JP6528789B2 - Numerical evaluation method, program and device - Google Patents

Description

  The present invention relates to a technology for recognizing a scoring result written on a test sheet and creating a scoring result table.

  In the in-class tests conducted by the teachers of each school, the teachers create questions and score and record the results of the tests for each class. In order to reduce the burden on teachers, it has been proposed to automate the recording of scoring results.

  The input position information of the character figure on the grading sheet is determined, the information of the character figure is recognized, the writing area for the marking sheet of the input information recognized according to the style information of the grading sheet is obtained, and the recognition result of the character figure and its writing area It has been proposed that, from information and scoring information indicated on the form information, the scoring results for the scoring sheet be added up and statistical processing be performed on each scoring summary value.

  In addition, when the calculation result of the recognition result of another field specified as a variable of the calculation formula defined by the form definition means is executed corresponding to the total field detected from the form, the character of the calculation result There is known a technique of comparing character strings of recognition results corresponding to a column and a total field, and indicating logical misreading when part or all of the characters are different.

JP-A-62-075578 JP 2005-18678 A

  However, in the above-mentioned technology, although the score written by handwriting in the frame of the test sheet is obtained by character recognition, it is necessary for the teacher to predetermine the field for entering the marking result on the test sheet from the computer. There is.

  Further, in order to detect logical misreading at the time of reading, the computer calculates whether the character recognition of the total column is correct or not by comparing with the total value of the character recognition results of the subtotal column. The test sheets used in elementary and junior high schools may change the position of the subtotal and total entry fields depending on the form. Therefore, it is time-consuming and labor-intensive for teachers to define the subtotal and total entry fields each time.

  If the entry field is not defined, the computer will be within the four rectangles even if there are score entry fields (for example, subtotal 1, subtotal 2, subtotal 3, and four rectangles for entering the total) on the test sheet. There are cases where it is not possible to recognize all the letters in the numbers, or because there are rectangles due to the graffiti etc. of the students other than the four rectangles, extra numbers are read and four figures can not be accurately obtained. was there.

  Therefore, the object of the present invention is to obtain the scoring result of the answer sheet.

According to an aspect of the present embodiment, a plurality of rectangular areas are specified from a predetermined area of an image by image recognition processing, and numerical values included in the specified plurality of rectangular areas are respectively extracted and extracted. And a predetermined arithmetic processing result for a plurality of other numerical values different from the extracted first numerical value, if the agreement is confirmed by the evaluation on the coincidence, the first numerical value A computer may execute the process of judging that the numerical value corresponds to a specific item, and the numerical value evaluation method may be performed.

  Further, as means for solving the above problems, it is possible to use a numerical value evaluation program, a numerical value evaluation device, and a recording medium recording the numerical value evaluation program.

  According to one aspect of the present embodiment, it is possible to obtain the scoring result of the answer sheet.

It is a figure which shows the hardware constitutions of a scoring result extraction apparatus. It is a figure which shows the function structural example of a grading result extraction apparatus. It is a figure (the 1) for explaining grading result extraction processing. It is a figure (the 2) for demonstrating grading result extraction processing. It is a figure (the 3) for demonstrating grading result extraction processing. It is a figure (the 4) for demonstrating grading result extraction processing. It is a figure which shows the example of a setting of a recognition area. It is a figure which shows the data structural example of a recognition result table. It is a figure which shows the data structural example of a value table. It is a figure which shows the example of a judgment result of whether the maximum value is a total value. It is a figure which shows the example of a total value and a subtotal value. It is a figure which shows the example of a data structure of a coordinate group table. It is a figure which shows the example of a result of integrated processing. It is a figure which shows the example of a result of an output process. It is a figure for demonstrating the effect concerning a present Example.

  Hereinafter, embodiments of the present invention will be described based on the drawings. In the present embodiment, for a test created by the teacher's own layout, a numerical value corresponding to each of the subtotal and the total is recognized and output from the test sheet on which the student's answer is scored by handwriting.

  The scoring result extraction device 100 according to the present embodiment has a hardware configuration as shown in FIG. FIG. 1 is a diagram showing a hardware configuration of a scoring result extraction device. In FIG. 1, the scoring result extraction device 100 is a terminal controlled by a computer, and is a CPU (Central Processing Unit) 11, a main storage device 12, an auxiliary storage device 13, an input device 14, and a display device 15. And an output device 16, a communication I / F (interface) 17, and a drive 18, which are connected to the bus B.

  The CPU 11 controls the scoring result extraction device 100 in accordance with a program stored in the main storage device 12. The main storage device 12 uses a RAM (Random Access Memory) or the like, and stores a program executed by the CPU 11, data necessary for processing by the CPU 11, data obtained by processing by the CPU 11, and the like. In addition, a partial area of the main storage device 12 is allocated as a work area used for processing by the CPU 11.

