CN114581934A

CN114581934A - Test paper image processing method, device and equipment

Info

Publication number: CN114581934A
Application number: CN202210158414.1A
Authority: CN
Inventors: 何鑫; 韩文源; 张帅一; 汪昆; 杨扬; 蒋冠军
Original assignee: Alibaba China Co Ltd
Current assignee: Alibaba China Co Ltd
Priority date: 2022-02-21
Filing date: 2022-02-21
Publication date: 2022-06-03

Abstract

One or more embodiments of the present specification provide a method, an apparatus, and a device for processing a test paper image, including: acquiring a test paper image set corresponding to a test paper; carrying out structural analysis on each test paper image in the test paper image set respectively to obtain a question set contained in each test paper image; integrating the question sets corresponding to the test paper images according to the page number sequence of the page numbers corresponding to the test paper images so as to generate the electronic test paper document corresponding to the test paper.

Description

Test paper image processing method, device and equipment

Technical Field

One or more embodiments of the present disclosure relate to the field of image processing, and in particular, to a method, an apparatus, and a device for processing a test paper image.

Background

In the related art, with the popularization of networks and the increasing development of internet technologies, various online education platforms emerge. Online education is becoming more and more popular because of the advantages of being free from time and place limitations. In addition to online education, business processing for test paper is also a very important component in online education scenarios. For example, a user uploads a test paper image to an online education platform, the online education platform performs image processing on the image, acquires a topic in the image, and performs subsequent business processing based on the acquired topic.

Disclosure of Invention

In view of this, one or more embodiments of the present disclosure provide a method, an apparatus, and a device for processing a test paper image

To achieve the above object, one or more embodiments of the present disclosure provide the following technical solutions:

according to a first aspect of one or more embodiments of the present specification, there is provided a method for processing a test paper image, including:

acquiring a test paper image set corresponding to a test paper;

carrying out structural analysis on each test paper image in the test paper image set respectively to obtain a question set contained in each test paper image;

integrating the question sets corresponding to the test paper images according to the page number sequence of the page numbers corresponding to the test paper images so as to generate the electronic test paper document corresponding to the test paper.

According to a second aspect of one or more embodiments of the present specification, there is provided a test paper image processing apparatus including:

the acquisition module is used for acquiring a test paper image set corresponding to the test paper;

the analysis module is used for respectively carrying out structural analysis on each test paper image in the test paper image set to obtain a question set contained in each test paper image;

and the generation module is used for integrating the question set corresponding to each test paper image according to the page number sequence of the page number corresponding to each test paper image so as to generate the electronic test paper document corresponding to the test paper.

According to a third aspect of one or more embodiments of the present specification, there is provided an electronic apparatus including:

a processor;

a memory for storing processor-executable instructions;

the processor executes the executable instructions to realize the test paper image processing method.

According to a fourth aspect of one or more embodiments of the present specification, a computer-readable storage medium is provided, on which computer instructions are stored, which when executed by a processor implement the above-mentioned method for processing a test paper image.

In this specification, an online education platform obtains a test paper image set corresponding to a test paper, and performs structured analysis on each test paper image in the test paper image set to obtain a question set included in each test paper image. Integrating the question set corresponding to each test paper image by the online education platform according to the page sequence of the page corresponding to each test paper image to generate an electronic test paper document corresponding to the test paper

The online education platform of the specification provides a method for analyzing by the granularity of test paper, because the online education platform firstly carries out structured analysis on a single-page test paper image to obtain a topic set corresponding to the single-page test paper image and then integrates the topic sets corresponding to a plurality of pages of test paper images. By taking the test paper as granularity, the context relationship among the questions can be considered, and the overall content of the test paper questions can be considered, so that the identified question set is more accurate.

Drawings

FIG. 1 is a networking architecture diagram of an online education system shown in an exemplary embodiment of the present description;

FIG. 2 is a flow chart illustrating a method of processing a test paper image in accordance with an exemplary embodiment of the present disclosure;

FIG. 3 is a schematic diagram of a test paper image shown in an exemplary embodiment of the present description;

FIG. 4 is a schematic diagram of another test paper image shown in an exemplary embodiment of the present description;

FIG. 5 is a diagram illustrating structured attributes of a test paper title and title in accordance with an exemplary embodiment of the present specification;

FIG. 6 is a hardware block diagram of a device shown in an exemplary embodiment of the present description;

fig. 7 is a block diagram of a device for processing a test paper image according to an exemplary embodiment of the present specification.

Detailed Description

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with one or more embodiments of the present specification. Rather, they are merely examples of apparatus and methods consistent with certain aspects of one or more embodiments of the specification, as detailed in the claims which follow.

It should be noted that: in other embodiments, the steps of the corresponding methods are not necessarily performed in the order shown and described herein. In some other embodiments, the method may include more or fewer steps than those described herein. Moreover, a single step described in this specification may be broken down into multiple steps for description in other embodiments; multiple steps described in this specification may be combined into a single step in other embodiments.

Conventional online education platforms are generally analyzed at a topic granularity, so the online education platforms generally support related processing services of topic images. For example, the user takes a picture to search for questions, etc.

On the one hand, analysis with topic as granularity can obtain a user and topic relation chain, and the relation chain can not completely and accurately describe and depict the user.

On the other hand, with the rapid development of online education, a business demand for the entire contents of the test paper appears. For example, an electronic test paper is generated based on a test paper image uploaded by a user. And performing various analyses (such as test paper difficulty, subject section analysis and the like) on the electronic test paper. Under such a requirement, the analysis method with the granularity of the example of the title is not applicable any more.

In view of this, the present specification proposes a method for processing a test paper image. In this specification, an online education platform obtains a test paper image set corresponding to a test paper, and performs structured analysis on each test paper image in the test paper image set to obtain a question set included in each test paper image. Integrating the question set corresponding to each test paper image by the online education platform according to the page sequence of the page number corresponding to each test paper image to generate an electronic test paper document corresponding to the test paper

The online education platform of the specification provides a method for analyzing by the granularity of test paper, because the online education platform firstly carries out structured analysis on a single-page test paper image to obtain a topic set corresponding to the single-page test paper image and then integrates the topic sets corresponding to a plurality of pages of test paper images. On one hand, a more complete relationship chain of the user, the test paper and the subject can be obtained by analyzing the test paper granularity, so that the online education platform can be better helped to describe and depict the user. On the other hand, the context relationship between the questions can be considered by taking the test paper as the granularity, and the overall content of the questions of the test paper is considered, so that the identified question set is more accurate, and the business requirements for the overall content of the test paper are better met.

Referring to fig. 1, fig. 1 is a networking architecture diagram of an online education system shown in an exemplary embodiment of the present specification.

As shown in fig. 1, the online education system includes: an online education platform and a plurality of electronic devices.

The online education platform is a platform for processing online education services and is the back end of the online education services. The online education platform may be constructed from a device having computing capabilities. For example, the computing-capable device may be a physical server comprising a standalone host, or may be a virtual server, a cluster of servers, a data center, a computer, or the like. The online education platform is exemplified herein and not particularly limited.

Several electronic devices are front ends of online education services, for example, an electronic device may refer to a user terminal installed with an online education client or logging in an online education web page. Wherein, user terminal includes: mobile phones, tablet devices, notebook computers, Personal Digital Assistants (PDAs), wearable devices (such as smart glasses, smart watches, etc.), and the like. The user terminal is only exemplified here and is not particularly limited.

In the online education system, the electronic device may transmit an image to an online education platform. The online education platform can acquire a test paper image set corresponding to the test paper, and respectively perform structural analysis on each test paper image in the test paper image set to obtain a question set contained in each test paper image. The online education platform can integrate the question set corresponding to each test paper image according to the page number sequence of the page number corresponding to each test paper image so as to generate an electronic test paper document corresponding to the test paper.

Referring to fig. 2, fig. 2 is a flowchart illustrating a method for processing a test paper image according to an exemplary embodiment of the present disclosure. The method can be applied to an online education platform in the online education system, and can comprise the following steps.

Step 202: and acquiring a test paper image set corresponding to the test paper.

