtain a second word combination and features in the broadcast video; A field determination module 204, configured to determine the news field to which the news belongs based on the first word combination, the second word combination, and the features; A dressing mapping module 205, configured to determine the dressing of the reference digital anchor based on the news field and map the dressing to the reference digital anchor to obtain a target digital anchor; A position mapping module 206, configured to determine the position of the target digital anchor based on the broadcast video and map the target digital anchor to the position in the broadcast video.

[0104] The embodiment of the present application discloses a digital anchor virtual image generation system 20 based on big data artificial intelligence. Among them, the acquisition module 201 acquires the broadcast text and broadcast video for subsequent analysis. The first processing module 202 performs word segmentation on the broadcast text to obtain a plurality of first word combinations that make up the broadcast text. The second processing module 203 performs text recognition and feature recognition on the broadcast video to obtain a second word combination in the broadcast video. The first word combination, the second word combination, and the features in the broadcast video are all key factors reflecting the news field to which the news belongs. Therefore, the field determination module 204 can accurately determine the news field to which the news belongs based on the first word combination, the second word combination, and the features. After determining the required news field, the dressing mapping module 205 determines the dressing on the reference digital anchor, so that the image of the digital anchor is more in line with the news field, and then maps the dressing to the reference digital anchor to obtain a target digital anchor. Since the picture changes at each position in the broadcast video are different, the position mapping module 206 determines the position of the target digital anchor on the broadcast video according to the broadcast video, that is, the position with the least information loss caused by the news broadcast, and maps the target digital anchor to the determined position, finally achieving the effect of reducing the information display loss caused by the occlusion of the digital anchor.

[0105] In a possible implementation manner of the embodiment of the present application, each news field corresponds to a preset word library and a preset feature library. When the field determination module 204 determines the news field to which the news belongs based on the word combination and the features, it is specifically used for: Determine the first quantity of multiple first word combinations and second word combinations in each preset word library; Determine the second quantity of multiple features in each preset feature library; Determine multiple target data sets, where each target data set is each second word combination and the features in the picture of the broadcast video at the time point when each second word combination appears; Determine the news field hit by each target data set and the third quantity of the number of times each hit news field appears, and determine the preferred field of each target data set as the news field with the most third quantity; Determine the news field to which the news belongs based on the first quantity, the second quantity, and the preferred field.

[0106] In a possible implementation manner of the embodiment of the present application, when the field determination module 204 determines the news field to which the news belongs based on the first quantity, the second quantity, and the preferred field, it is specifically used for: Sort the first quantity to obtain the target preset word library with the most first quantity; Sort the second quantity to obtain the target preset feature library with the most second quantity; Summarize the preferred fields of all target data sets to obtain the target preferred field with the most occurrences; If there are at least two identical news fields in the news field of the target preset word library, the news field of the target preset feature library, and the target preferred field, determine the at least two identical fields as the news field to which the news belongs.

[0107] In a possible implementation manner of the embodiment of the present application, when the position mapping module 206 determines the position of the target digital anchor based on the broadcast video, it is specifically used for: Perform edge detection on the target digital anchor to obtain the contour of the target digital anchor; Map the contour to the starting position in the lower left corner of the broadcast video, and horizontally translate the contour at a preset step length to obtain multiple candidate regions, where the candidate regions include the region where the contour is located when it is at the starting position; Segment the broadcast video according to each candidate region to obtain the regional video of each candidate region; Determine each frame of the picture of each regional video, calculate the first information entropy of each frame of the picture and the similarity between adjacent pictures; Draw a target region centered on the contour and with a preset width, and segment each frame of the picture of the broadcast video according to the target region to obtain the target region picture of each candidate region; Determine the pixel difference at the same position between two adjacent frames of the picture to obtain an absolute difference map, and calculate the differential entropy of each absolute difference map; Determine the first importance of each candidate region based on the first information entropy, differential entropy, and similarity; Calculate the second information entropy of each target region screen of each region video and determine the gray values of the pixels on both sides of the contour; Determine the second importance of each candidate region based on the second information entropy and gray values; Determine the total importance based on the first importance and the second importance, and determine the position of the target digital anchor as the candidate region with the lowest total importance.

[0108] In a possible implementation manner of the embodiment of the present application, when the position mapping module 206 determines the first importance of each candidate region based on the first information entropy, differential entropy, and similarity, it is specifically used for: Calculate the first average value of the information entropy of each region video and the average value of the differential entropy of each absolute difference map; Calculate the second average value of the similarity of adjacent screens of each region video and the similarity variance; Determine the first importance based on the first average value, the average value of the differential entropy, the second average value, the similarity variance, and their respective corresponding coefficients.

[0109] In a possible implementation manner of the embodiment of the present application, when the position mapping module 206 determines the second importance of each candidate region based on the second information entropy and gray values, it is specifically used for: Divide each target region screen of each region video into multiple sub-region screens; Determine the third average value of the gray values corresponding to both sides of the contour of each sub-region screen and determine the difference of the third average value; Sum the differences of each sub-region screen to obtain the total value of each target region screen with respect to the difference; Determine the fourth average value of each candidate region with respect to the total value based on the total value of each sub-region screen, determine the fifth average value of each candidate region with respect to the second information entropy based on the second information entropy of each sub-region screen, and determine the second importance of each candidate region based on the fourth average value, the fifth average value, and their respective corresponding coefficients.