  The auxiliary storage device 13 uses a hard disk drive and stores data such as programs for executing various processes. A part of the program stored in the auxiliary storage device 13 is loaded to the main storage device 12 and executed by the CPU 11 to realize various processing. The storage unit 130 includes a main storage unit 12 and / or an auxiliary storage unit 13.

  The input device 14 includes a mouse, a keyboard, and the like, and is used by a user as a teacher to input various information necessary for processing by the scoring result extraction apparatus 100. The display device 15 displays various information necessary under the control of the CPU 11. The output device 16 has a printer or the like, and is used to output various information in accordance with an instruction from a teacher. The communication I / F 17 is an apparatus connected to, for example, the Internet, a LAN (Local Area Network), etc., to control communication with an external apparatus. Communication by the communication I / F 17 is not limited to wireless or wired.

  A program for realizing the processing performed by the scoring result extraction device 100 is provided to the scoring result extraction device 100 by a storage medium 19 such as a CD-ROM (Compact Disc Read-Only Memory), for example. That is, when the storage medium 19 storing the program is set in the drive 18, the drive 18 reads the program from the storage medium 19, and the read program is installed in the auxiliary storage device 13 via the bus B. . Then, when the program is started, the CPU 11 starts the processing according to the program installed in the auxiliary storage device 13. The medium for storing the program is not limited to the CD-ROM, and any medium readable by a computer may be used. As a computer readable storage medium, in addition to a CD-ROM, a portable recording medium such as a DVD disk and a USB memory, or a semiconductor memory such as a flash memory may be used.

  The scoring result extraction device 100 may further include a reading device 20 such as a scanner device. The reading device 20 is used to read a graded test sheet (answer sheet) for the student response of the teacher's own test and store it in the storage unit 130. In the present embodiment, the image of the graded test sheet may be stored in the storage unit 130, and the reading device 20 is not necessarily required.

  FIG. 2 is a diagram showing an example of a functional configuration of the scoring result extraction device 100. As shown in FIG. In FIG. 2, the scoring result extraction device 100 has a scoring result extraction unit 40 as a processing unit. The scoring result extraction unit 40 is realized by the CPU 11 executing a corresponding program, that is, the CPU 11 functions as the scoring result extraction unit 40. The same applies to each processing unit in the scoring result extraction unit 40.

  The storage unit 130 includes an image file 31 including a test paper image 6a obtained by reading a graded test sheet for each class, a recognition area setting value 32, a recognition result table 33, a value table 34, a coordinate group table 35, and a marking result. The table 36 and the like are stored.

  The scoring result extraction unit 40 recognizes the numerical value scored by the teacher from the subtotal or / and one or more areas in which the total exists in the test sheet, and determines the certainty that the recognized numeric value is the subtotal or total. Processing unit that outputs the scoring result. The scoring result extraction unit 40 includes a recognition unit 41, a determination unit 41-2, and an output unit 45.

  The recognition unit 41 reads the image file 31 stored in the storage unit 130, and recognizes the subtotal value and the total value from the image of the test paper (hereinafter, test paper image 6a) scored by handwriting by the teacher It is. The recognition unit 41 includes an area setting unit 41a, a rectangle recognition unit 41b, and a character recognition unit 41c.

  The area setting unit 41a causes the teacher to set one or more recognition areas in which the subtotal value or / and the total value exists, and stores the recognition area setting value 32 indicating the set area in the storage unit 130. The scores of subtotals and total scores by teachers are often shown on the top, bottom, left, and right sides of the test sheet. In the present embodiment, as described later, the teacher uses the user interface displayed on the display device 15 by the area setting unit 41a to one or more ends (upper end, lower end, left end, Alternatively, the width of the recognition area can be set from the right end).

  The rectangle recognition unit 41b refers to the recognition area setting value 32 stored in the storage unit 130, and recognizes a rectangular area in which the subtotal value or the total value is handwritten and entered, included in the recognition area of the test paper image 6a. Processing unit. The character recognition unit 41c is a processing unit that recognizes characters in the rectangular area recognized by the rectangular recognition unit 41b. In the present embodiment, numerical values are recognized as characters. The rectangle recognition unit 41 b and the character recognition unit 41 c are performed for each of the test paper images 6 a. A recognition result table 33 indicating the results of the rectangle recognition unit 41 b and the character recognition unit 41 c is stored in the storage unit 130.

  The determination unit 41-2 is a processing unit that determines whether or not the numerical value recognized from the test paper image 6a (answer paper) is a scoring result, and the value analysis unit 42, the coordinate analysis unit 43, and the integration unit 44. And.

