Multi-blank test question recognition and correction method, storage medium and equipment

Abstract

The invention discloses a multi-empty test question recognition and correction method, a storage medium and equipment, and the method comprises the steps: firstly positioning each answer region of a multi-empty test question, intercepting an image of each answer region, carrying out the vertical splicing of the intercepted answer region images, and carrying out the recognition and correction through employing an OCR recognition technology. The problem that the answer result corresponding to each answer area cannot be distinguished due to the fact that OCR recognition is directly carried out on the whole test question answering area of the multi-empty test questions, so that the answer result of each answer area is difficult to accurately compare and correct with the corresponding standard answer can be solved. The answer result is obtained through vertical splicing and then one-time identification, the calling frequency of an identification interface can be reduced, the identification efficiency is improved, one-time identification and correction are carried out on all the answer area images of a multi-empty test question, and statistics and display of scores of each multi-empty test question are facilitated.

Description

Technical Field

[0001] The present invention relates to the field of intelligent education technology, and in particular to a method for identifying and correcting multiple-blank test questions, a storage medium and a device. Background Art

[0002] At present, artificial intelligence is developing rapidly, and educational informatization has made great progress. Thanks to the high accuracy of OCR text recognition, automatic grading of objective questions has become possible. However, since OCR can only obtain the coordinates of the entire text for continuous text, for multiple-blank questions such as fill-in-the-blank questions that usually contain several answer areas, the continuous text recognized by OCR causes multiple answer results to be mixed together, so the recognized answer results cannot correspond to the answer areas one by one, and it is impossible to effectively grade such questions. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention aims to provide a method, storage medium and device for identifying and correcting multiple-blank test questions.

[0004] In order to achieve the above object, the present invention adopts the following technical solution:

[0005] A method for identifying and correcting multiple-blank test questions comprises the following steps:

[0006] S1. Obtain an image of the answer sheet to be corrected, and obtain the coordinates of each answer area of ​​each multiple-blank test question on the answer sheet, and cut out the image of each answer area separately on the image of the answer sheet to be corrected according to the coordinates of each answer area; all the images of the answer areas cut out of a multiple-blank test question constitute an answer area image subset;

[0007] S2, arranging and splicing the images of each answer area in the answer area image subset in a vertical direction, so that each row has only one answer area image;

[0008] S3. Use the image spliced ​​in step S2 as the image to be recognized, perform OCR text recognition on the image to be recognized, obtain the answer results of each answer area, compare the answer results of each answer area with the corresponding standard answer text, and obtain the correction result.

[0009] Furthermore, in step S1, a set of answer sheet template parameters is pre-set, and the set of answer sheet template parameters at least includes coordinate data of each answer area of ​​each multiple-blank question and standard answer text corresponding to each answer area of ​​each multiple-blank question.

[0010] Furthermore, in step S1, when capturing the image of the answer area, first locate the range of each answer area on the image of the answer sheet to be graded according to the coordinate data of each answer area, then determine the range to be captured according to the set ratio of the range of the answer area, and then capture the image of the answer area according to the range to be captured.

[0011] Furthermore, in step S1, when capturing the image of the answer area, based on the coordinate data of each answer area, the answer areas with different vertical coordinates are captured in sequence from top to bottom. On this basis, the answer areas with the same vertical coordinates are captured in sequence from left to right, and the captured images of each answer area are placed into the answer area image subset in the order of capture; in step S2, when splicing, the images of each answer area are arranged in the vertical direction in the order of capture and then spliced.

[0012] Furthermore, in step S1, the captured image of each answer area is bound to the corresponding standard answer text according to the coordinate data of the answer area; in step S3, after completing OCR text recognition of the image to be recognized, the recognized answer results are matched one by one with the images of each answer area in the answer area image subset in order from top to bottom, and then the recognized answer results are compared with the standard answer text bound to the corresponding answer area image. If the comparison results are consistent, the grading result is correct, otherwise it is wrong.

[0013] Furthermore, in step S3, the answer results in the answer area are extracted using handwritten text OCR recognition.

[0014] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0015] The present invention also provides a computer device, comprising a processor and a memory, wherein the memory is used to store a computer program; and when the processor is used to execute the computer program, the above method is implemented.

[0016] The beneficial effect of the present invention is that the present invention first locates each answer area of ​​a multiple-blank test question and captures the image of each answer area, then vertically splices the captured answer area images and then uses OCR recognition technology to identify and correct them, which can solve the problem that the answer results corresponding to each answer area cannot be distinguished due to directly performing OCR recognition on the entire test answer area of ​​the multiple-blank test question, making it difficult to accurately compare and correct the answer results of each answer area with the corresponding standard answer. In addition, the present invention obtains the answer result by first vertically splicing and then identifying it at one time, which can reduce the number of calls to the recognition interface and speed up the recognition efficiency. In addition, all the answer area images of a multiple-blank test question are recognized and corrected at one time, which is convenient for the statistics and display of the score of each multiple-blank test question. DETAILED DESCRIPTION

[0017] The present invention will be further described below. It should be noted that this embodiment is based on the technical solution and provides a detailed implementation method and a specific operation process, but the protection scope of the present invention is not limited to this embodiment.

[0018] This embodiment provides a method for identifying and correcting multiple-space test questions, which specifically includes the following steps:

[0019] S1. Obtain an image of the answer sheet to be corrected, and obtain the coordinates of each answer area of ​​each multiple-blank test question on the answer sheet, and cut out the image of each answer area separately on the image of the answer sheet to be corrected according to the coordinates of each answer area; the images of all answer areas intercepted in a multiple-blank test question constitute an answer area image subset.

