CN113591743B - Handwriting video identification method, system, storage medium and computing device - Google Patents

Abstract

The application relates to a handwriting video identification method, a handwriting video identification system, a storage medium and a computing device, wherein the handwriting video identification method comprises the following steps: acquiring and processing initial handwriting video data; collecting and obtaining priori knowledge of the writing sequence of the cursive script and the cursive script symbol; extracting video key frame video pictures from initial handwriting video data by combining prior knowledge, and converting the key frame pictures into texts; vectorizing texts of the pictures to obtain multidimensional vectors of each text, splicing the multidimensional vectors generated by each picture according to a time sequence, and combining to form video vectors; and carrying out vector dimension reduction visualization processing on the video to finish classification recognition. The application can improve the accuracy of the video recognition of the cursive script and calligraphy and can be widely applied to the technical field of video data recognition.

Description

Calligraphy video recognition method, system, storage medium and computing device

Technical field

The present invention relates to the technical field of video data recognition, and in particular to a calligraphy video recognition method, system, storage medium and computing device for cursive scripts.

Background technique

Cursive script is a product that facilitates quick and continuous writing. Therefore, during the writing process, there will be phenomena such as connected strokes, deformation, simplification, and sloppy writing. This makes the calligraphy font shape different from the general simplified regular script writing method, which brings difficulties to the recognition of handwritten calligraphy. . However, this kind of simplification of writing is not arbitrary, it has certain rules. Since ancient times, the rules of simplification of cursive writing have not been fixed. However, in the gradual evolution, everyone will fix it by mutual agreement for a certain character or a certain structure. Come down with a cursive writing method. Cursive script has developed its unique technical system in the evolution of calligraphy style, the most important of which is the simplification of stroke order and cursive script symbols.

The adjustment of the stroke order makes cursive writing more natural and convenient when writing continuously. For example, the writing next to the vertical center can change from "first write the left dot, then write the right dot, and finally write the vertical dew" to "first write the short vertical, then fold the pen to write the short horizontal, then turn the pen to the upper left, and write with the flow. Long vertical". Cursive script symbols are radicals written in concise symbols that replace regular script. These cursive script components are constantly developed and evolved by the calligraphers of the past dynasties summarizing the rules. In modern times, Mr. Yu Youren summarized the writing methods of these cursive script components into standard cursive script symbols, and later expanded and proposed 71 radical cursive script symbols and 355 radical cursive script symbols in the book "Analysis of Cursive Script Calligraphy". Ordinary people must first understand some of these prior knowledge when understanding cursive handwriting calligraphy. Therefore, introducing information about cursive calligraphy stroke order and symbols is crucial in calligraphy recognition, especially the recognition of calligraphy videos. Video features include temporal changes in image features compared to common image features, and calligraphy writing videos can better reflect the stroke order information of cursive writing. Although there are many established methods in the field of video action recognition and video classification at this stage. In recent years, neural networks have achieved results that almost surpass human performance in computer vision tasks such as image recognition and object detection. Researchers are increasingly using neural networks in video tasks, such as neural networks based on three-dimensional convolution, Neural network based on dual streams, etc.

However, there are few studies in the field of calligraphy video recognition.

Contents of the invention

In response to the above problems, the purpose of the present invention is to provide a calligraphy video recognition method, system, storage medium and computing device, which improves the accuracy of cursive calligraphy video recognition.

In order to achieve the above object, the present invention adopts the following technical solution: a calligraphy video recognition method, which includes: obtaining initial calligraphy video data and processing; collecting and obtaining prior knowledge of cursive cursive writing stroke order and cursive symbols; combining the prior knowledge in Extract video key frame video pictures from the initial calligraphy video data, and convert the key frame pictures into text; vectorize the text of the picture to obtain a multi-dimensional vector of each text, and splice and combine the multi-dimensional vectors generated by each picture in time sequence. Form a video vector; perform vector dimensionality reduction and visualization processing on the video to complete classification and recognition.

Further, the acquisition and processing of the initial calligraphy video data include: using a crawler to crawl the initial calligraphy video data; filtering out videos that have clear video effects and do not block the written content beyond the preset range; and filtering the filtered videos. Extract the single-word video in the video.