  The value analysis unit 42 creates, in the storage unit 130, the value table 34 in which the value recognized for each test paper image 6a is one record from the recognition result table 33, and the value that can be the total value for each record of the value table 34 It is a processing unit to identify. In each record, it is assumed that the specified value is a total value, and the other values are subtotal values.

  The coordinate analysis unit 43 is a processing unit that refers to the recognition result table 33 and arranges the coordinates of a rectangle, which is a coordinate error range to be accepted as a group, into the same coordinate group with respect to all the records. The coordinate analysis unit 43 creates the coordinate group table 35 and stores the coordinate group table 35 in the storage unit 130.

  The integration unit 44 is a processing unit that specifies the coordinates at which the total value is written by integrating the analysis result obtained by the value analysis unit 42 and the analysis result obtained by the coordinate analysis unit 43. Further, by specifying the coordinates of the total value, the coordinates at which the subtotal value is written are also specified. If the coordinates at which the value assumed as the total value is detected belong to the same coordinate group with a certain probability, it is determined that the target coordinates are likely to be the coordinates of the total value. In addition, also in coordinates at which a value assumed as a subtotal value is detected, it is similarly determined that the coordinate of the subtotal value is likely to be within a certain probability range. If the coordinates at which the sum value is written do not exist in the same coordinate group, it is determined that the coordinates of the sum value do not exist.

  The output unit 45 is a processing unit that creates a grading result table 36 in which values having the same coordinate group for each student are set in the same column, and outputs the table 36 to the storage unit 130.

  The image file 31 is a data file storing a test paper image 6a obtained by scanning the test paper on which the answers of all the students who have undergone the test are scored. An image file 31 is created for each class and test. The image file 31 includes, for each class and test, a test paper image 6a that is scored for all students who have been tested.

  The recognition area setting value 32 indicates the area from the end of the test paper to the recognition of the rectangle in which the subtotal value and the total value are written. One or more regions of recognition are shown.

  The recognition result table 33 detects, for each sheet of the test paper image 6a stored in the image file 31, the value recognized in the area from the end of the sheet designated by the recognition area setting value 32 to the width, and the value It is a table which matched coordinates.

  The value table 34 is a table in which values are extracted one by one from the recognition result table 33. The coordinate group table 35 is a table obtained by acquiring the coordinates in order from the recognition result table 33 for each sheet of the test paper image 6a, and if they are the same as the coordinates of the previous image, they are collected as the same coordinate group. That is, the same coordinates are recorded in the same column for all the test sheet images 6a.

  The grading result table 36 is a table in which the subtotal values are shown as evaluation values per viewpoint, and each subtotal value is associated with the total value. When character recognition is performed from the test paper image 6a, the scoring result table 36 can be created by reading the student's attendance number and student name.

The process according to the present embodiment will be described with reference to FIGS. 3 to 6. 3 to 6 are diagrams for explaining the scoring result extraction process. The sequences and variables in FIGS. 3 to 6 are defined as follows.
· An array D (n, m) storing numeric data read by character recognition
(N = 1,...: Number of images of test paper to be subjected to extraction of scoring results (corresponding to the number of records))
(M = 1,...: Number of rectangles detected in the test paper image 6a (corresponding to the number of detected values))
· An array Z (n, m) storing the coordinates of the detected rectangle
(N = 1,...: Number of images of test sheets to be targeted (corresponding to the number of records))
(M = 1,...: Number of characters recognized in the test paper image 6a (corresponding to the number of detected values))
It is.
· Number of rectangles extracted M_n: Number of rectangles extracted in the nth test paper image 6a (hereinafter referred to as the number of rectangles)
Maximum value of the value D_max: Maximum value in the array D (n, m) of the numbers recognized in the n-th test paper image 6a Sum of the specified array of numbers D_SUM
An array MarkedMax (x) holding the place where the total point is detected
・ Number of records in which total points were detected x_max
・ Proportion of judging whether total points are statistically detected sufficiently
・ Coordinate error range to decide whether to include in the same coordinate group ρ 2
・ Coordinate group TmpZ (Y)
・ Number of coordinate groups y_max
・ A variable c that counts the number of coordinates close to the same coordinate group
・ Proportion of the coordinates of the total point (hereinafter, the total point coordinates) exist in the same coordinate group (a value close to 100)
z
In FIG. 4, the scoring result extraction unit 40 initializes the index number x of the number array in which the total score is detected (step S20).

  In FIG. 3, in the scoring result extraction unit 40, the recognition unit 41 displays a user interface on which the teacher can specify the path of the image file 31 on the display device 15, and acquires the path of the image file 31 from the teacher 31 is read from the storage unit 130 (step S11).