In an optional implementation manner, in practical application, due to operations such as user mistaken uploading, images uploaded by a user are not necessarily all test paper images containing test paper titles. For example, the user may upload non-test paper images, such as images completely unrelated to the test paper, answer sheets, etc., to the online education platform.

Therefore, in the embodiment of the present specification, the online education platform needs to screen out the test paper images from the images uploaded by the user to form a test paper image set, and then perform subsequent processing based on the test paper image set.

When the method is implemented, the online education platform can acquire at least one image uploaded by the user terminal, extract identification features from the at least one image, identify the test paper images in the at least one image based on the extracted identification features, and create a test paper image set based on the identified test paper images.

In an alternative implementation manner, in order to make the identification of the test paper image more accurate, the identification feature includes: image features and text features. Of course, the above-mentioned identification feature may also include an image feature or a text feature, and the test paper image is identified from at least one image uploaded by the user only by one of the image feature or the text feature.

Taking the identification features as image features and text features as an example, the detailed description of "extracting the identification features from the at least one image and identifying the test paper image in the at least one image based on the extracted identification features" is performed through steps a1 to a 2.

Step A1: and extracting image features from the at least one image, and identifying candidate images related to the test paper in the at least one image based on the image features.

In implementation, for each of at least one image, the online education platform may extract image features of the image based on the image feature extraction model. The image features of the image are then input to a trained first classification model that outputs a probability that the image is a candidate image associated with the test paper and a probability that the image is a non-candidate image not associated with the test paper. When the probability that the image is a candidate image is greater than a preset threshold, the online education platform determines the image as the candidate image.

The candidate image related to the test paper refers to a candidate image related to the test paper, such as a test paper image, an answer sheet image, and the like. Here, the candidate images related to the test paper are only exemplarily described, and are not particularly limited.

Wherein, the first classification model is obtained by training two sample label pairs. Wherein, the sample in one sample label pair is the image characteristic of the image relevant to the test paper, the label is the candidate image, the sample in the other sample label pair is the image characteristic of the image irrelevant to the test paper, and the label is the non-test paper image.

The image feature extraction model may be a model from which image features can be extracted. For example, the image feature extraction model may be a convolutional neural network, a ResNet (residual neural network), or the like, and the image feature extraction model is only described as an example and is not particularly limited.

In addition, in practical applications, the online education platform may also perform the determination of the candidate images by means of feature comparison, and here, the "identifying the candidate image related to the test paper in the at least one image based on the image features" is only exemplarily illustrated and is not specifically limited.

And A2, carrying out optical character recognition on the candidate image to obtain a plurality of text boxes corresponding to the candidate image, extracting text features from the text boxes, and further identifying the test paper image from the candidate image based on the image features and the text features extracted from the candidate image.

In implementation, for each candidate image, the online education platform may perform OCR (Optical Character Recognition) Recognition on the candidate image to obtain a plurality of text boxes corresponding to the candidate image. For example, as shown in fig. 3, the solid line boxes in fig. 3 are text boxes obtained by performing OCR recognition on the candidate image.

The online education platform may then extract text features from the text box. The online education platform can splice the text features and the image features of the candidate images and input the splicing result into the trained second classification model. The second classification model may output a probability that the candidate image is a test paper image and a probability that the candidate image is an answer sheet. When the probability that the candidate image is the test paper image is greater than the preset threshold, the online education platform may determine the candidate image as the test paper image.

Wherein, the second classification model is obtained by training two sample label pairs. Wherein, the sample in one sample label pair is the image characteristic of the test paper image, the label is the test paper, the sample in the other sample label pair is the image characteristic of the answer sheet image, and the label is the answer sheet.

Of course, in practical applications, the online education platform determines whether the candidate image is a test paper image by means of feature matching according to the image feature and the text feature of the candidate image, which is only exemplary and not specifically limited herein.

In the embodiment of the present specification, step a2 is mainly to distinguish the test paper image from the answer sheet, so the text feature is a text feature for distinguishing the test paper image from the answer sheet. Such as the position of the text information on the test paper image and the answer sheet is different. For example, the number and distribution of the printed forms and the handwritten forms on the test paper image and the answer sheet are different, for example, the printed form text on the test paper image is more, and the handwritten form text on the answer sheet is more.

Based on the difference between the test paper image and the answer sheet, the text features set by the present specification may include any one or a combination of more of the following: text information included in the image, the position of each text information in the image, the ratio of the handwritten text to the print text in the image, and the position distribution of the print text in the image.

In addition, it should be noted that, when screening out a test paper image from at least one test paper image uploaded by a user, on one hand, the present specification filters an image and an answer sheet irrelevant to the test paper by combining image features and text features, and screens out the test paper image, and since the present specification adopts a feature description image with multiple dimensions, the accuracy of the test paper image screened out from at least one image uploaded by the user is higher. On the other hand, compared with the image features and the text features based on the images, for directly screening the test paper images from at least one image, the test paper images are screened from the candidate images by the image features of the images, then the candidate images are subjected to OCR recognition, and the test paper images are screened from the candidate images based on the text features and the image features. The secondary screening mode can eliminate the workload of extracting the text features of most non-candidate images, splicing the text features and the image features of the non-candidate images and the like, and can effectively reduce the calculated amount of the online education platform.

It should be noted that, this is only an example of a manner for acquiring a test paper image set, and in practical applications, the user terminal may prompt the user to upload a test paper image before the user uploads an image each time. Or the user terminal can detect the image to be uploaded by the user first, and when the image to be uploaded by the user is detected to be the test paper image, the user terminal sends the test paper image to the online education platform. In this case, since the user terminal uploads the test paper images to the online education platform, the online education platform can receive the test paper images uploaded by the user terminal to form a test paper image set. Here, the sample image set corresponding to the test paper is only exemplarily described, and is not particularly limited.

Step 204: and respectively carrying out structural analysis on each test paper image in the test paper image set to obtain a question set contained in each test paper image.

The structured analysis is to recognize and correct texts contained in each test paper image and to structurally express the test paper based on the corrected result.

When the method is implemented, the online education platform identifies key text boxes in a plurality of text boxes corresponding to each test paper image, and partitions the test paper image through the key text boxes to obtain a test question area containing at least one question. Then, the online education platform extracts all questions from the question area to form a question set corresponding to the test paper image.

Step 204 is described in detail below by steps B1 through B4.

Step B1: and determining a key text box from a plurality of text boxes obtained by performing OCR recognition on the test paper image aiming at each test paper image, and positioning a test question area from the test paper image according to the key text box.

Step B1 is explained in detail through step B11 to step B12.

Step B11: for each test paper image, determining a key text box from a plurality of text boxes obtained by performing OCR recognition on the test paper image.

In step 202, the online education platform performs OCR recognition on each candidate image to obtain a plurality of text boxes corresponding to each candidate image, that is, each candidate image corresponds to at least one text box. Therefore, when the candidate image is determined as the test paper image, the text boxes corresponding to the candidate image are the text boxes corresponding to the test paper image.

And determining a key text box from a plurality of text boxes corresponding to each test paper image.

In this specification, the key text box includes a plurality of types of text boxes. For example, the key text boxes of the present specification may include: a test paper title text box, a title type text box, and a footer text box.

The title text box refers to a text box containing a title. For example, as shown in FIG. 3, the title of the test paper is "higher than the fourth month of the high level of the middle school in the week" and the text box containing "higher than the fourth month of the middle school in the week" and the text box containing "higher than the first month" are determined as the text boxes of the title of the test paper.

The title type text box refers to a text box containing a title type. Wherein the topic types can include: choice questions, non-choice questions, answer questions, etc. The topic type is merely exemplified here, and is not particularly limited. As shown in FIG. 3, the title type text box includes "one, choice title: the text box of the main topic total 12 subjects … …' is a topic type text box.

The footer text box refers to a text box containing page numbers. For example, as shown in FIG. 3, the footer text box may be the text box in FIG. 3 that contains "

page

1 and 6 pages".

In the embodiment of the present specification, a key text box may be determined from a plurality of text boxes corresponding to the test paper image in a characteristic comparison manner.