[0110] In a possible implementation manner of the embodiment of the present application, each news field belongs to multiple preset dressing combinations, and each preset dressing combination corresponds to multiple labels. When the dressing mapping module 205 determines the dressing of the reference digital anchor based on the news field to which it belongs, it is specifically used for: Determine the first word combination, the second word combination, and the fourth quantity of the labels that feature hits each preset dressing combination; Determine the preset dressing combination with the largest fourth quantity as the dressing of the reference digital anchor.

[0111] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the digital anchor virtual image generation system 20 based on big data artificial intelligence described above can refer to the corresponding process in the foregoing method embodiments, and will not be elaborated herein.

[0112] An embodiment of the present application provides an electronic device, such as Figure 3 shown, Figure 3 The electronic device 30 shown includes: a processor 301 and a memory 303. Among them, the processor 301 and the memory 303 are connected, such as connected through a bus 302. Optionally, the electronic device 30 may further include a transceiver 304. It should be noted that in actual applications, the transceiver 304 is not limited to one, and the structure of the electronic device 30 does not constitute a limitation to the embodiments of the present application.

[0113] The processor 301 may be a CPU (Central Processing Unit, central processor), a general-purpose processor, a DSP (Digital Signal Processor, data signal processor), an ASIC (Application Specific Integrated Circuit, application-specific integrated circuit), an FPGA (Field Programmable Gate Array, field programmable gate array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logical blocks, modules and circuits described in combination with the disclosure of the present application. The processor 301 may also be a combination that implements computing functions, such as a combination including one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0114] The bus 302 may include a path for transmitting information between the above components. The bus 302 may be a PCI (Peripheral Component Interconnect, peripheral component interconnect standard) bus or an EISA (Extended Industry Standard Architecture, extended industry standard architecture) bus, etc. The bus 302 may be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 3 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0115] The memory 303 can be a ROM (Read Only Memory), or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory), or other types of dynamic storage devices that can store information and instructions. It can also be an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0116] The memory 303 is used to store the application program code for executing the solution of this application, and is controlled by the processor 301 for execution. The processor 301 is used to execute the application program code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0117] Among them, the electronic device includes but is not limited to: mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. It can also be a server, etc. Figure 3 The electronic device shown is only an example and should not impose any limitations on the functions and usage scope of the embodiments of this application.

[0118] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When it runs on a computer, it enables the computer to execute the corresponding content in the foregoing method embodiment. Compared with the related art, in the embodiment of the present application, obtaining the broadcast text and the broadcast video facilitates subsequent analysis. The broadcast text is segmented to obtain a plurality of first word combinations that make up the broadcast text. The broadcast video is subjected to text recognition and feature recognition to obtain second word combinations in the broadcast video. The first word combinations, the second word combinations, and the features in the broadcast video are all key factors reflecting the news field to which the news belongs. Therefore, according to the first word combinations, the second word combinations, and the features, the news field to which the news belongs can be accurately determined. After determining the required news field, the clothing on the reference digital anchor is determined, so that the image of the digital anchor is more in line with the news field. Then, the clothing is mapped onto the reference digital anchor to obtain the target digital anchor. Since the changes in the images at various positions in the broadcast video are different, the position of the target digital anchor on the broadcast video is determined according to the broadcast video, that is, the position that causes the least information loss in the news broadcast, and the target digital anchor is mapped to the determined position, finally achieving the effect of reducing the information display loss caused by the occlusion of the digital anchor.

[0119] It should be understood that although the steps in the flowchart of the accompanying drawings are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps is not strictly limited in order, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

[0120] The above are only some embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A method for generating a digital anchor virtual image based on big data artificial intelligence, characterized in that: include: Get the text and video of news reports; Performing word segmentation processing on the broadcast text to obtain a plurality of first word combinations of the broadcast text; Performing text recognition and feature recognition on the broadcast video to obtain a second word combination and feature in the broadcast video; Determine the news field to which the news belongs based on the first word combination, the second word combination and the feature; Determine the attire of a benchmark digital anchor based on the news field to which it belongs and map the attire onto the benchmark digital anchor to obtain a target digital anchor; The position of the target digital anchor is determined based on the broadcast video, and the target digital anchor is mapped to the position in the broadcast video.

2. The method for generating a digital anchor virtual image based on big data artificial intelligence according to claim 1 is characterized in that: Each news field corresponds to a preset word library and a preset feature library, and the news field to which the news belongs is determined based on the word combination and the feature, including: Determine a first quantity of the plurality of first word combinations and second word combinations in each preset word library; Determining a second quantity of the plurality of features in each preset feature library; Determine a plurality of target data sets, each target data set being a feature of each second word combination and a screen in the broadcast video at a time point when each second word combination appears; Determine the third number of news fields hit by each target data set and the number of times each news field is hit, and determine the news field with the largest third number as the preferred field of each target data set; The news field to which the news belongs is determined based on the first number, the second number and the preferred field.