  In the recognition unit 41, the area setting unit 41a displays on the display device 15 a user interface that allows the teacher to specify the area for rectangle recognition, and acquires the area for rectangle recognition from the teacher (step S12). The area to be recognized as a rectangle corresponds to the width from one or more ends in the top, bottom, left, and right of the test sheet.

  The recognition unit 41 reads the test paper image 6a from the image file 31 for each count of n = 1, 2,..., N based on the number N of images of the test paper image 6a, and executes the following step S15. The rectangle recognition process is repeated (step S13: loop A).

  The recognition unit 41 reads the nth test paper image 6a from the image file 31 (step S14). Then, the recognition unit 41 extracts a rectangle from the area designated as the rectangle recognition target by the rectangle recognition unit 41 b for the n-th test paper image 6 a (step S 15). The rectangle recognition unit 41b corresponds to step S15.

  The rectangle recognition unit 41b inputs the test paper image 6a and the at least one rectangle recognition target area, executes rectangle recognition processing, and as a result, outputs the recognized coordinates Z (n, m) and the number M of rectangles Do. Coordinates Z (n, m) include coordinate information of the lower left and upper right of the rectangle. The number of rectangles M_n indicates the number of rectangles detected from the rectangle recognition target area.

  The recognition unit 41 designates coordinates Z (n, m) in the order of m = 1, 2,..., M_n based on the coordinates Z (n, m) and the number M of rectangles obtained by the rectangle recognition unit 41 b. Then, the character recognition unit 41c repeatedly performs the character recognition processing for reading the handwriting in the rectangle (step S16: loop B).

  The character recognition unit 41 c digitizes the number image in the rectangle specified by the coordinates Z (n, m) detected by the rectangle recognition unit 41 b by the character recognition technology (step S 17). A number (number) is read from a number image. The character recognition unit 41c corresponds to step S17.

  The character recognition unit 41c inputs the test paper image 6a and at least one coordinate Z (n, m), executes character recognition processing, and outputs an array D (n, m) of recognized numbers as a result thereof Do.

  When the recognition unit 41 does not finish the character recognition process for all the rectangles detected in each test paper image 6a, the process returns to step S16, increments m, and repeats the above-described process. Then, the loop B for performing the character recognition process on the test paper image 6a is ended (step S18).

  When the character recognition process is completed for all the rectangles detected in each test paper image 6a, the recognition unit 41 returns to step S13 when the rectangle recognition process is not completed for all the test paper images 6a. , N are repeated to repeat the above-described process, and when the process is completed, the loop A for performing the rectangular recognition process and the character recognition process on all the test paper images 6a of the image file 31 is ended (step S19). After completing the processing in the recognition unit 41, the scoring result extraction unit 40 proceeds to step S20 in FIG.

  In the scoring result extraction unit 40, the value analysis unit 42 detects a total score from a plurality of values for each count of n = 1, 2,..., N based on the number N of images of the test paper image 6a. The value analysis process is repeatedly performed (step S21: loop C).

  The value analysis unit 42 extracts the maximum value in the numeral array D (n, m) (m = 1, 2,... M_n) related to the n-th test paper image 6a (step S22). The value analysis unit 42 inputs the numeral array D (n, m) of the n-th test paper image 6a, performs value analysis processing, and calculates the maximum value D_max, the extraction number c_max of the maximum values, and the array number m_max of the maximum values. Output

  The value analysis unit 42 determines whether the extraction number c_max of the maximum value is 1 (step S23). If it is not 1, the value analysis unit 42 proceeds to step S28 where it determines the end of the loop C.

  On the other hand, in the case of 1, the value analysis unit 42 obtains the sum of the number array D (n, m) of the n-th test paper image 6a (step S24). The value analysis unit 42 receives the numerical array D (n, m) of the n-th test paper image 6a, sums up the numerical values shown in each array, and outputs a data sum D_SUM indicating the summed value.

  Then, the value analysis unit 42 determines whether the value obtained by dividing the data sum D_SUM including the maximum value into two equal parts is equal to D_max (step S25). In step S25, it is determined whether the maximum value and the sum of other numerical values are the same. If they are the same, it is determined that the n-th record indicates a total score.

  If it is determined in step S25 that the values are not equal, the value analysis unit 42 proceeds to step S28 in which the end of loop C is determined. On the other hand, if it is determined that they are equal, the value analysis unit 42 substitutes x + 1 into the index number x of the numeral array. One is added to the index number x (step S26).

  The value analysis unit 42 stores the index location of the number array from which the total point (maximum value) is extracted (step S27). The array MarkedMax (x) is set with the n-th and maximum value array number m_max (MarkedMax (x) = m_max).

  Then, when the value analysis processing has not been completed for all the test paper images 6a, the value analysis unit 42 returns to step S21, increments n to repeat the above-described processing, and ends the test. The value analysis process for each sheet image 6a is ended (step S28).

