A video retrieval and analysis system based on big data

Abstract

This invention discloses a video retrieval and analysis system based on big data, belonging to the field of video retrieval technology. It includes a tagging module, a matching module, a retrieval module, and a server. The tagging module is used to tag video sources with corresponding labels. The matching module is used to establish a matching graph for video sources. The retrieval module is used to perform corresponding video retrieval based on user input data, obtain the user-input retrieval data, extract key data from the retrieval data, establish a matching item table, divide the key data into several demand item data, calculate the matching value between each demand item data and each matching item in the matching item table, and combine the obtained matching values ​​into a corresponding target point set. The target point set is input into the matching graph, the video sources corresponding to the matching points within the summarization area are identified, the obtained video sources are integrated into target retrieval sources, the retrieval data is input into the target retrieval sources for video retrieval, several retrieval videos are obtained, and the obtained retrieval videos are recommended to the user.

Description

Technical Field

[0001] This invention belongs to the field of video retrieval technology, specifically a video retrieval and analysis system based on big data. Background Technology

[0002] With the development of multimedia technology, many multimedia web pages, multimedia applications, and clients have emerged that can provide users with video search functions. However, the increase in video sources has also brought some difficulties to users' video search. Therefore, how to determine the appropriate video source from among the many video sources is a problem that needs to be solved. Summary of the Invention

[0003] To address the problems of the above solutions, this invention provides a video retrieval and analysis system based on big data.

[0004] The objective of this invention can be achieved through the following technical solutions:

[0005] A video retrieval and analysis system based on big data includes a tag module, a matching module, a retrieval module, and a server;

[0006] The tagging module is used to assign corresponding tags to video sources; the matching module is used to create a matching graph of video sources.

[0007] The retrieval module is used to perform corresponding video retrieval based on user input data. It acquires the user-input retrieval data, extracts key data from the retrieval data, establishes a matching item table, divides the key data into several demand item data, calculates the matching value between each demand item data and each matching item in the matching item table, and combines the obtained matching values ​​into a corresponding target point set. The target point set is then input into a matching graph to identify the video sources corresponding to the matching points within the summarization area. The obtained video sources are integrated into a target retrieval source, and the retrieval data is input into the target retrieval source for video retrieval, resulting in several retrieved videos. The retrieved videos are then recommended to the user.

[0008] Furthermore, the working method of the tag module includes:

[0009] Obtain the available video sources, generate a corresponding video source detail table based on the obtained video sources, obtain the feature data of each video source based on the video source detail table, and set the source tag of the corresponding video source based on the obtained feature data.

[0010] Furthermore, the method for setting source labels for corresponding video sources based on the obtained feature data includes:

[0011] Establish a tag library, identify the feature data corresponding to each feature item in the feature data, input the feature data into the tag library for matching, obtain the corresponding single tag, and combine the obtained single tags to obtain the corresponding source tag.

[0012] Furthermore, methods for building a tag library include:

[0013] Identify the features, obtain the feature data based on the identified features, set a corresponding single label for each feature data, establish a first database, add the corresponding single label to the feature data, input the feature data into the first database for storage, and mark the current first database as a label library.

[0014] Furthermore, the matching module's working method includes:

[0015] Obtain source tags for each video source, convert the obtained source tags into corresponding matching points, establish a coordinate system corresponding to the matching points, input the obtained matching points into the coordinate system, and generate a coordinate graph based on the current coordinate system; set a summarization area in the coordinate graph, which is used to perform matching and summarization of corresponding matching points based on the target points input into the coordinate graph, and mark the current coordinate graph as a matching graph.

[0016] Furthermore, methods for converting the obtained source tags into corresponding matching points include:

[0017] Identify individual tags in the tag library, assign corresponding values ​​to each individual tag, and build a corresponding assignment matching table based on the assigned values;

[0018] Identify each individual label within the source label, match the corresponding assignment from the assignment matching table based on the identified individual labels, and integrate the obtained assignments into matching points according to the positions of each individual label within the source label.

[0019] Furthermore, the working methods of the inductive area include:

[0020] Identify the position of each target point in the coordinate graph, generate a corresponding range circle with the target point as the center, correct the generated range circle to obtain the range area, identify the matching points included in the range area, and include the identified matching points in the summarization area.