3. The method for generating a digital anchor virtual image based on big data artificial intelligence according to claim 2 is characterized in that: The determining the news field to which the news belongs based on the first quantity, the second quantity and the preferred field includes: Sorting the first quantities to obtain a target preset word library with the largest first quantity; Sorting the second quantities to obtain a target preset feature library with the largest second quantity; Summarize the preference areas of all target data sets to obtain the target preference areas with the most occurrences; If at least two of the news fields of the target preset word library, the news fields of the target preset feature library, and the target preference fields are the same, the at least two same fields are determined to be the news fields to which the news belongs.

4. The method for generating a digital anchor virtual image based on big data artificial intelligence according to claim 3 is characterized in that: The determining the position of the target digital anchor based on the broadcast video includes: Performing edge detection on the target digital anchor to obtain a contour of the target digital anchor; Mapping the outline to the starting position of the lower left corner of the broadcast video, and translating the outline horizontally according to a preset step length to obtain a plurality of candidate areas, wherein the candidate areas include the area where the outline is located when it is at the starting position; Segment the broadcast video according to each to-be-selected area to obtain a regional video of each to-be-selected area; Determine each frame of the video in each region, calculate the first information entropy of each frame and the similarity between adjacent frames; Draw a target area with the outline as the center and according to a preset width, and divide each frame of the broadcast video according to the target area to obtain a target area image of each to-be-selected area; Determine the pixel difference at the same position between two adjacent frames, obtain an absolute difference map, and calculate the difference entropy of each absolute difference map; Determine a first importance of each candidate region based on the first information entropy, difference entropy and similarity; Calculate the second information entropy of each target area picture of each area video and determine the grayscale values ​​of pixels on both sides of the outline; Determine a second importance of each candidate area based on the second information entropy and the gray value; The total importance is determined based on the first importance and the second importance, and the candidate area with the lowest total importance is determined as the location of the target digital anchor.

5. The method for generating a digital anchor virtual image based on big data artificial intelligence according to claim 4 is characterized in that: The determining the first importance of each candidate region based on the first information entropy, the difference entropy and the similarity includes: Calculate the first average value of the information entropy of each region video and the average value of the difference entropy of each absolute difference map; Calculate the second average value and similarity variance of the similarities of adjacent frames of each regional video; The first importance is determined based on the first average value, the difference entropy average value, the second average value, the similarity variance, and the respective corresponding coefficients.

6. The method for generating a digital anchor virtual image based on big data artificial intelligence according to claim 5 is characterized in that: The determining the second importance of each candidate region based on the second information entropy and the grayscale value includes: Separate each target area screen of each area video into a plurality of sub-area screens; Determine the third average value of the grayscale values ​​corresponding to both sides of the contour of each sub-region picture and determine the difference of the third average values; Summing the difference values ​​of each sub-region picture to obtain a total value of each target region picture with respect to the difference values; Based on the sum value of each sub-region picture, the fourth average value of each candidate area with respect to the sum value is determined; based on the second information entropy of each sub-region picture, the fifth average value of each candidate area with respect to the second information entropy is determined; and based on the fourth average value, the fifth average value and their respective corresponding coefficients, the second importance of each candidate area is determined.

7. The method for generating a digital anchor virtual image based on big data artificial intelligence according to claim 1 is characterized in that: Each news field corresponds to a plurality of preset clothing combinations, and each preset clothing combination corresponds to a plurality of tags. The method of determining the clothing of the benchmark digital anchor based on the news field includes: Determine a fourth number of the first word combination, the second word combination, and the label of each preset clothing combination that the feature hits; The fourth largest number of preset clothing combinations is determined as the clothing of the benchmark digital anchor.

8. A digital anchor virtual image generation system based on big data artificial intelligence, characterized in that: include: The acquisition module is used to obtain the text and video of the news report; A first processing module, configured to perform word segmentation processing on the broadcast text to obtain a plurality of first word combinations of the broadcast text; A second processing module is used to perform text recognition and feature recognition on the broadcast video to obtain a second word combination and feature in the broadcast video; A field determination module, used for determining the news field to which the news belongs based on the first word combination, the second word combination and the feature; A clothing mapping module, used to determine the clothing of a benchmark digital anchor based on the news field and map the clothing to the benchmark digital anchor to obtain a target digital anchor; The position mapping module is used to determine the position of the target digital anchor based on the broadcast video, and map the target digital anchor to the position in the broadcast video.

9. An electronic device, characterized in that: It includes: at least one processor; Memory; At least one application, wherein the at least one application is stored in the memory and is configured to be executed by the at least one processor, and the at least one application is used to execute the method for generating a digital anchor virtual image based on big data artificial intelligence according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed in a computer, the computer is caused to execute the method for generating a digital anchor virtual image based on big data artificial intelligence as described in any one of claims 1 to 7.

Images