Further, the prior knowledge includes: stroke order information in cursive script that is different from regular script writing.

Further, the method of extracting video key frame video pictures from the initial calligraphy video data by combining prior knowledge includes: calling the opencv package, intercepting video frames at preset intervals, and saving them as pictures; and according to the stroke order information of running script and cursive script. Cursive symbols, obtain the approximate progress position of key frames in the video; automatically filter each calligraphy video according to the key frames to obtain a fixed number of key frame video pictures.

Further, converting the key frame pictures into text includes: converting the characteristic information of the pixel points of each picture into text storage; standardizing the picture to a fixed length and width, and performing grayscale processing; extracting the image of the picture Numeric matrix and generate its transpose matrix, concatenate the image numerical matrix and its transpose matrix to obtain the text of the picture.

Further, the combination forms a video vector, including: calling the gensim package, using the Doc2Vec document embedding model to vectorize the image text, and pre-setting the length and window parameters of the text vector; traversing the vector dimensions and window parameters to form a Doc2Vec document The embedding model determines the optimal parameters; the vectors generated from each picture of the same video are spliced in chronological order and combined to form a video vector.

Further, performing vector dimensionality reduction and visualization processing on the video includes: performing manifold learning on the video vector, performing dimensionality reduction visualization, converting a high-dimensional matrix into a two-dimensional vector group, and treating each document as a Scatter points are drawn into a graph; the vectors of the same word obtained on the dimensionality reduction result graph are gathered in similar places on the graph, and classification and recognition are completed based on the obtained graph.

A calligraphy video recognition system, which includes: an initial data acquisition module, a priori knowledge collection module, a text conversion module, a vectorization module and a recognition module; the initial data acquisition module is used to acquire initial calligraphy video data and process it; The prior knowledge collection module is used to collect and obtain prior knowledge of cursive writing stroke order and cursive symbols; the text conversion module combines prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and converts the Key frame pictures are converted into text; the vectorization module vectorizes the text of the picture to obtain a multi-dimensional vector of each text, splices the multi-dimensional vectors generated by each picture in time sequence, and combines them to form a video vector; the identification module, The video is subjected to vector dimensionality reduction and visualization processing to complete classification and recognition.

A computer-readable storage medium storing one or more programs including instructions that, when executed by a computing device, cause the computing device to perform any of the above methods.

A computing device comprising: one or more processors, a memory, and one or more programs, wherein one or more programs are stored in the memory and configured for execution by the one or more processors, so The one or more programs include instructions for performing any of the methods described above.

Since the present invention adopts the above technical solutions, it has the following advantages:

1. The present invention raises the problem of static character recognition based on artificial intelligence to a problem of cursive character recognition using prior knowledge of dynamic stroke order.

2. The present invention introduces prior knowledge such as the order of writing strokes and cursive symbols of cursive script, and improves the accuracy of video recognition of cursive calligraphy.

3. The present invention uses an unsupervised algorithm for video and image embedding training, which is conducive to the promotion of the application of the present invention.

Description of drawings

Figure 1 is a schematic flow chart of a calligraphy video recognition method in an embodiment of the present invention;

Figure 2 is a schematic flow chart of a method for identifying running script and cursive script in an embodiment of the present invention;

Figure 3 is a schematic diagram of storing crawled videos in a local disk in an embodiment of the present invention;

Figure 4 is a schematic diagram of prior knowledge of cursive symbols in an embodiment of the present invention;

Figure 5 is a schematic structural diagram of a computing device in an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the described embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art fall within the scope of protection of the present invention.

It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular forms are also intended to include the plural forms unless the context clearly indicates otherwise. Furthermore, it will be understood that when the terms "comprises" and/or "includes" are used in this specification, they indicate There are features, steps, operations, means, components and/or combinations thereof.