For example, if the key text box includes multiple key text boxes, the text box feature corresponding to the key text box is set for each key text box. The online education platform may establish a library of standard text box features for each type of key text box. When established, the online education platform may use the features corresponding to different key text boxes in the standard test paper in the sample as the standard text box features. And then, the online education platform extracts the text box characteristics of each text box corresponding to the test paper image, carries out similarity calculation on the text box characteristics and the standard text box characteristics, and determines the text box as the key text box when the similarity is greater than a preset threshold value.

For example, the key text boxes include a test paper title text box, a title type text box, and a footer text box.

Taking the footer text box as an example, text box features matched with the footer text box can be set for the footer text box. Text box features such as footer text box are: a feature for indicating that the text contains a specified character of "XX page in total on the XX page", a feature for indicating the position of the lower end of the text box in the test paper image, a feature for indicating the percentage of the height of the text box to the height of the test paper image, and a feature for indicating the percentage of the width of the text box to the width of the test paper.

The online education platform can extract the text box characteristics from the footer text boxes in the standard test paper in the sample, and the extracted text box characteristics are used as standard text box characteristics to form a standard text box characteristic library. And then, the online education platform extracts the text box characteristics aiming at each text box in the test paper image uploaded by the user, then carries out similarity calculation on the extracted text box characteristics and the standard text box characteristics, and determines the text box as a footer text box when the similarity calculation is greater than a preset threshold value.

Similarly, the online education platform can also set corresponding text box characteristics for the test paper title text box and set corresponding text box characteristics for the question type text box, and the two types of text boxes are determined in a characteristic comparison mode.

Certainly, in practical application, a classification model may also be trained in advance, after the text box features of any text box corresponding to the test paper image are input to the classification model, the classification model may output probability values that the text box belongs to various key text boxes, and when the probability value belonging to a certain key text box is greater than a preset threshold, it is determined that the text box belongs to the type of key text box. For example, if the classification model outputs that the probability that the text box belongs to the test question heading text box is 80%, the probability that the text box belongs to the footer text box is 5%, the probability that the text box belongs to the topic type is 5%, and the preset threshold value is 70%, it is determined that the text box belongs to the test question heading text box.

Step B12: and positioning a test question area from the test paper image according to the key text box.

Wherein, the test question area refers to an area containing a plurality of questions.

In implementations, the online education platform may determine an area between two adjacent topic type textboxes as a question area, and/or the online education platform may determine an area between a topic type textbox and a footer textbox as a question area.

For example, as shown in FIG. 3, the online education platform may take an area between a topic type text box (i.e., a text box containing "choice topic: this general topic and 12 small topics … …") and a footer text box (i.e., a text box "

page

1 and 6") as a test question area. The test question area is the area circled by the dotted line in fig. 3.

Step B2: extracting shape and position characteristics corresponding to at least one text box contained in the test question area; and identifying a reference text box serving as a topic segmentation point from the at least one text box based on the shape and position characteristics and the semantic characteristics corresponding to the text contained in the at least one text box.

In the embodiment of the present specification, since the reference text box is a text box indicating the topic division point, the reference text box needs to have the characteristic of the topic division point. The online education platform can extract the characteristics reflecting the characteristics of the question segmentation points to determine a reference text box from a plurality of text boxes corresponding to the test paper images. And dividing the test question area into a plurality of question areas by taking the reference text box as a segmentation point of the question.

The reference text box may be a text box indicating the start of a topic or a text box indicating the end of a topic.

For example, as shown in FIG. 3, the text box "2, inequality

A solution of () indicates a text box where a title begins, and this type of text box can be used as a reference text box.

For another example, as shown in fig. 3, a text box containing "C, [0,1 ∞ D, (∞, 0], [1 ∞ and +∞)" indicates a text box in which a title ends, and this type of text box may be used as a reference text box.

The reference text box is only exemplified here and is not particularly limited.

Step B2 is described in detail below with reference to steps B21 through B22.

Step B21: extracting shape and position characteristics corresponding to at least one text box contained in the test question area; and semantic features corresponding to the text contained in the at least one text box.

Since the configuration and semantic features are used to determine the reference text box as the topic segmentation point, the configuration and phonetic features can express the features of the topic segmentation point.

The shape and position characteristics refer to shape characteristics and position characteristics of the text box.

Wherein the shape features include: and the size characteristic is used for representing the relation between the size of each text box and the size of the test paper image. For example, the size characteristic includes a ratio of a height of the text box to a height of the test paper image. The examples are illustrative only and not intended to be limiting.

The location features include a combination of one or more of:

and the position characteristics represent the position relation of the text boxes.

Representing the position characteristics of each text box in the test paper image where the text box is located;

the position characteristics of the position relation between each text box and the text box marked as the key text box on the test paper image are represented;

the position characteristics of the position relation among the text boxes comprise: a feature that characterizes whether the current text box and the previous text box are in a line, a text box that characterizes the beginning of a new line of the current text box, a feature that characterizes the distance between the current text box and the previous text box, and the like. The examples are illustrative only and not intended to be limiting.

The position feature for representing the position relationship between each text box and the text box marked as the key text box on the test paper image comprises: the characteristic of the position relation between the current text box and the previous theme type text box is represented, and the characteristic of the position relation between the current text box and the next theme type text box is represented. This is merely an example and is not particularly limited.

The semantic features include one or more of the following: whether the text box contains the characteristics of the specified characters and the position characteristics of the specified characters or not is indicated; and the characteristic shows whether the text information in the text box is continuous with the text information in the previous text box in semantics.

The designated characters may be, for example, question numbers, option symbols (e.g., A, B, C, D), answer points (e.g., in a choice question, answer points are "()", and in a fill-in question, answer points are "-"). The specific characters are only exemplified and not particularly limited herein.

It should be noted that, for example, if the reference text box is an item start text box, the text box includes an item number, and the text in the text box is not semantically continuous with the text in the previous text box. For another example, if the reference text box is an end-of-title text box, the text box contains the D option, and the text box is continuous with the semantics of the previous text box.

Step B22: and identifying a reference text box serving as a segmentation point of the title from the at least one text box based on the shape and position features and the semantic features.

In implementation, the online education platform may input the shape and location characteristics and the semantic characteristics of each text box into the third classification model, and the third classification model may output a probability value that the text box is a reference text box, and when the probability value is greater than a preset threshold, the online education platform may use the text box as the reference text box.

Step B3: and determining a text box set corresponding to each question contained in the test question area based on the reference text box, and combining the text boxes in the text box set to obtain a question area corresponding to each question contained in the test question area.

In implementation, the online education platform may use a text box located between two reference text boxes and one text box of the two reference text boxes as a text box set corresponding to one topic.

And combining the text box sets corresponding to each topic by the online education platform to obtain a topic area corresponding to the topic.

For example, when the reference text box is an topic start text box, the online education platform may use a text box located between two topic start text boxes, and a topic start text box located at a front position in the two topic start text boxes, as a text box set corresponding to one topic. And then, the online education platform combines the textboxes in the textbox set corresponding to the theme to obtain the theme area of the theme.

As shown in FIG. 3, assume that the two title start textboxes are textbox 1 and textbox 2, respectively.

Text box 1 is "2, inequality

The solution set of (c) is () ".

The text box 2 is "3", and the set a ═ 0,1,2}, if

Then the contrary negation proposition of B number () "of the eligible set is ()".

The online education platform may set, as the text box set of topic 2, the text boxes included between the text box 1 and the text box 2, i.e., the text box including "a, (∞, 0], (∞) B, [0, + ∞)" and the text box including "C, [0,1) < u > (1, + ∞) D, (∞, 0] < u > 1, + ∞)" and the text box located at the top among the two topic start text boxes, i.e., the text box 1.

Then, the online education platform merges the textboxes in the textbox set of the 2 nd question to obtain a question area of the 2 nd question. For example, the dashed area of the title 2 in fig. 3 is the title area.

For another example, if the reference text box is an item ending text box, the online education platform may use a text box located between two item ending text boxes and an item ending text box located at a rear position of the two item ending text boxes as a text box set corresponding to one item. And then, combining the textboxes in the textbox set corresponding to the title by the online education platform to obtain a test question area of the title.

