CN113705509A

CN113705509A - Method and device for acquiring analysis information of test questions

Info

Publication number: CN113705509A
Application number: CN202111028206.1A
Authority: CN
Inventors: 陈天; 梁桂浩
Original assignee: Beijing Yundie Zhixue Technology Co ltd
Current assignee: Beijing Yundie Zhixue Technology Co ltd
Priority date: 2021-09-02
Filing date: 2021-09-02
Publication date: 2021-11-26

Abstract

The invention provides a method and a device for acquiring analysis information of test questions, wherein the method comprises the following steps: acquiring a test question analysis information image; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image; processing the character answer image sequence to generate a first HDR image; comparing the HDR image with a preset model to obtain the matching degree of the HDR image and the preset model; determining the score of the character answer according to the matching degree; judging whether the filling area of the answer image is filled; and determining the score of the machine-readable card answer according to the judgment result. Therefore, the accuracy and efficiency of paper marking are improved.

Description

Method and device for acquiring analysis information of test questions

Technical Field

The invention relates to the technical field of data processing, in particular to a method and a device for acquiring analysis information of test questions.

Background

With the continuous development of computer and mobile terminal technologies, more and more electronic terminal-based applications are greatly enriching the lives of people.

In daily learning, students usually adopt a mode of test questions to consolidate learning knowledge, the submitted test questions of the students generally consist of two parts, including a choice question part and a non-choice question part, for the choice question part, a paper reading teacher needs to evaluate one by one, but the efficiency is low, and the problem of misjudgment is easy to occur, so that the accuracy of paper reading is influenced.

For non-choice question parts, such as text parts, scoring needs to be performed according to subjective evaluation of teachers, so that final scoring is easily influenced by subjective emotion.

Therefore, how to improve the efficiency and accuracy of paper marking becomes a problem to be solved urgently.

Disclosure of Invention

The embodiment of the invention aims to provide a method and a device for acquiring analysis information of test questions, so as to solve the problem of low examination paper reading efficiency in the prior art.

In order to solve the above problem, in a first aspect, the present invention provides a method for acquiring analysis information of test questions, where the method includes:

acquiring a test question analysis information image; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image;

processing the character answer image sequence to generate a first HDR image;

comparing the HDR image with a preset model to obtain the matching degree of the HDR image and the preset model;

determining the score of the character answer according to the matching degree;

judging whether the filling area of the answer image is filled;

and determining the score of the machine-readable card answer according to the judgment result.

Preferably, the text answer image sequence includes: a first non-reference picture, a reference picture, and a second non-reference picture; the exposure duration of the first test question of the first non-reference image, the reference image and the second non-reference image is sequentially increased.

Preferably, the preset model comprises keywords of standard answers of the test question analysis information;

the comparing the HDR image with a preset model to obtain the matching degree of the HDR image and the preset model comprises:

acquiring character information in the HDR image;

comparing the character information with the keywords;

and grading the analysis information of the test questions according to the comparison result of the character information and the keywords.

Preferably, the scoring the test question analysis information according to the comparison result between the text information and the keyword includes:

and comparing the comparison result with a preset scoring standard, and scoring the analysis information of the test questions according to the comparison result.

Preferably, before the determining whether the filling area of the answer image is filled, the method further includes:

carrying out gray level processing on the machine-readable card answer image to generate a gray level image of the machine-readable card answer image;

identifying anchor point information of the gray-scale image, correcting the gray-scale image according to the anchor point information, and generating a corrected image of the machine-readable card answer image;

carrying out binarization processing on the corrected image to generate a binarized image of the machine-readable card answer image;

and analyzing the concentration and the area of a filling area on the machine-readable card answer image to identify whether the filling area is filled or not.

In a second aspect, the present invention provides an apparatus for acquiring test question analysis information, where the apparatus includes:

the acquisition unit is used for acquiring a test question analysis information image; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image;

the processing unit is used for processing the character answering image sequence to generate a first HDR image;

the comparison unit is used for comparing the HDR image with a preset model;

the obtaining unit is further configured to obtain a matching degree between the HDR image and a preset model according to a comparison result of the comparing unit;

the determining unit is used for determining the score of the character answer according to the matching degree;

the judging unit is used for judging whether a filling area of the answer image is filled;

the determining unit is further used for determining the score of the machine-readable card answer according to the judgment result.

the acquiring unit is further configured to acquire text information in the HDR image;

the comparison unit is also used for comparing the character information with the keywords;

and the scoring unit is used for scoring the test question analysis information according to the comparison result of the character information and the keywords.