[0021] Compared with the prior art, the beneficial effects of the present invention are: through the cooperation between the tag module, the matching module and the retrieval module, when a user performs a video search, the target retrieval source can be quickly determined from the current large number of video sources based on the user's input data, and the search can be performed in the target retrieval source, which greatly reduces the search volume and increases the accuracy and security of the search, and avoids the problem of the retrieved video data carrying viruses when searching directly on the Internet. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a block diagram illustrating the principle of the present invention. Detailed Implementation

[0024] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0025] like Figure 1 As shown, a video retrieval and analysis system based on big data includes a tag module, a matching module, a retrieval module, and a server;

[0026] The tagging module is used to tag video sources with corresponding labels, and the specific methods include:

[0027] The process involves acquiring video sources, which refer to websites, platforms, etc., that contain video resources; generating a detailed video source table based on the acquired video sources; obtaining feature data for each video source based on the detailed video source table; and setting source tags for the corresponding video sources based on the obtained feature data.

[0028] The method for obtaining feature data of each video source based on the video source details table is as follows: a corresponding feature analysis model is established based on a CNN network or a DNN network to obtain the corresponding feature data from the Internet according to the video source name. Corresponding feature items are set, and analysis is performed based on the set feature items, such as feature data such as advertisements, memberships, and video types. Specifically, the corresponding training set is set manually for training, and the feature analysis model is analyzed after successful training to obtain the corresponding feature data.

[0029] Methods for setting source tags for corresponding video sources based on the obtained feature data include:

[0030] Establish a tag library, identify the feature data corresponding to each feature item in the feature data, input the feature data into the tag library for matching, obtain the corresponding single tag, and combine the obtained single tags to obtain the corresponding source tag.

[0031] Methods for building a tag library include:

[0032] Identify existing features, obtain potential feature data based on the identified features, assign a corresponding single label to each feature data, establish a first database, add the corresponding single label to the feature data, input the feature data into the first database for storage, and mark the current first database as a label library.

[0033] The matching module is used to establish a matching graph for the video source, and the specific methods include:

[0034] Obtain the source tags of each video source, convert the obtained source tags into corresponding matching points, establish a coordinate system corresponding to the matching points, input the obtained matching points into the coordinate system, and generate a coordinate graph based on the current coordinate system; set a summarization area in the coordinate graph, which is used to perform matching and summarization of corresponding matching points based on the target points input into the coordinate graph. The summarization area is a separate area, generally located in the blank area next to the coordinate system; mark the current coordinate graph as the matching graph.

[0035] Methods for converting obtained source tags into corresponding matching points include:

[0036] Identify individual tags in the tag library, assign corresponding values ​​to each individual tag, and use existing common sense to assign values ​​accordingly. Then, build a corresponding assignment matching table based on the assigned values.

[0037] Identify each individual label within the source label, match the corresponding assignment from the assignment matching table based on the identified individual labels, and integrate the obtained assignments into matching points according to the positions of each individual label within the source label.

[0038] The working methods of the summarization area include:

[0039] Identify the position of each target point in the coordinate graph, generate a corresponding range circle with the target point as the center, correct the generated range circle to obtain the range area, identify the matching points included in the range area, and include the identified matching points in the summarization area.

[0040] Methods for generating a corresponding range circle centered on the target point include:

[0041] The coordinate graph is divided into several regions based on each coordinate axis. A corresponding region radius is set for each region. The region radius corresponding to the region where each target point is located is identified. A corresponding range circle is generated with the target point as the center and the identified region radius as the radius.

[0042] The coordinate graph is divided into several regions based on each coordinate axis. This division is done progressively according to the coordinate axes, mainly based on the meaning of the data items corresponding to the coordinate axes and the distribution of their values. Specifically, the division is done manually.

[0043] Setting a corresponding radius for each region is done manually when dividing the regions.

[0044] The method for correcting the generated range circle is as follows: a corresponding correction model is built based on a CNN network or a DNN network, and the corresponding training set is set manually for training. The range circle is corrected by the correction model after successful training. Specifically, the correction is made according to the progressive relationship between the position of the range circle and the distribution of the radius of each region, especially for range circles with more than two regions.