As shown in FIG. 3, assume that the two title end text boxes are text box 3 and text box 4, respectively.

Assume that text box 3 contains "a. {0} B. { -3, -4} C. { -1, -2} D.

"a text box;

the text box 4 contains "C, [0,1 ] U (1, + ∞) D, (∞, 0)]Text box of U [1, + ∞) ". The online education platform may include a text box (i.e., including a "2, inequality") between text box 3 and text box 4

Is a () "and contains" A, (∞, 0)]Text boxes of u (1, +∞) B, [0, +∞) ] and text boxes located at the rear positions in the two title end text boxes (i.e. containing "C, [0,1) [ u (1, +∞) D, (∞, 0)]Text boxes of u [1, + ∞) ") as a set of text boxes of topic 2.

Step B4: and respectively extracting topics from topic areas corresponding to the topics contained in the test question areas, and sequencing the extracted topics based on the position sequence of the topic areas in the test question areas to obtain a topic set contained in the test paper image.

When the online education platform is realized, aiming at each topic area in the test question area, the online education platform extracts the topic from the topic area. The title can include a title number and title content.

In practical applications, since there are writing traces of the user on the test paper image, the question numbers of some questions are difficult to identify, which causes difficulty in sorting the questions based on the question numbers. Therefore, in the present specification, the online education platform can sort the extracted topics according to the position order of the topic areas in the test question areas. For example, the online education platform may sort the topics extracted from the topic area according to the sequence of the topic area from top to bottom in the test question area, so as to obtain a topic set included in the test paper image.

As shown in fig. 3, the test paper image shown in fig. 3 includes 6 dashed boxes, and the online education platform may sort the titles in the 6 dashed boxes according to the position order of the 6 dashed boxes from top to bottom, so as to generate the title set of the test paper image.

Of course, in practical applications, the online education platform may also sort the subjects according to the subject numbers of the subjects in the test paper images, and the sorting manner is exemplarily described herein, but is not particularly limited thereto.

Step 206: integrating the question sets corresponding to the test paper images according to the page number sequence of the page numbers corresponding to the test paper images so as to generate the electronic test paper document corresponding to the test paper.

When the online education platform is implemented, the page number corresponding to each test paper image can be identified from the page footer text box in each test paper image.

If the page numbers of all the test paper images are identified, the online education platform can sort the question sets corresponding to the test paper images according to the page number sequence of the page numbers corresponding to the test paper images to generate question sequences corresponding to the test paper, and generates electronic test paper documents according to the question sequences;

in the embodiment of the present specification, the number of the test paper is increased. In other words, the larger the page number, the larger the average value of the question numbers of the test sheet images. Based on this characteristic, in this specification, if the page numbers of a part of the test paper images are identified, the online education platform may sort the question sets corresponding to the test paper images according to the order of the average value of the question numbers in the question sets on the test paper images to generate question sequences corresponding to the test paper images, and generate an electronic test paper document according to the question sequences.

For example, assume that there are 3 test paper images, i.e., test paper image 1, test paper image 2, and test paper image 3. Assume that the page number is recognized as the 1 st page from the test paper image 1, and no page number is recognized from the

test paper images

2 and 3. The online education platform can calculate the average value of the question numbers on the

test paper images

1,2 and 3 and sort according to the average value.

Suppose that the test paper image 1 includes questions 1 to 4, and the average value of the question numbers is (1+2+3+ 4)/4-2.5.

The test sheet image 2 includes questions 15 to 21, and the average value of the question numbers is (15+16+17+18+19+20+ 21)/7-18.

The test paper image 3 includes questions 5 to 14, and the average value of the question numbers is (5+6+7+8+9+10+11+12+13+14)/10 is 9.5.

It can be seen that the order of the average question number is 2.5 < 9.5 < 18, and therefore the question sets of the three test paper images can be sorted in the order of the question set of the test paper image 1, the question set of the test paper image 3, and the question set of the test paper image 2, and the question sequence of the test paper can be obtained.

However, in practical applications, after the topic sets corresponding to the test paper images are sorted to obtain the topic sequences of the test paper, errors may occur in the topic sequences of the test paper. For example, some test questions have missing question numbers and repeated question numbers. In order to obtain a more accurate question sequence, the online education platform can correct the question sequence according to the question number of the question in the question sequence and generate an electronic test paper document according to the corrected question sequence.

The following describes "correcting a topic sequence according to the topic number of a topic in the topic sequence" in detail.

Case 1: aiming at the questions with the recognized question numbers in the question sequence, when the question numbers of any two adjacent questions are discontinuous, correcting the question sequence according to the question numbers of the any two adjacent questions

When the online education platform is realized, if the question numbers of the two questions are the same, the two questions are combined by the online education platform.

If the question with the identified question number in the question sequence is not continuous and the question numbers of two adjacent questions are the same, the two questions are the same question and are simply split by mistake. Therefore, in this case, two titles can be combined.

For example, when the last topic of the previous page is the same as the first topic of the next page, the topic sets of the two pages are sorted, and the obtained topic sequence is [ q1, q2, q2, q3 ]. Wherein q1, q2, and q3 represent the question numbers of the questions in the question sequence.

As can be seen from the question sequence, the second question mark and the third question mark in the question mark sequence are the same, and in this case, two questions represented by q2 can be combined to obtain a question sequence [ q1, q2, q3 ].

In an embodiment of the present specification, if the two topic question numbers are different, the online education platform re-segments a target topic located at a front position in the two topics, and replaces the target topic with a plurality of topics formed by segmentation.

In the implementation process, if the question numbers of the two questions are not continuous and the question numbers of the two questions are different, it indicates that a missed question-cutting error occurs, that is, the question positioned at the front position in the two questions is combined with multiple questions, and the question positioned at the front position needs to be re-divided.

During segmentation, the online education platform can acquire the text boxes in the topic areas corresponding to the topics positioned at the front positions, then acquire the probability values of the text boxes output by the third classification model, and reduce the preset threshold corresponding to the third classification model. And for each text box, if the probability value of the text box is greater than the reduced preset threshold value, determining that the text box is a reference text box serving as a title segmentation point.

Then, the online education platform divides the topic areas again based on the reference text box to obtain a plurality of new topic areas, extracts the topics from the new topic areas, and replaces the topics located at the front positions in the topic sequence with the extracted topics.

For example, suppose the topic sequence is [ q1, q2, q4, q5], and it can be seen from the topic sequence that the topic numbers of the second topic and the third topic are not consecutive, and the topic numbers of the two topics are different. At this point, the q2 topic can be re-segmented.

Suppose, as shown in FIG. 4, the title region corresponding to the q2 title is the region circled by the dashed box in FIG. 4. The title region includes 5 text boxes, namely a text box 401, a text box 402, a text box 403, a text box 404, and a text box 405.

The online education platform may acquire a probability value that the text box 401 output by the third classification model is the reference text box, a probability value that the text box 402 is the reference text box, a probability value that the text box 403 is the reference text box, a probability value that the text box 404 is the reference text box, and a probability value that the text box 405 is the reference text box.

The online education platform may then decrease the preset threshold corresponding to the third classification model. Assuming that the probability values of the text box 401 and the text box 404 are greater than the reduced preset threshold, the text box 401 and the text box 404 are used as reference text boxes.

Then, the online education platform can combine the text boxes 401 to 403 to generate a theme region corresponding to the theme q2, and extract the theme q2 from the theme region.

The online education platform may combine the textboxes 404 to 405 to generate a topic area corresponding to the topic q3 and extract the topic q3 from the topic area.

Then, the online education platform can replace q2 in the sequence of topics with q2, q3 to obtain a corrected sequence of topics [ q1, q2, q3, q4, q5 ].

Case 2: and aiming at the questions with missing question numbers in the question sequence, correcting the questions with missing question numbers according to the continuity of the question numbers of the existing questions.

In an optional implementation manner, if it is determined that the question number adjacent to the question with the missing question number is not continuous, the question number of the question with the missing question number is complemented according to the question number adjacent to the question with the missing question number.

In this case, the question number of the question with the missing question number can be complemented according to the question number adjacent to the question with the missing question number.