The present invention is a method of applying video embedding to identify cursive calligraphy videos based on prior knowledge of cursive hand order and cursive symbols, and involves technologies and methods such as network video acquisition, video processing, video embedding, and picture feature extraction. The present invention only involves the extraction of time-series visual modal information in the video. It needs to ignore the occlusion, jitter, and perspective changes caused by video background changes and shooting, and accurately identify the text writing information in the calligraphy video. The focus of the present invention is the recognition of calligraphy videos, and the problem of static character recognition based on artificial intelligence is raised to a problem of cursive character recognition using prior knowledge of dynamic stroke order. Since the single-character calligraphy video needs to be segmented first, building a self-built database is time-consuming. Therefore, the method used in this invention is an unsupervised method for small-scale data sets.

In one embodiment of the present invention, as shown in Figure 1, a calligraphy video recognition method is provided. This embodiment illustrates the application of this method to a terminal. It can be understood that this method can also be applied to a server. It can be applied to systems including terminals and servers, and is implemented through the interaction between terminals and servers. The recognition method provided in this embodiment can not only be used to identify cursive calligraphy videos, but can also be applied to other fields to identify other video data. For example, it can also be used to identify cursive calligraphy videos. This embodiment takes cursive calligraphy as an example. , there are no restrictions on other calligraphy types. In this embodiment, the method includes the following steps:

Step 1. Obtain initial calligraphy video data and process it;

Step 2. Collect and acquire prior knowledge of cursive writing strokes and cursive symbols;

Step 3. Combine the prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and convert the key frame pictures into text;

Step 4. Vectorize the text of the picture to obtain a multi-dimensional vector of each text, splice the multi-dimensional vectors generated from each picture in time sequence, and combine them to form a video vector;

Step 5: Perform vector dimensionality reduction and visualization on the video to complete classification and recognition.

In a preferred embodiment, obtaining and processing the initial calligraphy video data in step 1 includes the following steps:

Step 11. Use a crawler to crawl the initial calligraphy video data;

Crawling calligraphy writing videos on short video websites and caching them into local files. Some results are shown in Figure 3.

Step 12. For the training effect, select videos with clear video effects and no occlusion of the written content beyond the preset range; in this embodiment, the preset range occlusion is preferably 25% occlusion;

Step 13: Intercept the single-character videos in the filtered videos so that each video only contains the writing process of a single Chinese character, name the videos, and intercept the user watermarks before and after deleting the videos.

Specifically: in this embodiment, a crawler is used to crawl the initial calligraphy video data and process it. There are a large number of calligraphy writing videos on today's short video websites, but the videos on these websites are unevenly shot, and all downloaded videos contain a few seconds of user watermark at the end. Therefore, after downloading the videos, these videos need to be processed, artificially filtered to obtain videos with clear video effects and no significant obstruction to the written content, and the watermarks of the last few seconds are uniformly deleted. Because one video in some videos contains the writing process of multiple words, single-word videos are processed and recognized in this embodiment, so these videos need to be intercepted.

In a preferred embodiment, in step 2, prior knowledge of cursive writing stroke order and cursive symbols is collected and acquired. The prior knowledge includes information on stroke order in cursive writing that is different from regular script writing.

Specifically: refer to the website's complete collection of cursive radicals, the most complete cursive radical writing method and books Yu Youren's "Standard Cursive Script", Liu Dongqin's "Analysis of Cursive Script", Sun Baowen's "A Practical Dictionary of Running Script" and other reference materials, combined with daily writing usage habits , collecting commonly used stroke order information and cursive symbols that are different from regular script writing in cursive script.

Because standard reference materials prefer not to create confusion in cursive writing when formulating standards for cursive symbols, symbols with unique correspondences are promoted, but this is not the case in actual use. Therefore, the present invention combines daily usage habits and summarizes commonly used cursive symbols in running script and cursive script and their corresponding representative radicals and used words. Since the present invention focuses on the communication of ideas of solutions, only 35 sets of commonly used cursive symbols are collected as examples.

Since the stroke order information of cursive script is collected, the method of the present invention can not only identify single-word videos of cursive calligraphy, but also distinguish regular script calligraphy videos and cursive calligraphy videos of a certain word.