  The value analysis unit 42 substitutes the number of indexes x into the number of records x_max in which the total score is detected (step S29).

  The value analysis unit 42 determines whether the number of records x_max in which the total score is detected is が 1% or more of the number of read records N (that is, the number of test paper images 6a in the image file 31) (step S30) ). When it is less than 1% 1%, the scoring result extraction unit 40 outputs that the place of the total point can not be determined by the output unit 45 (step S30-2 in FIG. 6), and ends the scoring result extraction process.

  On the other hand, when it is 1% 1% or more, the scoring result extraction unit 40 proceeds to step S31 in FIG. In FIG. 5, the scoring result extraction unit 40 sets the coordinate group TmpZ (on the condition that the coordinate analysis unit 43 exists near coordinates Z (n, m) (m = 1, 2,..., M_n). y) is classified (step S31). The coordinate analysis unit 43 inputs the array Z (n, m) of rectangular coordinates and the coordinate error range 22 to be accepted as a group, performs coordinate analysis processing, coordinates the coordinate group TmpZ (y), and the number y of coordinate groups Output

  Thereafter, the scoring result extraction unit 40 performs integration processing by the integration unit 44. The integration unit 44 repeatedly performs the process described below for each count of y = 1, 2,..., Y_max based on the number y_max of coordinate groups (step S32: loop D).

  The integration unit 44 substitutes 0 (zero) into the variable c (step S33). Then, the integration unit 44 repeats the process described below for each count of x = 1, 2,..., X_max based on the number of records x_max in which the total score is detected (step S34: loop E).

  The integration unit 44 determines whether the coordinate Z (n, m_max) with respect to MarkedMax (x) indicating the position where the maximum value is detected is sufficiently close to the coordinate TmpZ (y) of the coordinate group and returns the result (step S35 ). The integration unit 44 inputs Z (n, m_max) where the MarkedMax (x) is set, the coordinate group TmpZ (y), and the coordinate error range 2 2 and determines whether the coordinates are sufficiently close or not. As a result, it outputs NearTF indicating that the coordinates are close enough.

  Thereafter, the integration unit 44 determines whether NearTF is true (step S36). If NearTF is false, the process proceeds to step S38. On the other hand, if NearTF is true, c + 1 is substituted for the coordinate number c close to the same coordinate group (step S37). The coordinate number c is incremented by one. The number of coordinates close to the same coordinate group is counted.

  The integration unit 44 determines whether x has reached the number of records x_max in which the total point has been detected, and if it has not reached, the process returns to step S34, performs the processing in loop E described above, and has reached , Loop E is finished, and the process proceeds to step S39.

  When the loop E is ended, the integration unit 44 determines whether c / x_max is equal to or more than a ratio z% of total point coordinates existing in the same coordinate group (step S39). The ratio of the number of records in which the total point is detected to the number c of coordinates close to the same coordinate group counted is compared, and if it is a predetermined ratio or more, TmpZ (y) is the coordinate where the total point is written It is determined that

  If it is determined that the ratio is less than z%, the integration unit 44 determines whether y has reached the number y_max of coordinate groups, and if it has not, the process returns to step S32, and the loop D described above is performed. The processing is performed, and if reached, the loop D is ended. The processing in the integration unit 44 ends. In this case, the scoring result extraction unit 40 proceeds to step S30-2 in FIG. 6, outputs that the place of the total score can not be determined, and ends the scoring result extraction processing.

  On the other hand, when it is determined that the ratio is z% or more, the scoring result extraction unit 40 proceeds to step S41 in FIG. In FIG. 6, the scoring result extraction unit 40 causes the output unit 45 to perform the processes of steps S41 to S43.

  The output unit 45 determines that the TmpZ (y) coordinates and the MarkedMax (x) = Z (n, m_max) coordinates are coordinates to which the total value is input (step S41). This coordinate is taken as Z_max.

  Further, the output unit 45 determines that the coordinates of the same record other than the coordinate Z (n, m_max) corresponding to MarkedMax (x) are the coordinates of the subtotal value (step S42). Let these coordinates be Z_1, Z_2,.

  Then, the output unit 45 creates a grading result table 36 in which each value corresponding to the Z_max coordinates of the total value and the Z_1, Z_2, ... coordinates of the subtotal values is associated with each student, and is stored as a storage unit. It outputs to 130 (step S43). The output unit 45 creates a scoring result table 36 in association with each student for each class. In association with each student, a database for managing a student ID (for example, an attendance number) for specifying the student, a student name and the like is created in advance in the storage unit 130, and the scoring result is associated with the student. At the end of the process by the output unit 45, the scoring result extraction unit 40 ends the scoring result extraction process.