For example, the topic sequence is [ q1, q2, q? Q4, q5, q6], wherein q? Indicates the question with missing question number. In this example, the topics adjacent to the topic with the missing topic number are q2 and q 4. The question numbers of the subjects q2 and q4 are not continuous, so the question numbers of subjects with missing question numbers can be complemented by the question numbers of q2 and q 4. In this example, since q? Between q2 and q4, so q? The title sequence after completion of the title is q3 [ q1, q2, q3, q4, q5, q6 ].

In another optional implementation manner, if it is determined that the topic numbers adjacent to the topic with the missing topic number are continuous, the topics with the missing topic number are combined with the topics located at the front positions in the adjacent topics.

When the problem number is determined to be continuous with the problem number adjacent to the problem with the missing problem number, the problem is indicated to be mistakenly cut, namely, the problem originally serving as the same problem is cut into two problems. In this case, the online education platform may merge the topic number missing topic with the topic located at the front position in the adjacent topics.

For example, the topic sequence is [ q1, q2, q? Q3, q4], wherein, q? Indicates the question with missing question number. In this example, the topics adjacent to the topic with the missing topic number are q2 and q 3. The question numbers of the subjects q2 and q3 are consecutive, indicating q2 and q? Originally the same topic, but is mistakenly cut into two, and then q2 and q? And combining to obtain a corrected topic sequence [ q1, (q2+ q.

In practical applications, the subject sequence may be corrected in other ways, which are only exemplary and not specifically limited.

It should be noted that, in the conventional structured analysis with topic granularity, only a single topic is focused on, but the association relationship between different topics is not considered, so that some problems (such as false topic, missed topic, etc.) may occur in the identified topic. The structured analysis is carried out by the test paper granularity, and the question sequence is corrected according to the incidence relation among the questions in the test paper question sequence, so that the more accurate question sequence is obtained by the structured analysis carried out by the test paper granularity.

In addition, in the embodiment of the present specification, in addition to obtaining the question sequence, the online education platform may further analyze the obtained question sequence to obtain attribute information of the test paper, perform statistics on the attribute information of the test paper, and display the result through the visualization element statistics, so that the user may clearly and clearly obtain the attribute information of the test paper.

This is explained in detail through step C1 to step C3.

Step C1: and carrying out multi-dimensional classification on all questions in the generated electronic test paper document to obtain the attribute value of each question under at least one preset question attribute.

When the method is implemented, the online education platform can classify each topic in the electronic test paper document in a multi-dimensional way through a natural language processing technology to obtain the attribute value of each topic under at least one preset topic attribute. Wherein each classification dimension corresponds to each preset topic attribute.

Wherein the preset theme attributes include one or more of the following: subject segment, knowledge point, difficulty, problem solving method, problem type, etc.

For example, assume that the preset topic attributes include: subject segment, difficulty.

For each topic, the online education platform can classify the topic in the dimension of the discipline section. If the classification result of the topic is that the topic belongs to high mathematics, the topic attribute and the attribute value thereof corresponding to the topic are ' subject segment ═ high mathematics ', wherein the subject segment ' is a preset topic attribute, and ' high mathematics ' is an attribute value.

In addition, the online education platform can classify the subject in the difficulty dimension. If the classification result of the topic is that the topic belongs to the second class difficulty, the topic attribute corresponding to the topic and the attribute value thereof are "difficulty is equal to the second class difficulty", wherein "difficulty" is a preset topic attribute, and "second class difficulty" is an attribute value.

In addition, in the embodiment of the present specification, after obtaining the attribute value of each topic under at least one preset topic attribute, the online education platform may further correct the attribute value of the topic in order to ensure the accuracy of the attribute value.

When the method is realized, a specified topic attribute is specified in at least one topic attribute in advance, namely the specified topic attribute is one or more types below the preset topic attribute. For example, the preset theme attributes include: subject segment, difficulty. The pre-specified discipline segment is a specified topic attribute.

Aiming at the appointed subject attribute, the online education platform can count the number of various attribute values under the appointed subject attribute, then takes the attribute value with the maximum number as a standard attribute value, and corrects the attribute values of a plurality of subjects under the appointed attribute according to the standard attribute value.

For example, suppose the specified topic attribute is a subject section, and suppose that 15 topics are shared by topics written in the electronic test paper document, wherein the attribute values of the 1 st to 10 th, 14 th and 15 th topics in the subject section are high mathematics, and the attribute values of the 11 th to 13 th topics in the subject section are high mathematics.

The online education platform may count the number of attribute values of high mathematics (i.e., 12) and count the number of attribute values of high mathematics (i.e., 3). Since the high numerology is the largest in the number of attribute values, the high numerology is taken as the standard attribute value. Then, the property values of the 11 th to 13 th subjects under the discipline section are modified from higher first mathematics to higher second mathematics.

It should be noted that, the present specification analyzes with the test paper granularity, and by considering the context relationship between the topics, the attribute value of the topic can be effectively corrected, so that the attribute value of each topic identified by the online education platform is more accurate.

Step C2, identifying the test paper title from any test paper image in the test paper set, and identifying the attribute value under the preset title attribute from the test paper title

When the online education platform is implemented, the online education platform can identify the test paper title from the text information in the test paper title text box of any test paper image, and identify the attribute value under the preset title attribute from the test paper title.

The preset title attribute refers to information for describing a title of the test paper. For example, the preset title attribute may include one or more of the following: year, school time, subject section and test paper type.

For example, assume that the identified test paper titles are: the second school period of the university of the four colleges, 2020-2021, and 5 months of the school year, the examination questions are high in the number of two characters.

Assume that the preset title attributes include: year, school time, subject section and test paper type. The preset title attribute and the attribute value thereof identified from the test paper title are shown in table 1:

preset title attribute	Attribute value
		Year of year	2020-2021
School time	Second school term
		Discipline section	High number of characters
Type of test paper	Moon examination

TABLE 1

Step C3: writing the attribute value of each question and the attribute value of the test paper title into the electronic test paper document

When the method is implemented, the electronic equipment can write the attribute values of the test paper titles under the attribute of each title into the electronic test paper document. Such as writing the contents shown in table 1 to an electronic test paper document.

In addition, the attribute value of each topic of the electronic equipment under each preset topic attribute is written into the electronic test paper document. For example, consider "first topic: the subject section is high second-degree number, difficulty is second-class difficulty, knowledge point is analytic geometry, and question type is answer, and is written into the electronic test paper document.

Of course, in practical application, the online education platform can also display the relevant statistical information of the test paper to the user, so that the user can know the test paper condition more intuitively.

During implementation, the electronic device counts the number of attribute values appearing under each topic attribute, and displays the statistical result according to the visual display elements.

The visual display elements refer to elements capable of visually displaying the statistical results, and include graphic elements (such as sector diagrams, column diagrams, and the like), tables, and the like. The visual display elements are only exemplary and not specifically limited.

For example, if the question attribute is a question type, and the attribute values under the question attribute are a single-choice question, an answer question, and others, the statistical attribute value is the number of questions of the single-choice question (assumed to be 12 questions), the statistical attribute value is the number of questions of the answer question (assumed to be 6 questions), and the statistical attribute value is the number of questions of other types (assumed to be 4 questions). As shown in fig. 5, the online education platform may display the statistics of the topic types and their respective attribute values in the form of a table.

For example, the topic attribute is the difficulty, the attribute values under the attribute are first class difficulty, second class difficulty and third class difficulty, and if the test paper has 20 questions, 15 questions in the 20 questions are second class difficulty, 2 questions are first class difficulty, and 3 questions are third class difficulty, the topic proportion of the first class difficulty is 10%, the topic proportion of the second class difficulty is 75%, and the topic proportion of the third class difficulty is 15%. The online education platform may then display the statistical result in a fan shape as shown in fig. 5.

In addition, in practical applications, a user may upload test paper images of different sets of test papers to the online education platform. Therefore, before returning the electronic test paper document to the user terminal, the online education platform can also judge whether the test paper images in the test paper image set are the same test paper according to the information identified from the images in the test paper image set.

When the method is realized, before the electronic test paper document is returned to the user terminal, the online education platform can detect whether the test paper images in the test paper image set meet the preset conditions representing the same set of test paper.