In a preferred embodiment, step 3 combines prior knowledge to extract video key frame video pictures from the initial calligraphy video data, including the following steps:

Step 311: Call the opencv package to intercept video frames at preset intervals and save them as pictures;

Step 312: According to the stroke order information and cursive symbols of running script and cursive script, obtain the approximate progress position of the key frame in the video;

Although the video length and the number of extracted pictures are different, the writing speed of each stroke when a person writes is similar. For a set of writing videos of each text to be recognized, according to the stroke order rules and cursive symbols of running script and cursive script, a rough series of key frame progress positions in the video are set.

Step 313: Automatically filter each calligraphy video according to key frames to obtain a fixed number of key frame video pictures.

Specifically: due to training time issues, it is impossible to propose and train every frame in the video, so this embodiment combines previous prior knowledge to extract single-word calligraphy video key frames. For a set of writing videos of each text to be recognized, according to the stroke order rules and cursive symbols of running script and cursive script, a rough series of key frame progress positions in the video are set. According to the set key frames, each calligraphy video is automatically filtered to obtain a fixed number of key frame video pictures.

In a preferred embodiment, converting keyframe images into text in step 3 includes the following steps:

Step 321: Convert the feature information of the pixels of each picture into text storage;

Step 322: Standardize the image to a fixed length and width, and perform grayscale processing;

Step 323: Extract the image value matrix (i.e., pixel matrix) of the picture and generate its transpose matrix, and splice the image value matrix and its transpose matrix to obtain the text of the picture.

In this embodiment, the Doc2Vec algorithm in unsupervised learning is used, so the key frame pictures must be converted into text first. First, the image is grayscaled so that the image only contains brightness information and does not contain redundant color information. Among them, the value of the white point is 255, the value of the black point is 0, and 0 to 255 are grayscale points. Extract the image numerical matrix of the picture and generate its transpose matrix. In order to extract the horizontal and vertical features of the picture at the same time, the image numerical matrix and its transpose matrix are spliced. Save the image textualization result in a txt file and store it locally.

In a preferred embodiment, combining to form a video vector in step 4 includes the following steps:

Step 41. Call the gensim package, use the Doc2Vec document embedding model to vectorize the image text, and preset the length and window parameters of the text vector;

Since the Doc2Vec model can create a fixed-length vectorized representation of a document, regardless of its length. Use the Doc2Vec function in the gensim package to input the text representation of the image into the function, and preset parameters such as the length and window of the document vector.

Step 42: Traverse the vector dimensions and window parameters to determine the optimal parameters for the Doc2Vec document embedding model;

Step 43: Splice the vectors generated from each picture of the same video in chronological order and combine them to form a video vector.

Specifically: train the Doc2Vec model for image text, input the text representation of the image into the function, and preset parameters such as the length and window of the document vector. Using the spatial representation of the PV-DM model training vector in Doc2Vec, the model output obtains a multi-dimensional vector for each text. Each dimension represents a hidden feature of the image represented by the text. These features summarize the calligraphy image represented by the text. Horizontal and vertical features.

In a preferred embodiment, performing vector dimensionality reduction and visualization processing on the video in step 5 includes the following steps:

Step 51: Perform manifold learning on the video vectors, perform dimensionality reduction visualization, convert the high-dimensional matrix into a two-dimensional vector group, treat each document as a scatter point, and draw it into a graph;

Among them, in this embodiment, the T-SNE method is used for dimensionality reduction visualization.

Step 52: Obtain vectors of the same word on the dimensionality reduction result map and gather them in similar places on the map, and complete classification and recognition based on the obtained map.

Specifically: after using unsupervised learning to represent the single-character calligraphy video into a vector form. Perform manifold learning on the generated video vectors, use the T-SNE method for dimensionality reduction visualization, convert the high-dimensional matrix into a two-dimensional vector group, treat each document as a scatter point, and draw it into a graph. In the picture of the dimensionality reduction result, it can be seen that the vectors of the same word are gathered in similar places on the picture.

Applying this vector visualization method, classification experiments can also be performed on labeled single-word videos. For unrecognized calligraphy single-character videos, the method of the present invention can be used for identification.