  Below, the example in each process part is shown. FIG. 7 is a diagram showing an example of setting of a recognition area. In FIG. 7, the width from each end of the recognition area is defined for the test paper image 6 a by the area setting unit 41 a. Each width 7a, 7b, 7c and 7d can be changed by the teacher. As a setting value, with respect to the widths from the upper and lower ends, the ratio to the length in the vertical direction of the test paper image 6a may be indicated. Similarly, with respect to the widths from the left and right ends, the ratio to the horizontal length of the test paper image 6a may be shown.

  The recognition area 8a is an area defined by the width 7a, the recognition area 8b is an area defined by the width 7b, the recognition area 8c is an area defined by the width 7c, and the recognition area 8d is defined by the width 7d Area.

  The teacher sets the recognition area on the test paper image 6a simply by clicking and selecting one or more of the upper and lower recognition areas 8a to 8d with a mouse or the like. The example of FIG. 7 shows the case where only the upper recognition area 8a is selected and the case where only the upper recognition area 8a and the lower recognition area 8c are selected. You may set three places or all four places as a recognition area including right and left. A recognition area setting value 32 indicating a recognition area selected by the teacher is stored in the storage unit 130. Hereinafter, the recognition areas 8a to 8d are collectively referred to simply as the recognition area 8.

  The recognition result table 33 obtained by the rectangle recognition unit 41b and the character recognition unit 41c in steps S13 to S19 in FIG. 3 will be described with reference to FIG. FIG. 8 is a view showing an example of the data configuration of the recognition result table. The recognition result table 33 illustrated in FIG. , Value detection No. , Items, and values such as rectangular coordinates.

  Record No. Is a number for identifying each test paper image 6a. A unique number is given to each test paper image 6a. Multiple identical record numbers Indicates that a plurality of rectangles and numbers have been recognized from the same test paper image 6a.

  Value detection No. Indicates a serial number in the order in which numbers are recognized in a rectangle of each test paper image 6a.

  The values are managed in array D (n, m). Also, rectangular coordinates are managed by the array Z (n, m). The array Z (n, m) indicates the coordinates of the lower left corner and upper right corner on the diagonal of each rectangle. In the values and rectangular coordinates, n indicates the number of images (record No.), and m indicates the number of recognized characters (the number of detected values).

  In this example, the record No. In the test paper image 6 a of “1”, it shows that three rectangles from the recognition area 8 and that values are recognized from within each rectangle. Record No. The test paper image 6 a of “2” indicates that four rectangles from the recognition area 8 and that values are recognized from within each rectangle.

  Regarding value analysis processing by the value analysis unit 42 in steps S21 to S28 in FIG. 4, an example data configuration of the value table 34 will be described with reference to FIGS. 9, 10, and 11. FIG. 9 is a diagram showing an example of the data configuration of the value table. The value table 34 exemplified in FIG. It is a table in which values recognized in one record are associated with each other.

  In FIG. 9, the record No. of the recognition result table 33 is the first record. D (1,1), D (1,2), and D (1,3) indicating the value of “1” are associated with each other. In the second record, the record No. of the recognition result table 33 is displayed. D (2,1), D (2,2), D (2,3), and D (2,4) indicating the value of “2” are associated with each other. The same applies to other records.

  In the data configuration example of FIG. 9, the value analysis unit 42 extracts the maximum value from each record. The array D indicating the maximum value is shown shaded. In the first record, D (1, 3), in the second record, D (2, 3), in the third record, D (3, 1), ... are the maximum values.

  FIG. 10 is a diagram showing an example of the determination result as to whether or not the maximum value is a total value. The value analysis unit 42 extracts the maximum value for each record using the value table 34 and determines whether or not it is a total point. In the determination result, the index location indicating the maximum value determined to be the total point is stored.

  FIG. 10 shows an example of the determination result for each record. In the first record, D (1, 3) indicating the maximum value does not match the sum of the values of D (1, 1) and D (1, 2). In the second record, D (2,3) indicating the maximum value matches the sum of the values of D (2,1), D (2,2) and D (2,4). In the third record, D (3,1) indicating the maximum value does not match the sum of the values of D (3,2) and D (3,3). Thus, a determination is made for each record whether the maximum value matches the sum of the other values.

  FIG. 11 is a diagram illustrating an example of the total value and the subtotal value. In the record in which it is assumed that the maximum value is the total value using the value table 34 of FIG. 10, FIG. 11 shows an example in which the total value and the subtotal value are marked, but at least the total value can be marked. Just do it.

  In the example of FIG. 11, the index location of the array assumed to be a sum value is stored in MarkedMax (x). The first record is D (1, 3), the second record is D (2, 3), the third record is D (3, 1), the fourth record is D (4, 3), ... The value of · is the maximum value determined to be a total point, and is stored in MarkedMax (x). Also, if it is determined to be a total point, it can be assumed that other values of the same record are subtotal values.