Wherein the preset conditions include a combination of one or more of the following:

one of the at least one test paper image includes a test paper title;

the standard attribute value under the attribute of the specified subject identified from each subject is consistent with the attribute value under the attribute of the specified subject identified from the test paper title; wherein the specified title attribute is the same as the specified title attribute;

the number of repeated question marks is less than a preset threshold.

The following explains "the standard attribute value under the specified topic attribute identified from each topic coincides with the attribute value under the specified topic attribute identified from the test sheet title".

For example, the title attribute is designated as a discipline segment, and the title attribute is designated as a discipline segment.

Assuming that the test paper title is "the second school of the university of the four colleges, 2020-2021, second school year, test question of 5 months and high number of letters", the subject section identified from the title of the test paper is high number of letters.

Suppose that the test paper has 15 questions, the attribute values of the 1 st to 10 th questions, the 14 th and the 15 th questions in the subject section are high second literature, and the attribute values of the 11 th to 13 th questions in the subject section are high first literature. The online education platform may count the number of each attribute value and then select the attribute value having the largest number as the standard attribute value. In this example, if the online education platform counts that the number of questions with the attribute value of high number of second letters is 12, and the number of questions with the attribute value of high number of first letters is 3, the online education platform can determine the number of high numbers of second letters as the standard attribute value.

Then, the online education platform can detect whether the standard attribute value (i.e., high number of letters) under the specified topic attribute of the subject section identified from the topic is consistent with the attribute value (i.e., high number of letters) under the specified topic attribute identified from the test paper title, and in this example, the standard attribute value (i.e., high number of letters) under the specified topic attribute of the subject section identified from the topic is consistent with the attribute value (i.e., high number of letters) under the specified topic attribute identified from the test paper title, so that the second condition of the preset conditions is met.

In the embodiment of the present specification, if a test paper image in a test paper image set meets a preset condition representing the same set of test paper, the electronic test paper document is returned to the user terminal.

And if the test paper images in the test paper image set do not meet the preset condition representing the same set of test paper, correcting the electronic test paper document, and returning the corrected electronic test paper document to the user terminal. For example, if two test paper images in the test paper set include test paper titles, the electronic test paper document is cut into two electronic test paper documents, and each electronic test paper document corresponds to one set of paper. And the online education platform returns the two electronic test paper documents to the user terminal.

From the above description, it can be seen that, since the online education platform performs structured analysis on a single-page test paper image to obtain the topic sets corresponding to the single-page test paper image, and then integrates the topic sets corresponding to the multiple-page test paper image, the online education platform of this specification provides a method for analyzing with the test paper granularity.

On one hand, a more complete relationship chain of the user, the test paper and the subject can be obtained by analyzing the test paper granularity, so that the online education platform can be better helped to describe and depict the user.

On the other hand, the overall contents of the questions of the test paper can be considered according to the context relationship among the questions by taking the test paper as the granularity, so that the question sequence obtained after integrating the questions corresponding to the multiple pages of test paper images and the question attributes of all the questions can be corrected, and the obtained question sequence and the obtained question attribute values are more accurate.

Fig. 6 is a schematic block diagram of an online education platform provided in an exemplary embodiment. Referring to fig. 6, at the hardware level, the apparatus includes a processor 602, an internal bus 604, a network interface 606, a memory 608 and a non-volatile memory 610, but may also include hardware required for other services. One or more embodiments of the present description may be implemented in software, such as by processor 602 reading corresponding computer programs from non-volatile memory 610 into memory 608 and then executing. Of course, besides software implementation, the one or more embodiments in this specification do not exclude other implementations, such as logic devices or combinations of software and hardware, and so on, that is, the execution subject of the following processing flow is not limited to each logic unit, and may also be hardware or logic devices.

Referring to fig. 7, the apparatus for processing test paper image may be applied to the device shown in fig. 6 to implement the technical solution of the present specification. Wherein, the device for processing the test paper image can comprise:

an obtaining module 701, configured to obtain a test paper image set corresponding to a test paper;

an analysis module 702, configured to perform structured analysis on each test paper image in the test paper image set, respectively, to obtain a question set included in each test paper image;

the generating module 703 is configured to integrate the question sets corresponding to the test paper images according to the page sequence of the page corresponding to each test paper image, so as to generate an electronic test paper document corresponding to the test paper.

Optionally, the obtaining module 701 is configured to obtain at least one image uploaded by the user terminal when obtaining a test paper image set corresponding to a test paper; and extracting identification features from the at least one image, identifying a test paper image in the at least one image based on the extracted identification features, and creating a test paper image set based on the identified test paper image.

Optionally, the identification features include image features and text features;

the obtaining module 701 is configured to, when extracting an identification feature from the at least one image and identifying a test paper image in the at least one image based on the extracted identification feature, extract an image feature from the at least one image and identify a candidate image related to the test paper in the at least one image based on the image feature; and carrying out optical character recognition on the candidate image to obtain a plurality of text boxes corresponding to the candidate image, extracting text features from the text boxes, and further recognizing the test paper image from the candidate image based on the image features extracted from the candidate image and the text features.

Optionally, the text feature includes any one or a combination of more of the following:

text information included in the image, the position of each text information in the image, the ratio of the handwritten text to the print text in the image, and the position distribution of the print text in the image.

Optionally, the analysis module 702 is configured to, when performing structured analysis on each test paper image in the test paper image set to obtain a question set included in each test paper image, determine, for each test paper image, a key text box from a plurality of text boxes obtained by performing optical character recognition on the test paper image, and locate a test question area from the test paper image according to the key text box; extracting the shape and position characteristics corresponding to at least one text box contained in the test question area; semantic features corresponding to texts contained in the at least one text box are identified from the at least one text box on the basis of the shape and position features and the semantic features, and a reference text box serving as a title segmentation point is identified from the at least one text box; determining a text box set corresponding to each question contained in the test question area based on the reference text box, and combining the text boxes in the text box set to obtain a question area corresponding to each question contained in the test question area; and respectively extracting the questions from the question areas corresponding to the questions contained in the test question areas, and sequencing the extracted questions based on the position sequence of the question areas in the test question areas to obtain a question set contained in the test paper image.

Optionally, the key text box includes: a title type text box, a page number text box;

the analysis module 702 is configured to, when a test question area is located from the test paper image according to the key text box, use an area between two adjacent topic type text boxes and/or an area between a topic type text box and a page number text box as the test question area.

Optionally, the form and position characteristics include any one or a combination of more of the following:

the position characteristics represent the position relation between the text boxes;

the size characteristic of the relation between the size of each text box and the size of the test paper image is represented;

the position characteristics of the position relation between each text box and the key text box marked on the test paper image are represented;

the semantic features include any one or a combination of more of the following:

a feature that characterizes whether the text box contains a specified character;

characterizing a position feature of a specified character in the text box;

and the characteristic that whether the text information in the text box is continuous with the text information in the previous text box is represented.

Optionally, the key text box comprises a page number text box; the titles in the title set corresponding to the test paper images comprise: the subject content and the subject number thereof;

the generating module 703 is configured to, when integrating the question set corresponding to each test paper image according to the page sequence of the page number corresponding to each test paper image to generate the electronic test paper document of the test paper, identify the page number corresponding to each test paper image from the text information of the page number text box of each test paper image; if the page numbers corresponding to all the test paper images are identified, sorting the question sets corresponding to all the test paper images according to the page number sequence of the page numbers corresponding to all the test paper images to generate question sequences corresponding to the test paper, and generating electronic test paper documents according to the question sequences; if the page numbers corresponding to partial test paper images are identified, the question sets corresponding to the test paper images are sequenced according to the sequence of the average value of the question numbers in the question sets on the test paper images to generate question sequences corresponding to the test paper, and electronic test paper documents are generated according to the question sequences.

Optionally, the generating module 703 is configured to correct the question sequence according to the question number of the question in the question sequence when generating the electronic test paper document according to the question sequence, and generate the electronic test paper document according to the corrected question sequence.