In one embodiment of the present invention, a calligraphy video recognition system is provided, which includes: an initial data acquisition module, a priori knowledge collection module, a text conversion module, a vectorization module and a recognition module;

The initial data acquisition module is used to obtain initial calligraphy video data and process it;

The prior knowledge collection module is used to collect and obtain prior knowledge of cursive writing strokes and cursive symbols;

The text conversion module combines prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and converts the key frame pictures into text;

The vectorization module vectorizes the text of the picture to obtain a multi-dimensional vector of each text, splices the multi-dimensional vectors generated by each picture in time sequence, and combines them to form a video vector;

The recognition module performs vector dimensionality reduction and visualization processing on the video to complete classification and recognition.

The system provided by this embodiment is used to execute each of the above method embodiments. Please refer to the above embodiments for specific processes and details, which will not be described again here.

As shown in Figure 5, it is a schematic structural diagram of a computing device provided in an embodiment of the present invention. The computing device may be a terminal, which may include: a processor (processor), a communication interface (Communications Interface), a memory (memory), a display screens and input devices. Among them, the processor, communication interface, and memory complete communication with each other through the communication bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. When the computer program is executed by the processor, it implements an identification method; the internal memory is non-volatile. Provides an environment for the operation of operating systems and computer programs in permanent storage media. The communication interface is used for wired or wireless communication with external terminals. The wireless mode can be implemented through WIFI, manager network, NFC (Near Field Communication) or other technologies. The display screen may be a liquid crystal display or an electronic ink display. The input device may be a touch layer covered on the display screen, or may be a button, trackball or touch pad provided on the housing of the computing device, or may be an external Keyboard, trackpad or mouse, etc. The processor can call logical instructions in memory to perform methods such as:

Obtain initial calligraphy video data and process it; collect and obtain prior knowledge of cursive writing stroke order and cursive symbols; combine the prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and convert the key frame pictures into text; The text of the picture is vectorized to obtain a multi-dimensional vector of each text, and the multi-dimensional vectors generated by each picture are spliced in time sequence and combined to form a video vector; the video is subjected to vector dimensionality reduction and visualization processing to complete classification and recognition.

In addition, the above-mentioned logical instructions in the memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. .

Those skilled in the art can understand that the structure shown in Figure 5 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computing device to which the solution of the present application is applied. The specific computing device can May include more or fewer parts than shown, or combine certain parts, or have a different arrangement of parts.

In one embodiment of the present invention, a computer program product is provided. The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions When executed by a computer, the computer can execute the methods provided in each of the above method embodiments, including, for example: obtaining initial calligraphy video data and processing it; collecting and obtaining prior knowledge of the cursive writing stroke order and cursive symbols; combining the prior knowledge in the initial Extract video key frame video pictures from the calligraphy video data, convert the key frame pictures into text; vectorize the text of the picture to obtain a multi-dimensional vector of each text, and splice the multi-dimensional vectors generated by each picture in time sequence to form a combination Video vector; perform vector dimensionality reduction and visualization processing on the video to complete classification and recognition.

In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided. The non-transitory computer-readable storage medium stores server instructions. The computer instructions cause the computer to execute the methods provided in the above embodiments, for example, including : Obtain initial calligraphy video data and process it; collect and obtain prior knowledge of cursive cursive writing strokes and cursive symbols; combine prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and convert the key frame pictures into text ;Vectorize the text of the picture to obtain a multi-dimensional vector of each text, splice the multi-dimensional vectors generated by each picture in time sequence, and combine them to form a video vector; perform vector dimensionality reduction and visualization processing on the video to complete classification and recognition.

The implementation principles and technical effects of the computer-readable storage medium provided by the above embodiments are similar to those of the above method embodiments, and will not be described again here.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each process and/or block in the flowchart illustrations and/or block diagrams, and combinations of processes and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing device produce a use A device for realizing the functions specified in one process or multiple processes of the flowchart and/or one block or multiple blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory that causes a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction means, the instructions The device implements the functions specified in a process or processes of the flowchart and/or a block or blocks of the block diagram.