[0045] The retrieval module is used to perform corresponding video retrieval based on user input data, and the specific methods include:

[0046] The process involves acquiring user-input search data and extracting key data. Key data refers to the content within the search data that limits the videos to be searched. Corresponding prompt templates can be set to facilitate user input of search data. Existing recognition and extraction technologies can be used to obtain the corresponding key data from the search data, so this will not be described in detail. A matching item table is established, which is built based on the data items corresponding to each coordinate axis of the matching points. The key data is divided into several demand item data, which are the directional data that limits the videos, such as tutorial videos for using production equipment in a certain field. The matching values ​​between each demand item data and each matching item in the matching item table are calculated, and the obtained matching values ​​are combined to form the corresponding target point set. The target point set is input into the matching graph, and the video sources corresponding to the matching points in the summarization area are identified. The obtained video sources are integrated into the target search source, and the search data is input into the target search source to perform video retrieval, obtaining several search videos. The obtained search videos are then recommended to the user.

[0047] Methods for calculating the matching values ​​between each requirement item data and each matching item in the matching item table include:

[0048] The matching value is derived based on the inclusion relationship between the range corresponding to the matching item and the data of the demand item. For example, the matching degree is determined based on the inclusion relationship, and the corresponding similar assignment is located according to the assignment in the assignment matching table. Specifically, the corresponding matching model is built based on CNN network or DNN network, and the corresponding training set is built manually for training. The matching model after successful training is analyzed to obtain the matching value between the data of the demand item and each matching item.

[0049] The obtained matching values ​​are combined to form the corresponding target point set. That is, the matching values ​​corresponding to each requirement item data are combined to obtain the corresponding target points. After summarizing, the corresponding target point set is obtained. During the combination, matching values ​​that do not meet the threshold of the corresponding matching item are removed. That is, a threshold is set manually for each matching item to delete matching values ​​that do not meet the matching requirements, thereby reducing the amount of subsequent data processing.

[0050] The above formulas are all numerical calculations after removing dimensions. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0051] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A video retrieval and analysis system based on big data, characterized in that, It includes a tag module, a matching module, a search module, and a server; The tagging module is used to tag the video source with the corresponding tags; the matching module is used to create a matching graph of the video source. The retrieval module is used to perform corresponding video retrieval based on user input data, obtain the user-input retrieval data, extract key data from the retrieval data, establish a matching item table, divide the key data into several demand item data, calculate the matching value between each demand item data and each matching item in the matching item table, and combine the obtained matching values ​​into a corresponding target point set; input the target point set into the matching graph, identify the video source corresponding to the matching point in the summarization area, integrate the obtained video sources into the target retrieval source, input the retrieval data into the target retrieval source to perform video retrieval, obtain several retrieval videos, and recommend the obtained retrieval videos to the user; The matching module works by including: Obtain the source tags of each video source, convert the obtained source tags into corresponding matching points, establish a coordinate system corresponding to the matching points, input the obtained matching points into the coordinate system, generate a coordinate graph based on the current coordinate system, set a summarization area in the coordinate graph, the summarization area is used to perform matching summarization of corresponding matching points based on the target points input into the coordinate graph, and mark the current coordinate graph as a matching graph; Methods for converting obtained source tags into corresponding matching points include: Identify individual tags in the tag library, assign corresponding values ​​to each individual tag, and build a corresponding assignment matching table based on the assigned values; Identify each individual label within the source label, match the corresponding assignment from the assignment matching table based on the identified individual labels, and integrate the obtained assignments into matching points according to the positions of each individual label within the source label; The working methods of the summarization area include: Identify the position of each target point in the coordinate graph, generate a corresponding range circle with the target point as the center, correct the generated range circle to obtain the range area, identify the matching points included in the range area, and include the identified matching points in the summarization area.

2. The video retrieval and analysis system based on big data according to claim 1, characterized in that, The working methods of the tag module include: Obtain the available video sources, generate a corresponding video source detail table based on the obtained video sources, obtain the feature data of each video source based on the video source detail table, and set the source tag of the corresponding video source based on the obtained feature data.

3. The video retrieval and analysis system based on big data according to claim 2, characterized in that, Methods for setting source tags for corresponding video sources based on the obtained feature data include: Establish a tag library, identify the feature data corresponding to each feature item in the feature data, input the feature data into the tag library for matching, obtain the corresponding single tag, and combine the obtained single tags to obtain the corresponding source tag.

4. The video retrieval and analysis system based on big data according to claim 3, characterized in that, Methods for building a tag library include: Identify the features, obtain the feature data based on the identified features, set a corresponding single label for each feature data, establish a first database, add the corresponding single label to the feature data, input the feature data into the first database for storage, and mark the current first database as a label library.

Images