Optionally, the generating module 703 is configured to correct the question sequence according to the question number of each question in the question sequence when the question sequence is corrected, and correct the question sequence according to the question number of any two adjacent questions when the question numbers of any two adjacent questions are not continuous. And aiming at the questions with missing question numbers in the question sequence, correcting the questions with missing question numbers according to the continuity of the existing question numbers.

Optionally, the generating module 703 is configured to, when the question sequence is corrected according to the question numbers of the two adjacent questions, merge the two questions if the question numbers of the two questions are the same; if the two topic question numbers are different, the target topic positioned at the front position in the two topics is divided again, and the target topic is replaced by a plurality of topics formed by division.

Optionally, the generating module 703 is configured to, when correcting an item with a missing item number according to the continuity of existing item numbers, if it is determined that an item number adjacent to the item with the missing item number is not continuous, complement the item number of the item with the missing item number according to the item number adjacent to the item with the missing item number; and if the question numbers adjacent to the questions with the missing question numbers are determined to be continuous, merging the questions with the missing question numbers with the questions positioned at the front positions in the adjacent questions.

Optionally, the generating module 703 is further configured to perform multidimensional classification on each topic in the generated electronic test paper document to obtain an attribute value of each topic under at least one preset topic attribute; identifying a test paper title from any test paper image in the test paper set, and identifying an attribute value under a preset title attribute from the test paper title; and writing the attribute value of each title and the attribute value of the test paper title into the electronic test paper document.

Optionally, the generating module 703 is further configured to count the number of multiple attribute values appearing under the attribute of the specified topic before writing the attribute value of each topic and the attribute value of the test paper title into the electronic test paper document; and taking the attribute value with the maximum number as a standard attribute value, and correcting the attribute values of the plurality of titles under the attribute of the specified title according to the standard attribute value.

Optionally, the preset title attribute includes one or more of the following: year, school time, subject section, type of test paper;

the preset theme attributes comprise one or more of the following: subject section, knowledge point, difficulty, problem solving method and problem type.

Optionally, the generating module 703 is further configured to count, for each topic attribute, the number of attribute values appearing under the topic attribute, and display a statistical result according to a visual display element

Optionally, the generating module 703 is further configured to detect whether the test paper images in the test paper image set satisfy a preset condition representing the same set of test paper, if so, return the electronic test paper document to the user terminal, and if not, correct the electronic test paper document, and return the corrected electronic test paper document to the user terminal.

Optionally, the preset condition includes a combination of one or more of the following:

one test paper image in the test paper image set comprises a test paper title;

the standard attribute value under the attribute of the designated subject identified from each subject is consistent with the attribute value under the attribute of the designated subject identified from the test paper title; wherein the specified title attribute is the same as the specified title attribute;

the number of repeated question marks is less than a preset threshold.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. A typical implementation device is a computer, which may take the form of a personal computer, laptop computer, cellular telephone, camera phone, smart phone, personal digital assistant, media player, navigation device, email messaging device, game console, tablet computer, wearable device, or a combination of any of these devices.

In a typical configuration, a computer includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

Computer-readable media, including both permanent and non-permanent, removable and non-removable media, may implement the information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The terminology used in the description of the one or more embodiments is for the purpose of describing the particular embodiments only and is not intended to be limiting of the description of the one or more embodiments. As used in one or more embodiments of the present specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, third, etc. may be used in one or more embodiments of the present description to describe various information, such information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of one or more embodiments herein. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.

The above description is intended only to be exemplary of the one or more embodiments of the present disclosure, and should not be taken as limiting the one or more embodiments of the present disclosure, as any modifications, equivalents, improvements, etc. that come within the spirit and scope of the one or more embodiments of the present disclosure are intended to be included within the scope of the one or more embodiments of the present disclosure.

Claims

1. A method for processing a test paper image comprises the following steps:

acquiring a test paper image set corresponding to a test paper;

2. The method of claim 1, wherein the obtaining of the test paper image set corresponding to the test paper comprises:

acquiring at least one image uploaded by a user terminal;

and extracting identification features from the at least one image, identifying a test paper image in the at least one image based on the extracted identification features, and creating a test paper image set based on the identified test paper image.

3. The method of claim 2, the identifying features comprising image features and text features;

the extracting identification features from the at least one image and identifying the test paper image in the at least one image based on the extracted identification features comprises:

extracting image features from the at least one image, and identifying candidate images related to the test paper in the at least one image based on the image features;

and carrying out optical character recognition on the candidate image to obtain a plurality of text boxes corresponding to the candidate image, extracting text features from the text boxes, and further recognizing a test paper image from the candidate image based on the image features extracted from the candidate image and the text features.

4. The method of claim 3, the textual features comprising a combination of any one or more of:

5. The method of claim 1, wherein the performing a structured analysis on each test paper image in the test paper image set to obtain a topic set included in each test paper image comprises:

determining a key text box from a plurality of text boxes obtained by carrying out optical character recognition on the test paper image aiming at each test paper image, and positioning a test question area from the test paper image according to the key text box;

extracting the shape and position characteristics corresponding to at least one text box contained in the test question area; semantic features corresponding to texts contained in the at least one text box are identified from the at least one text box on the basis of the shape and position features and the semantic features, and a reference text box serving as a title segmentation point is identified from the at least one text box;

determining a text box set corresponding to each question contained in the test question area based on the reference text box, and combining the text boxes in the text box set to obtain a question area corresponding to each question contained in the test question area;

and respectively extracting the questions from the question areas corresponding to the questions contained in the test question areas, and sequencing the extracted questions based on the position sequence of the question areas in the test question areas to obtain a question set contained in the test paper image.

6. The method of claim 5, the key text box comprising: a title type text box, a page number text box;

the positioning of the test question area from the test paper image according to the key text box comprises the following steps:

and taking the area between two adjacent theme type textboxes and/or the area between the theme type textbox and the page number textbox as a test question area.

7. The method of claim 5, the topographical features comprising any one or a combination of more of the following:

characterizing a position feature of a specified character in the text box;

and the characteristic that whether the semantics of the text information in the text box and the semantics of the text information in the previous text box are continuous is represented.

8. The method of claim 5, the key text box comprising a page number text box; the questions in the question set corresponding to the test paper image comprise: the subject content and the subject number thereof;

integrating the question set corresponding to each test paper image according to the page sequence of the page corresponding to each test paper image to generate the electronic test paper document of the test paper, comprising:

identifying the page number corresponding to each test paper image from the text information of the page number text box of each test paper image;

if the page numbers corresponding to all the test paper images are identified, sorting the question sets corresponding to all the test paper images according to the page number sequence of the page numbers corresponding to all the test paper images to generate question sequences corresponding to the test paper, and generating electronic test paper documents according to the question sequences;

if the page numbers corresponding to partial test paper images are identified, the question sets corresponding to the test paper images are sequenced according to the sequence of the average value of the question numbers in the question sets on the test paper images to generate question sequences corresponding to the test paper, and electronic test paper documents are generated according to the question sequences.

9. The method of claim 8, the generating an electronic test paper document in accordance with the sequence of topics, comprising:

and correcting the question sequence according to the question number of the question in the question sequence, and generating an electronic test paper document according to the corrected question sequence.

10. The method of claim 9, wherein correcting the sequence of topics by topic numbers of the topics in the sequence of topics comprises:

aiming at the questions with the identified question numbers in the question sequence, when the question numbers of any two adjacent questions are discontinuous, correcting the question sequence according to the question numbers of the any two adjacent questions;

and aiming at the questions with missing question numbers in the question sequence, correcting the questions with missing question numbers according to the continuity of the existing question numbers.

11. The method of claim 10, wherein the correcting the sequence of topics according to the topic numbers of any two adjacent topics comprises:

if the question numbers of the two questions are the same, combining the two questions;

if the two topic question numbers are different, the target topic positioned at the front position in the two topics is divided again, and the target topic is replaced by a plurality of topics formed by division.

12. The method of claim 10, wherein correcting the problem with missing question number according to the continuity of the existing question numbers comprises:

if the problem number adjacent to the problem with the missing problem number is determined to be discontinuous, filling the problem number of the problem with the missing problem number according to the problem number adjacent to the problem with the missing problem number;

and if the question numbers adjacent to the questions with the missing question numbers are determined to be continuous, merging the questions with the missing question numbers with the questions positioned at the front positions in the adjacent questions.