Preferably, the comparison unit is further configured to compare the comparison result with a preset scoring standard;

the scoring unit is further used for scoring the analysis information of the test questions according to the comparison result of the comparison unit.

Preferably, the processing unit is further configured to perform gray scale processing on the machine-readable card answer image to generate a gray scale image of the machine-readable card answer image;

the identification unit is used for identifying the anchor point information of the gray-scale image, correcting the gray-scale image according to the anchor point information and generating a corrected image of the machine-readable card answer image;

the processing unit is also used for carrying out binarization processing on the corrected image to generate a binarization image of the machine-readable card answer image;

the processing unit is further used for analyzing the concentration and the area of a filling area on the machine-readable card answer image for the binary image so as to identify whether the filling area is filled or not.

By applying the method for acquiring the test question analysis information provided by the embodiment of the invention, a test question analysis information image is acquired; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image; processing the character answer image sequence to generate a first HDR image; comparing the HDR image with a preset model to obtain the matching degree of the HDR image and the preset model; determining the score of the character answer according to the matching degree; judging whether the filling area of the answer image is filled; and determining the score of the machine-readable card answer according to the judgment result. Therefore, the accuracy and efficiency of paper marking are improved.

Drawings

Fig. 1 is a flowchart of a method for acquiring analysis information of test questions according to an embodiment of the present invention;

fig. 2 is a schematic structural diagram of an apparatus for acquiring test question analysis information according to an embodiment of the present invention.

Detailed Description

The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Fig. 1 is a flowchart of a method for acquiring test question analysis information according to an embodiment of the present invention. The executing body of the method for acquiring the test question analysis information can be a terminal, an acquiring device of the test question analysis information and the like. As shown in fig. 1, the method for acquiring analysis information of test questions includes the following steps:

step 110, obtaining a test question analysis information image; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image.

The answer sheet can comprise an A surface and a B surface, the A surface is machine-readable card answer content, the B surface is character answer content, the machine-readable card answer content corresponds to a selected question answer, and the character answer content corresponds to a non-selected question answer.

And 120, processing the character answer image sequence to generate a first High-Dynamic Range (HDR) image.

Wherein, the image sequence of the character answer includes: a first non-reference picture, a reference picture, and a second non-reference picture; the exposure duration of the first test question of the first non-reference image, the reference image and the second non-reference image is sequentially increased. Because at least three character answer images are shot and then the three character answer images are synthesized into the HDR image, the problems that the collected image is not clear and needs to be collected for multiple times when one character answer image is collected are solved, the time is saved, and the efficiency is improved.

It should be noted that, the present application only takes three images as an example for description, it is understood that the number of the images may be more than three, when the number of the images is more than three, the exposure time lengths of the first non-reference image and the second non-reference image are closest to the reference image, the exposure time length of the first non-reference image is less than that of the reference image, and the exposure time length of the second non-reference image is greater than that of the reference image.

Next, how to acquire the first non-reference image, the reference image, and the second non-reference image will be described in detail.

First, three images with gradually increasing exposure time of a target scene (test question parsing information), namely, an original first non-reference image, an original reference image, and an original second non-reference image, may be captured by a terminal device, where the terminal device may be a device with a camera, including but not limited to a camera (e.g., a Digital camera), a video camera, a mobile phone (e.g., a smart phone), a tablet computer (Pad), a Personal Digital Assistant (PDA), a portable device (e.g., a portable computer), a wearable device, and the like, and this is not particularly limited in the embodiments of the present invention.

Next, the RGB images of the original first non-reference image, the original reference image, and the original second non-reference image are converted into grayscale images, respectively.

Then, on a brightness channel, Histogram Matching (HM) is performed on the gray scale map of the original first non-reference image by taking the gray scale map of the original reference image as a standard, the gray scale map of the original first non-reference image is brightened, HM is performed on the gray scale map of the original second non-reference image by taking the gray scale map of the original reference image as a standard, and the original second non-reference image is darkened.