  Regarding the coordinate analysis processing by the coordinate analysis unit 43 in step S31 of FIG. 5, an example of the data configuration of the coordinate group table 35 will be described with reference to FIG. FIG. 12 is a diagram showing an example of the data configuration of the coordinate group table.

  In FIG. 12, first, the coordinate analysis unit 43 refers to the recognition result table 33 shown in FIG. The rectangular coordinates of “1” are detected as the value detection No. They are arranged in the coordinate group table 35 in the order of. And, the next record No. The rectangular coordinates of “2” are value detection No. When arranged in the order of the coordinate group table 35 in the case of the same coordinates as the previous coordinates, they are put together as a group of the same coordinates. If the previous record does not have the same coordinates, a new coordinate group is created in the coordinate group table 35.

  The same coordinates indicate the coordinates within the coordinate error range ρ2 for determining whether or not to be included in the same coordinate group.

  In the coordinate group table 35 illustrated in FIG. 12, four coordinate groups of TmpZ (1), TmpZ (2), TmpZ (3), and TmpZ (4) are created.

  An example of the result of the integration process by the integration unit 44 in steps S32 to S40 in FIG. 5 will be described with reference to FIG. FIG. 13 is a diagram showing an example of the result of the integration process. When the coordinates of a rectangle whose maximum value is assumed to be a total value are included in a coordinate group having a predetermined ratio or more, the integration unit 44 identifies the coordinate group as coordinates of the total value. In the coordinate group table 35 illustrated in FIG. 13, TmpZ (3) is specified as the coordinates of the total value.

  By using the tables 33 to 35 illustrated in FIG. 8 to FIG. 13, records in which the number of character-recognized numerical values is insufficient, records in which the number is excessive, and the total of numerical values are incorrect. Records can be identified.

  An example of the result of the output process by the output unit 45 in steps S41 to S43 of FIG. 5 will be described with reference to FIG. FIG. 14 is a diagram showing an example of the result of the output process. The scoring result table 36 illustrated in FIG. 14 is a table that is output and stored in the storage unit 130 as a result of the output process by the output unit 45.

  In FIG. 14, the scoring result table 36 has items such as an attendance number, a name, a viewpoint 1, a viewpoint 2, a viewpoint 3, and a total. For the sum, values corresponding to the coordinates included in the coordinate group specified as the coordinates of the total value are acquired from the recognition result table 33 and set. The viewpoints 1 to 3 correspond to subtotals. A value corresponding to the coordinate determined to be the coordinate of the subtotal value is acquired from the recognition result table 33 and set in the scoring result table 36.

  FIG. 15 is a figure for demonstrating the effect concerning a present Example. In the test paper image 6a illustrated in FIG. 15, one recognition area 8-2 is designated. In the scoring result extraction process according to the present embodiment, three rectangles 51a, 52a and 53a are recognized and detected in the recognition area 8-2, and the coordinates of each rectangle are stored in the array Z. When the rectangle is recognized, processing in the area other than the recognition area 8-2 is not performed.

  Numbers are detected from the number images in the areas of the detected rectangles 51a, 52a and 53a, respectively, and stored in the array D. Three values "40", "35" and "75" are detected and stored in array D. Since numbers are detected only from the rectangles 51a, 52a and 53a in the recognition area 8-2, processing for numbers "8", "22", "10" etc. other than the recognition area 8-2 is suppressed, and the processing time is Can be reduced. Moreover, the setting of the recognition area to be processed may be performed only by specifying an area including one or more rectangles to be detected instead of each rectangle, and the setting can be performed concisely even when the teacher sets it. .

  In this embodiment, numbers are detected in the set recognition area 8-2 in each test paper image 6a, and the probability of the position (coordinates) of the total value is statistically confirmed from the detected numbers. Therefore, the teacher does not have to set a value type such as whether the subtotal value is to be written or whether the total value is to be written for each rectangle.

  As described above, according to the present embodiment, the marking result in the predetermined area of the test sheet image obtained by reading the marked test sheet is recognized with a small burden on the teacher, and the total value is indicated among the marking results. A scoring result table can be created automatically to determine the numbers statistically.

  The present invention is not limited to the specifically disclosed embodiments, and various modifications and changes are possible without departing from the scope of the claims.