13. The method of claim 1, further comprising:

performing multi-dimensional classification on all questions in the generated electronic test paper document to obtain an attribute value of each question under at least one preset question attribute; and the number of the first and second groups,

identifying a test paper title from any test paper image in the test paper set, and identifying an attribute value under a preset title attribute from the test paper title;

and writing the attribute value of each title and the attribute value of the test paper title into the electronic test paper document.

14. The method of claim 13, prior to said writing the attribute values of the titles and the attribute values of the test paper titles to the electronic test paper document, further comprising:

counting the number of the multiple attribute values appearing under the specified title attribute;

and taking the attribute value with the maximum number as a standard attribute value, and correcting the attribute values of the plurality of titles under the attribute of the specified title according to the standard attribute value.

15. The method of claim 13, the preset title attributes comprising one or more of: year, school time, subject section, type of test paper;

16. The method of claim 13, further comprising:

and counting the number of attribute values appearing under each topic attribute, and displaying a counting result according to the visual display element.

17. The method of claim 14, further comprising:

and detecting whether the test paper images in the test paper image set meet preset conditions representing the same set of test paper, if so, returning the electronic test paper document to the user terminal, and if not, correcting the electronic test paper document and returning the corrected electronic test paper document to the user terminal.

18. The method of claim 17, the preset conditions comprising a combination of one or more of:

one test paper image in the test paper image set comprises a test paper title;

the number of repeated question marks is less than a preset threshold.

19. A device for processing a test paper image, comprising:

20. The apparatus according to claim 19, wherein the obtaining module, when obtaining the test paper image set corresponding to the test paper, is configured to obtain at least one image uploaded by the user terminal; and extracting identification features from the at least one image, identifying a test paper image in the at least one image based on the extracted identification features, and creating a test paper image set based on the identified test paper image.

21. The apparatus of claim 20, the identifying features comprising image features and text features;

the acquisition module is used for extracting image features from the at least one image and identifying candidate images related to the test paper in the at least one image based on the image features when the identification features are extracted from the at least one image and the test paper image in the at least one image is identified based on the extracted identification features; and carrying out optical character recognition on the candidate image to obtain a plurality of text boxes corresponding to the candidate image, extracting text features from the text boxes, and further recognizing the test paper image from the candidate image based on the image features extracted from the candidate image and the text features.

22. The apparatus of claim 21, the textual features comprising a combination of any one or more of the following:

23. The apparatus according to claim 19, wherein the analysis module, when performing structured analysis on each test paper image in the test paper image set to obtain a question set included in each test paper image, is configured to determine, for each test paper image, a key text box from a plurality of text boxes obtained by performing optical character recognition on the test paper image, and locate a test question area from the test paper image according to the key text box; extracting the shape and position characteristics corresponding to at least one text box contained in the test question area; semantic features corresponding to texts contained in the at least one text box are identified from the at least one text box on the basis of the shape and position features and the semantic features, and a reference text box serving as a title segmentation point is identified from the at least one text box; determining a text box set corresponding to each question contained in the test question area based on the reference text box, and combining the text boxes in the text box set to obtain a question area corresponding to each question contained in the test question area; and respectively extracting the questions from the question areas corresponding to the questions contained in the test question areas, and sequencing the extracted questions based on the position sequence of the question areas in the test question areas to obtain a question set contained in the test paper image.

24. The apparatus of claim 23, the key text box comprising: a title type text box, a page number text box;

and the analysis module is used for taking an area between two adjacent theme type text boxes and/or an area between the theme type text box and the page number text box as a test question area when the test question area is positioned from the test paper image according to the key text box.

25. The apparatus of claim 23, the topographical features comprising any one or a combination of more of the following:

characterizing a position feature of a specified character in the text box;

26. The apparatus of claim 23, the key text box comprising a page text box; the questions in the question set corresponding to the test paper image comprise: the subject content and the subject number thereof;

the generation module is used for identifying the page number corresponding to each test paper image from the text information of the page number text box of each test paper image when integrating the question set corresponding to each test paper image according to the page number sequence of the page number corresponding to each test paper image so as to generate the electronic test paper document of the test paper; if the page numbers corresponding to all the test paper images are identified, sorting the question sets corresponding to all the test paper images according to the page number sequence of the page numbers corresponding to all the test paper images to generate question sequences corresponding to the test paper, and generating electronic test paper documents according to the question sequences; if the page numbers corresponding to partial test paper images are identified, the question sets corresponding to the test paper images are sequenced according to the sequence of the average value of the question numbers in the question sets on the test paper images to generate question sequences corresponding to the test paper, and electronic test paper documents are generated according to the question sequences.

27. The apparatus of claim 26, wherein the generating module, when generating the electronic test paper document according to the topic sequence, is configured to correct the topic sequence according to the topic number of the topic in the topic sequence, and generate the electronic test paper document according to the corrected topic sequence.

28. The apparatus according to claim 27, wherein the generating module, when correcting the question sequence according to the question number of each question in the question sequence, is configured to correct, for a question with a recognized question number in the question sequence, when the question numbers of any two adjacent questions are not consecutive, the question sequence according to the question numbers of the any two adjacent questions; and aiming at the questions with missing question numbers in the question sequence, correcting the questions with missing question numbers according to the continuity of the existing question numbers.

29. The apparatus according to claim 28, wherein the generating module, when correcting the sequence of topics according to the topic numbers of any two adjacent topics, is configured to merge the two topics if the topic numbers of the two topics are the same; if the two topic question numbers are different, the target topic positioned at the front position in the two topics is divided again, and the target topic is replaced by a plurality of topics formed by division.

30. The apparatus according to claim 28, wherein the generating module, when correcting the problem with missing problem number according to the continuity of the existing problem number, is configured to, if it is determined that the problem number adjacent to the problem with missing problem number is not continuous, complement the problem number of the problem with missing problem number according to the problem number adjacent to the problem with missing problem number; and if the question numbers adjacent to the questions with the missing question numbers are determined to be continuous, merging the questions with the missing question numbers with the questions positioned at the front positions in the adjacent questions.

31. The apparatus according to claim 19, wherein the generating module is further configured to perform multidimensional classification on the topics in the generated electronic test paper document to obtain an attribute value of each topic under at least one preset topic attribute; identifying a test paper title from any test paper image in the test paper set, and identifying an attribute value under a preset title attribute from the test paper title; and writing the attribute value of each title and the attribute value of the test paper title into the electronic test paper document.

32. The apparatus according to claim 31, wherein the generating module is further configured to count the number of the plurality of attribute values appearing under the specified topic attribute before writing the attribute value of each topic and the attribute value of the test paper title into the electronic test paper document; and taking the attribute value with the maximum number as a standard attribute value, and correcting the attribute values of the plurality of titles under the attribute of the specified title according to the standard attribute value.

33. The apparatus of claim 31, the preset title attributes comprising one or more of: year, school time, subject section, type of test paper;

34. The apparatus of claim 31, wherein the generating module is further configured to, for each topic attribute, count a number of attribute values appearing under the topic attribute, and display a statistical result according to a visual display element.

35. The apparatus of claim 32, wherein the generating module is further configured to detect whether the test paper images in the test paper image set satisfy a preset condition representing the same set of test paper, if so, return the electronic test paper document to the user terminal, and if not, modify the electronic test paper document and return the modified electronic test paper document to the user terminal.

36. The apparatus of claim 35, the preset conditions comprising a combination of one or more of:

one test paper image in the test paper image set comprises a test paper title;

the standard attribute value under the attribute of the designated subject identified from each subject is consistent with the attribute value under the attribute of the designated subject identified from the test paper title; the specified title attribute is the same as the specified title attribute;

the number of repeated question marks is less than a preset threshold.

37. An electronic device, comprising:

a processor;

a memory for storing processor-executable instructions;

wherein the processor implements the method of any one of claims 1-18 by executing the executable instructions.

38. A computer readable storage medium having stored thereon computer instructions which, when executed by a processor, carry out the steps of the method according to any one of claims 1 to 18.