And finally, acquiring homography matrixes of the brightened original first non-reference image, the original reference image and the darkened original second non-reference image by using a surf characteristic point detection algorithm, and mapping the original first non-reference image, the original reference image and the original second non-reference image by using the homography matrixes respectively to align the images, eliminate the movement caused by camera shake and obtain the first non-reference image, the reference image and the second non-reference image.

The method for processing the character answering image sequence comprises the following steps:

first, a first motion area and a second motion area of a character answer image sequence are obtained.

The RGB maps of the first non-reference image, the reference image and the second non-reference image are converted into grayscale maps, respectively. And performing HM on a brightness channel, performing HM on the gray scale map of the first non-reference image by taking the gray scale map of the reference image as a standard, turning up the gray scale map of the first non-reference image, performing histogram matching on the gray scale map of the second non-reference image, and turning down the gray scale map of the second non-reference image. Finally, comparing the gray value difference between the gray level image of the first non-reference image after the brightness adjustment and the gray level image of the reference image, determining all the pixels with the gray level difference larger than the preset threshold value as the first motion area, correspondingly, comparing the gray value difference between the gray level image of the second non-reference image after the brightness adjustment and the gray level image of the reference image, determining all the pixels with the gray level difference larger than the preset threshold value as the second motion area, for example, the gray value of the first pixel in the gray level image of the second non-reference image after the brightness adjustment is 200, the gray value of the first pixel in the gray level image of the reference image is 100, and the gray value difference between the two is 100, and when the preset threshold is 50, the difference between the gray values of the two is larger than the preset threshold, so that the pixel value of the first pixel point is 1, and after all the pixel points with the pixel values of 1 are obtained, the dried second motion area is obtained through threshold filtering and corrosion expansion operation.

Secondly, determining a target motion area according to a comparison result between the first proportion and a second threshold value and a comparison result between the second proportion and the second threshold value.

The first proportion is the proportion occupied by the first motion area in the first non-reference image, the second proportion is the proportion occupied by the second motion area in the second non-reference image, and the target motion area comprises at least one connected area.

And finally, obtaining a first HDR image according to the first weight value and the second weight value.

The first weighted value is the sum of the product of the saturation, the contrast and the exposure degree of each pixel point in the first non-reference image and the product of the saturation, the contrast and the exposure degree of each pixel point in the second non-reference image; or, the first weight value is a product of saturation, contrast and exposure of each pixel point in the first non-reference image; or, the first weight value is a product of saturation, contrast and exposure of each pixel point in the second non-reference image; the second weight value is the product of the saturation, the contrast and the exposure degree of each pixel point in the reference image.

The first proportion is the proportion of a first motion area in the first non-reference image, the second proportion is the proportion of a second motion area in the second non-reference image, and the target motion area comprises at least one connected area. In the field of image processing, a Connected Component generally refers to an image region composed of foreground pixels having the same pixel value and located adjacent to each other in an image.

Further, determining the target motion region according to a comparison result between the first proportion and a first threshold and a comparison result between the second proportion and the first threshold, including:

when the first proportion and the second proportion are not larger than the first threshold value, overlapping the first motion area and the second motion area, and determining the overlapped area as a target motion area; alternatively, the first and second electrodes may be,

when the first ratio is not larger than the first threshold value and the second ratio is larger than the first threshold value, determining that the first motion area is a target motion area; alternatively, the first and second electrodes may be,

and when the first ratio is larger than a first threshold value and the second ratio is not larger than the first threshold value, determining the second motion area as a target motion area.

Step 130, comparing the HDR image with a preset model to obtain a matching degree of the HDR image and the preset model.

In one embodiment, the preset model comprises keywords of standard answers of the test question parsing information; the comparing the HDR image with a preset model to obtain the matching degree of the HDR image and the preset model comprises:

acquiring character information in the HDR image; comparing the character information with the keywords; and grading the analysis information of the test questions according to the comparison result of the character information and the keywords.

Wherein the keyword may be a key score point. Therefore, the character information in the HDR image can be acquired through the image recognition technology, and then the character information is compared with the keywords, so that the recognition of character answers is realized, the scores are automatically obtained, and the student level evaluation and the scoring processing are facilitated.

And step 140, determining the score of the character answer according to the matching degree.

Specifically, a relation table between the matching degree and the score may be preset, and the relation table between the matching degree and the score may be as shown in table 1, where a represents the test question full score.