These computer program instructions may also be loaded onto a computer or other programmable data processing device, causing a series of operating steps to be performed on the computer or other programmable device to produce computer-implemented processing, thereby executing on the computer or other programmable device. Instructions provide steps for implementing the functions specified in a process or processes of a flowchart diagram and/or a block or blocks of a block diagram.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be used Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent substitutions are made to some of the technical features; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (8)

1. A calligraphy video recognition method, characterized by including: Obtain initial calligraphy video data and process it; Collect and acquire prior knowledge of cursive writing strokes and cursive symbols; Combining prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and convert the key frame video pictures into text; Vectorize the text of the picture to obtain a multi-dimensional vector of each text, splice the multi-dimensional vectors generated from each picture in time sequence, and combine to form a video vector; Perform vector dimensionality reduction and visualization processing on the video to complete classification and recognition; The prior knowledge includes: information on the stroke order in cursive script that is different from regular script writing; The method of extracting video key frame video pictures from the initial calligraphy video data by combining prior knowledge includes: Call the opencv package to intercept video frames at preset intervals and save them as pictures; According to the stroke order information and cursive symbols of running script and cursive script, the progress position of the key frame in the video is obtained; Automatically filter each calligraphy video according to key frames to obtain a fixed number of key frame video pictures. 2. The identification method according to claim 1, characterized in that said obtaining and processing the initial calligraphy video data includes: Use a crawler to crawl the initial calligraphy video data; Filter out videos that have clear video effects and do not block the written content beyond the preset range; Extract single-word videos from the filtered videos. 3. The identification method according to claim 1, wherein converting the key frame video picture into text includes: Convert the feature information of the pixels of each image into text storage; Standardize the image to a fixed length and width and perform grayscale processing; Extract the image numerical matrix of the picture and generate its transpose matrix, and concatenate the image numerical matrix and its transpose matrix to obtain the text of the picture. 4. The identification method according to claim 1, characterized in that the combination to form a video vector includes: Call the gensim package, use the Doc2Vec document embedding model to vectorize image text, and preset the length and window parameters of the text vector; Traverse the vector dimensions and window parameters to determine the optimal parameters for the Doc2Vec document embedding model; The vectors generated from each picture of the same video are spliced in chronological order and combined to form a video vector. 5. The identification method according to claim 1, characterized in that said performing vector dimensionality reduction and visualization processing on the video includes: Perform manifold learning on the video vectors, perform dimensionality reduction visualization, convert the high-dimensional matrix into a two-dimensional vector group, treat each document as a scatter point, and draw it into a graph; The vectors of the same word obtained on the dimensionality reduction result map are gathered in similar places on the map, and classification and recognition are completed based on the obtained map. 6. A calligraphy video recognition system, characterized by including: an initial data acquisition module, a priori knowledge collection module, a text conversion module, a vectorization module and a recognition module; The initial data acquisition module is used to acquire initial calligraphy video data and process it; The prior knowledge collection module is used to collect and obtain prior knowledge of cursive writing strokes and cursive symbols; The text conversion module combines prior knowledge to extract video key frame video pictures from the initial calligraphy video data, and converts the key frame video picture into text; The vectorization module vectorizes the text of the picture to obtain a multi-dimensional vector of each text, splices the multi-dimensional vectors generated by each picture in time sequence, and combines them to form a video vector; The identification module performs vector dimensionality reduction and visualization processing on the video to complete classification and identification; The prior knowledge includes: information on the stroke order in cursive script that is different from regular script writing; The method of extracting video key frame video pictures from the initial calligraphy video data by combining prior knowledge includes: Call the opencv package to intercept video frames at preset intervals and save them as pictures; According to the stroke order information and cursive symbols of running script and cursive script, the progress position of the key frame in the video is obtained; Automatically filter each calligraphy video according to key frames to obtain a fixed number of key frame video pictures. 7. A computer-readable storage medium storing one or more programs, characterized in that the one or more programs include instructions that, when executed by a computing device, cause the computing device to perform as claimed Any of the methods described in 1 to 5. 8. A computing device, characterized by comprising: one or more processors, a memory, and one or more programs, wherein one or more programs are stored in the memory and configured as the one or more The processor executes, and the one or more programs include instructions for executing any one of the methods according to claims 1 to 5.