[0020] S2, arranging and splicing the images of each answer area in the answer area image subset in a vertical direction, so that each row has only one answer area image, so as to ensure that the OCR recognition results are independent.

[0021] S3. Use the image spliced ​​in step S2 as the image to be recognized, perform OCR text recognition on the image to be recognized, obtain the answer results of each answer area, compare the answer results of each answer area with the corresponding standard answer text, and obtain the correction result.

[0022] In this embodiment, in step S1, a set of answer sheet template parameters is pre-set, and the set of answer sheet template parameters at least includes coordinate data of each answer area of ​​each multiple-blank question and standard answer text corresponding to each answer area of ​​each multiple-blank question. The coordinate data of the answer area may include the coordinates of the upper left corner and the lower right corner of the answer area, and the rectangular range of the answer area can be determined by the coordinates of the upper left corner and the lower right corner.

[0023] In this embodiment, in step S1, when capturing the image of the answer area, the range of each answer area is first located on the image of the answer sheet to be corrected according to the coordinate data of each answer area, and then the range to be captured is determined according to the set ratio of the range of the answer area, and then the image of the answer area is captured according to the range to be captured. By adding a certain amount of redundant space to capture the image of the answer area, it is possible to prevent errors in the range of the answer area determined according to the coordinate data, and prevent the complete answer area image from being captured.

[0024] In this embodiment, in step S1, when intercepting the image of the answer area, according to the coordinate data of each answer area, for answer areas with different ordinates, intercept from top to bottom in sequence, on this basis, for answer areas with the same ordinate, intercept from left to right in sequence, and put the intercepted images of each answer area into the answer area image subset in the order of interception; in step S2, when splicing, the images of each answer area are arranged in the vertical direction in the order of interception and then spliced. According to the general typesetting order and reading order, the answer area of ​​each question is arranged from top to bottom on the answer sheet, and the logical order of each answer area in each multiple-blank question is from left to right. The order of listing the standard answers corresponding to each answer area in each multiple-blank question is generally from left to right. Therefore, intercepting the images of the answer area from top to bottom and from left to right in sequence and putting them into the answer area image subset in the order of interception can help improve the accuracy and efficiency of the correspondence between the image of the answer area and the standard answer text.

[0025] In this embodiment, in step S1, the image of each answer area captured is bound to the corresponding standard answer text according to the coordinate data of the answer area; in step S3, after the OCR text recognition is completed for the image to be recognized, the recognized answer results are matched one by one with the images of each answer area in the answer area image subset in order from top to bottom, and then the recognized answer results are compared with the standard answer text bound to the corresponding answer area image, if the comparison result is consistent, the correction result is correct, otherwise it is wrong. Furthermore, the correction result can be further converted into a corresponding score.

[0026] In this embodiment, in step S3, the answer results in the answer area are extracted using handwritten text OCR recognition.

[0027] For those skilled in the art, various corresponding changes and modifications can be made according to the above technical solutions and concepts, and all of these changes and modifications should be included in the protection scope of the claims of the present invention.

Claims

1. A method for identifying and correcting multiple-blank test questions, characterized in that: The steps include: S1. Obtain an image of the answer sheet to be corrected, and obtain the coordinates of each answer area of ​​each multiple-blank test question on the answer sheet, and cut out the image of each answer area separately on the image of the answer sheet to be corrected according to the coordinates of each answer area; all the images of the answer areas cut out of a multiple-blank test question constitute an answer area image subset; S2, arranging and splicing the images of each answer area in the answer area image subset in a vertical direction, so that each row has only one answer area image; S3. Use the image spliced ​​in step S2 as the image to be recognized, perform OCR text recognition on the image to be recognized, obtain the answer results of each answer area, compare the answer results of each answer area with the corresponding standard answer text, and obtain the correction result.

2. The method for identifying and correcting multiple-blank test questions according to claim 1, characterized in that: In step S1, a set of answer sheet template parameters is preset, and the set of answer sheet template parameters at least includes coordinate data of each answer area of ​​each multiple-blank question and standard answer text corresponding to each answer area of ​​each multiple-blank question.

3. The method for identifying and correcting multiple-blank test questions according to claim 1, characterized in that: In step S1, when capturing the image of the answer area, first locate the range of each answer area on the image of the answer sheet to be graded according to the coordinate data of each answer area, then determine the range to be captured according to the set ratio of the range of the answer area, and then capture the image of the answer area according to the range to be captured.

4. The method for identifying and correcting multiple-blank test questions according to claim 1, characterized in that: In step S1, when capturing the image of the answer area, based on the coordinate data of each answer area, the answer areas with different vertical coordinates are captured from top to bottom. On this basis, the answer areas with the same vertical coordinates are captured from left to right, and the captured images of each answer area are placed in the order of capture into the answer area image subset; in step S2, when splicing, the images of each answer area are arranged in the vertical direction in the order of capture and then spliced.

5. The method for identifying and correcting multiple-blank test questions according to claim 2, characterized in that: In step S1, the captured image of each answer area is bound to the corresponding standard answer text according to the coordinate data of the answer area; in step S3, after completing OCR text recognition on the image to be recognized, the recognized answer results are matched one by one with the images of each answer area in the answer area image subset in order from top to bottom, and then the recognized answer results are compared with the standard answer text bound to the corresponding answer area image. If the comparison results are consistent, the grading result is correct, otherwise it is wrong.

6. The method for identifying and correcting multiple-blank test questions according to claim 1, characterized in that: In step S3, the answer results in the answer area are extracted using handwritten text OCR recognition.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

8. A computer device, characterized in that: The method comprises a processor and a memory, wherein the memory is used to store a computer program; when the processor is used to execute the computer program, the method described in any one of claims 1 to 6 is implemented.