Degree of matching	Score of
		95％	10a
85％-95％	9a
		75％-85％	8a
65％-75％	7a
		……	……

TABLE 1

And 150, judging whether the filling area of the answer image is filled.

Before the step of judging whether the filling area of the answer image is filled, the method further comprises the following steps:

carrying out gray level processing on the machine-readable card answer image to generate a gray level image of the machine-readable card answer image; identifying anchor point information of the gray-scale image, correcting the gray-scale image according to the anchor point information, and generating a corrected image of the machine-readable card answer image; carrying out binarization processing on the corrected image to generate a binarized image of the machine-readable card answer image; and analyzing the concentration and the area of a filling area on the machine-readable card answer image to identify whether the filling area is filled or not.

Specifically, firstly, the machine-readable card answer image is converted into a gray scale image, which aims to remove the interference of variegated colors on the collected machine-readable card answer information, so that the machine-readable card image is as uniform and sharp as possible, and the gray scale image is converted in a plurality of specific implementation modes, such as an averaging method.

Secondly, the method of searching the outline by using the OPEN CV library can be used for identifying the anchor point information on the gray level image, wherein the four anchor points are the four corners of the machine-readable card respectively.

And thirdly, after acquiring the anchor point information, judging whether the anchor point information meets the specification. Judging whether the four anchor points meet the specification or not according to the sizes and the shapes of the four anchor points acquired by scanning; if the relative size and shape of the four anchor points are the same as those designed when the machine-readable card is manufactured, the four anchor points conform to the specification.

The gray scale image can then be corrected from the anchor point information using the principle of projective transformation. Therefore, the correction processing of the gray level image is realized. Firstly, judging whether the relative proportion of black and white stripes in the four anchor point images in the gray-scale image is in accordance with 1:1:3:1:1, if not, adjusting the gray-scale image by using a projection transformation principle to enable the relative proportion of the black and white stripes in the four anchor point images to be in accordance with the proportion of 1:1:3:1:1, namely completing the correction of the gray-scale image.

And then, carrying out binarization processing on the corrected gray level image, analyzing the concentration and the area of a filling area on the answer sheet according to the gray level image after binarization processing to identify whether the filling area is filled, and counting the score of the answer sheet according to an identification result. The binarization processing is to change the number of the colorimetric values of the corrected gray-scale image into two, such as black and white. The specific process is as follows: and comparing the gray value of each pixel point in the gray map with a preset gray threshold value, and if the gray value is greater than the preset gray threshold value, updating the gray value of the pixel point to a first preset gray value (such as black). And if the gray value is smaller than the preset gray threshold value, updating the gray value of the pixel point to a second preset gray value (such as white). In practical application, it is also possible to adopt further optimization measures, such as dividing the entire gray image into N small square formats (for example, the size of each grid is 50 × 50 pixels), then calculating the average gray value of the pixel points in each grid, taking the average gray value as a binarization threshold, setting a point with a gray value smaller than the threshold as a preset gray value, and setting a point with a gray value larger than the threshold as another preset gray value. Therefore, through binarization processing, an optimal binarization image can be obtained, the accuracy of subsequent filling and throwing area identification is improved, and the reliability of machine-readable card answer information identification is improved.

And step 160, determining the score of the machine-readable card answer according to the judgment result.

Specifically, the relationship between the determination result and the score of the machine-readable card answer may be as shown in table 2. In table 2, b represents the score of one choice question, and c represents the discount when the choice question is wrong.

Correct number of	Number of errors	Score of
			10	0	10b
9	1	9b+c
			8	2	8b+2c
7	3	7b+3c
			……	……

TABLE 2

The judgment result and the machine-readable card answer score may be calculated as shown in table 3, where d is the correct number in table 3.

TABLE 3

Fig. 2 is a schematic structural diagram of an apparatus for acquiring test question analysis information according to an embodiment of the present invention. As shown in fig. 2, the apparatus for acquiring the test question analysis information includes: an obtaining unit 210, a processing unit 220, a comparing unit 230, a determining unit 240 and a judging unit 250.

The obtaining unit 210 is configured to obtain a test question analysis information image; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image.

The processing unit 220 is configured to process the text answer image sequence to generate a first HDR image.

The comparing unit 230 is configured to compare the HDR image with a preset model.

The obtaining unit 210 is further configured to obtain a matching degree between the HDR image and a preset model according to a comparison result of the comparing unit.