Further, the following appendices will be disclosed regarding the embodiment including the above-described example.
(Supplementary Note 1)
Image recognition processing of the answer sheet to extract one or more numerical information,
It is determined that numerical value information indicating the maximum value among the extracted one or more numerical value information is the scoring result described on the answer sheet.
A scoring result acquisition method characterized by
(Supplementary Note 2)
The one or more numerical information is numerical information recognized in a predetermined area of the answer sheet,
The scoring result acquisition method according to claim 1, characterized in that:
(Supplementary Note 3)
A numerical value obtained by the character recognition based on a recognition result table stored in a storage unit, in which coordinates of the numeral image extracted in the predetermined area of the answer sheet are associated with a numerical value recognized from the numeral image. It is assumed that the numerical value indicating the maximum value of
Create a coordinate group in which the coordinates of the numeric image are combined within a predetermined range among a plurality of the answer sheets,
In the case where the coordinates of the numerical value assumed to have the possibility of the total value belong to the same coordinate group at a predetermined ratio or more, it is determined that the numerical value is a total value, and the scoring result acquisition method according to supplementary note 2 .
(Supplementary Note 4)
Displaying on the display device a user interface that allows the user to set one or more of the predetermined areas defined by the width from the upper, lower, left, and right ends of the answer sheet;
The recognition result table according to appendix 3, characterized in that a recognition result table in which the coordinates of the numeral image extracted in the predetermined area are associated with the numerical value recognized from the numeral image is stored in the storage unit. Acquisition method.
(Supplementary Note 5)
Recognize a rectangle within the predetermined area, and set two diagonal vertices of the rectangle as coordinates of the numeric image,
The marking result acquisition method according to claim 4, wherein the numerical value is character-recognized in the rectangle specified by the coordinates.
(Supplementary Note 6)
It is determined that numerical values other than those determined to be the total value are subtotal values,
The scoring result acquisition method according to any one of appendices 3 to 5, wherein a result table indicating numerical values corresponding to each item of the subtotal value and the total value is output.
(Appendix 7)
It is determined whether the number of numerical values assumed to have the possibility of the total value is greater than or equal to a ratio of statistical judgment criteria with respect to the number of sheets of the answer sheet.
The scoring result acquisition method according to any one of appendices 3 to 6, wherein creation of the coordinate group is enabled when the ratio is equal to or more than the ratio of the statistical determination standard.
(Supplementary Note 8)
Image recognition processing of the answer sheet to extract one or more numerical information,
A marking result acquisition program which causes a computer to execute processing of determining numerical value information indicating the maximum value among the extracted one or more pieces of numerical information as the marking result described on the answer sheet.
(Appendix 9)
A recognition unit that performs image recognition processing on the answer sheet to extract one or more numerical information;
The marking result acquisition device which has a judgment part which judges the numerical value information which shows the maximum value among the extracted one or a plurality of numerical information as the marking result indicated on the above-mentioned answer sheet.

6a test paper image 8a, 8b, 8c, 8d recognition area 11 CPU
12 main storage device 13 auxiliary storage device 14 input device 15 display device 16 output device 17 communication I / F
18 drive 19 storage medium 20 reader 31 image file 32 recognition area setting value 33 recognition result table 34 value table 35 coordinate group table 36 grading result table 40 grading result extraction unit 41 recognition unit 41a area setting unit 41b rectangle recognition unit 41c character recognition Part 42 Value analysis part 43 Coordinate analysis part 44 Integration part 45 Output part 100 Scoring result extraction device 130 Storage part

Claims (5)

Identify a plurality of rectangular areas from a predetermined area of the image by image recognition processing,
Extracting the numerical values contained in the plurality of specified rectangular areas,
First and numerical extracted, with the predetermined arithmetic processing results for different other multiple numbers and extracted the first numerical value, if a match of the implementation of the evaluation of matching has been confirmed, Determining that the first numerical value corresponds to a specific item ,
A numerical evaluation method characterized in that a computer executes processing.   The numerical value evaluation method according to claim 1, wherein the predetermined area is an area included in a predetermined range from at least one end of upper, lower, left, and right of the image. The predetermined arithmetic processing is calculation processing of a total value,
The numerical evaluation method according to claim 1 or 2, characterized in that: Identify a plurality of rectangular areas from a predetermined area of the image by image recognition processing,
Extracting the numerical values contained in the plurality of specified rectangular areas,
First and numerical extracted, with the predetermined arithmetic processing results for different other multiple numbers and extracted the first numerical value, if a match of the implementation of the evaluation of matching has been confirmed, Determining that the first numerical value corresponds to a specific item ,
A numerical evaluation program that causes a computer to execute a process. An extraction unit which identifies a plurality of rectangular areas from a predetermined area of the image by image recognition processing, and extracts numerical values included in the plurality of identified rectangular areas respectively;
First and numerical extracted, with the predetermined arithmetic processing results for different other multiple numbers and extracted the first numerical value, if a match of the implementation of the evaluation of matching has been confirmed, An analysis unit that determines that the first numerical value corresponds to a specific item ;
The numerical evaluation device characterized by having.

Images