The determining unit 240 is configured to determine a score of the text answer according to the matching degree.

The judging unit 250 is used for judging whether the filling area of the answer image is filled.

The determining unit 240 is further configured to determine a score of the machine-readable card answer according to the determination result.

Further, the text answer image sequence includes: a first non-reference picture, a reference picture, and a second non-reference picture; the exposure duration of the first test question of the first non-reference image, the reference image and the second non-reference image is sequentially increased.

Further, the preset model comprises keywords of standard answers of the test question analysis information;

the obtaining unit 210 is further configured to obtain text information in the HDR image;

the comparing unit 230 is further configured to compare the text information with the keyword;

the scoring unit 260 is configured to score the test question analysis information according to the comparison result between the text information and the keyword.

Further, the comparing unit 230 is further configured to compare the comparison result with a preset scoring standard;

the scoring unit 260 is further configured to score the analysis information of the test questions according to the comparison result of the comparing unit.

Further, the processing unit 220 is further configured to perform gray processing on the machine-readable card answer image to generate a gray image of the machine-readable card answer image;

the identification unit 270 is configured to identify anchor point information of the grayscale image, correct the grayscale image according to the anchor point information, and generate a corrected image of the machine-readable card answer image;

the processing unit 220 is further configured to perform binarization processing on the corrected image to generate a binarized image of the machine-readable card answer image;

the processing unit 220 is further configured to analyze, on the binarized image, the density and the area of a filling area on the machine-readable card answer image to identify whether the filling area is filled.

By applying the device for acquiring the test question analysis information provided by the embodiment of the invention, the acquisition unit acquires the test question analysis information image; the test question analysis information image comprises a character answer image sequence and a machine-readable card answer image; the processing unit processes the character answering image sequence to generate a first HDR image; the comparison unit compares the HDR image with a preset model; the obtaining unit obtains the matching degree of the HDR image and a preset model according to the comparison result of the comparison unit; the determining unit is used for determining the score of the character answer according to the matching degree; the judging unit is used for judging whether the filling area of the answer image is filled; the determining unit is also used for determining the score of the machine-readable card answer according to the judgment result. Therefore, the problems that in the prior art, the examination paper marking efficiency is low and the accuracy is influenced by subjective emotion are solved, and the accuracy and the efficiency of examination paper marking are improved.

Those of skill would further appreciate that the various illustrative components and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied in hardware, a software module executed by a processor, or a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The above embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, it should be understood that the above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the scope of the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims

1. A method for acquiring analysis information of test questions is characterized by comprising the following steps:

processing the character answer image sequence to generate a first HDR image;

determining the score of the character answer according to the matching degree;

judging whether the filling area of the answer image is filled;

2. The method for acquiring analysis information on test questions according to claim 1, wherein the sequence of the image of the text answers includes: a first non-reference picture, a reference picture, and a second non-reference picture; the exposure duration of the first test question of the first non-reference image, the reference image and the second non-reference image is sequentially increased.

3. The method for acquiring test question parsing information according to claim 1, wherein the preset model includes keywords of standard answers of the test question parsing information;

acquiring character information in the HDR image;

comparing the character information with the keywords;

4. The method for acquiring analysis information of test questions according to claim 1, wherein the step of scoring the analysis information of test questions according to the comparison result between the text information and the keywords comprises:

5. The method for acquiring analysis information on test questions according to claim 1, wherein before determining whether or not a filling area of the answer image is filled, the method further comprises:

6. An apparatus for obtaining analysis information of test questions, the apparatus comprising:

the comparison unit is used for comparing the HDR image with a preset model;

7. The apparatus for obtaining analysis information on test questions according to claim 6, wherein the sequence of images of the text answers includes: a first non-reference picture, a reference picture, and a second non-reference picture; the exposure duration of the first test question of the first non-reference image, the reference image and the second non-reference image is sequentially increased.

8. The apparatus for obtaining the analysis information of the test questions according to claim 6, wherein the preset model includes keywords of standard answers of the analysis information of the test questions;

9. The apparatus for obtaining analysis information on test questions according to claim 6, wherein the comparing unit is further configured to compare the comparison result with a preset scoring criterion;

10. The apparatus according to claim 6, wherein the processing unit is further configured to perform grayscale processing on the machine-readable card answer image to generate a grayscale image of the machine-readable card answer image;