JP2012033124A - Ballot paper sorting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a ballot paper sorting device that can automatically judge whether or not the type of election stated on a ballot paper is correct.SOLUTION: A ballot paper sorting device comprises an imaging unit 5 that performs imaging of a ballot paper and generates image information I, a region extracting section 9 that extracts an election type region R that includes information representing the type of election from the image information I, an evaluated character string generating section 10 that generates an evaluated character string Se from the election type region R, a reference character string generating section 11 that generates a reference character string Sr from the election type region R in a ballot paper for practice use provided in advance, a memory section 12 that stores the reference character string Sr, and a judging section 13 that judges whether or not the type of election stated on a ballot paper is correct on the basis of whether or not the evaluated character string Se and the reference character string Sr are identical.

Description

  The present invention relates to a ballot paper classification device.

  2. Description of the Related Art A voting paper classification apparatus that classifies voting paper by reading a candidate name or the like entered on an election ballot is known (for example, Patent Document 1). In such a ballot paper classification device, a ballot entry surface (a surface on which a voter enters a candidate name, etc.) is imaged by a scanner or the like to generate image information, and the candidate name is entered from this image information. Read the contents and classify the ballots based on the reading results.

Japanese Patent Laid-Open No. 2004-5022

  In recent years, there are cases where a plurality of elections are held on the same schedule and at the same venue. In such cases, different types of ballots are used for each type of election, such as “○ △ Prefectural Council Election”, “□ × City Council Election”, etc. Therefore, it is necessary for the person in charge of the operation to visually check whether the election type of the ballot is correct, which is a heavy burden.

  The present invention has been made to solve such a problem, and an object of the present invention is to provide a ballot paper classification device that can automatically determine whether the election type of a ballot is correct.

In order to solve the above problems, a ballot paper classification device according to the present invention includes: an image pickup unit that picks up a ballot paper and generates image information in the ballot paper classification device that classifies the ballot paper based on the entry content; An area extracting means for extracting an election type area including information representing an election type from the image information generated by the imaging means, an evaluation information generating means for generating evaluation information from the election type area extracted by the area extracting means, Reference information generating means for generating reference information from the election type area in the image information of a given correct election type ballot, storage means for storing the reference information generated by the reference information generating means, and a voting paper to be determined For voting to be determined based on the evaluation information generated from the election type area in the image information of the image and the reference information stored in the storage means Wherein the of and a determining means for determining whether the election type is correct.
Thereby, it can be automatically determined whether or not the election type of the ballot is correct.

  The evaluation information is the first character string extracted from the election type area in the image information of the ballot paper to be determined, and the reference information is from the election type area in the image information of the ballot paper of the correct election type given in advance. Whether or not the election type of the ballot to be determined is correct based on whether or not the first character string and the second character string match. May be determined.

  The evaluation information is a set of pixel values of the election type area in the image information of the ballot paper to be determined, and the reference information is a set of pixel values of the election type area in the image information of the correct election type ballot given in advance. The determination means determines whether the election type of the vote ballot to be determined is based on the result of applying the mask pattern to the set of pixel values generated from the vote ballot to be determined. You may determine whether it is correct.

  According to the ballot paper classification device according to the present invention, it is possible to automatically determine whether or not the election type of the ballot paper is correct.

It is a figure which shows the structure of the ballot paper classification | category apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows the internal structure of the ballot paper classification device based on Embodiment 1 of this invention. It is a flowchart which shows the process of learning mode in the ballot paper classification device concerning Embodiment 1 of this invention. It is a figure which shows an example of the image information produced | generated from the ballot for learning in the ballot paper classification device concerning Embodiment 1 of this invention. It is a flowchart which shows the process of the determination mode in the ballot paper classification device based on Embodiment 1 of this invention. It is a figure which shows an example of the image information produced | generated from the ballot paper of determination object in the ballot paper classification device concerning Embodiment 1 of this invention. It is a schematic diagram which shows the internal structure of the ballot paper classification device based on Embodiment 2 of this invention. It is a flowchart which shows the process of learning mode in the ballot paper classification device concerning Embodiment 2 of this invention. It is a figure which shows an example of the process in which a mask pattern is produced | generated from the ballot for learning in the ballot paper classification device based on Embodiment 2 of this invention. It is a flowchart which shows the process of the determination mode in the ballot paper classification device concerning Embodiment 2 of this invention. It is a figure which shows the example of the result of having applied the mask pattern with respect to the array of the evaluation pixel value produced | generated from the ballot paper of determination object in the ballot paper classification device concerning Embodiment 2 of this invention.

Embodiments of the present invention will be described below with reference to the accompanying drawings.
Embodiment 1 FIG.
A voting paper classification device according to Embodiment 1 of the present invention will be described with reference to FIGS.
The ballot paper classification device classifies ballot papers based on the contents of entries such as candidate names entered on the ballot paper. As shown in FIG. 1, the main body 1 of the voting paper classification apparatus includes a hopper 3 that stores a plurality of voting papers 2 to be classified and a plurality of rollers 4 a, and a conveyance path 4 that conveys the voting paper 2. An imaging unit 5 that is configured by a scanner or the like and that captures an entry surface of the voting form 2 to generate image information I, and five stackers 6 that classify and store the voting form 2 based on the entry contents; And a monitor 7 for displaying the classification result.

  In addition, as schematically shown in FIG. 2, a calculation process for determining whether the election type of the voting paper is correct based on the image information I generated by the imaging unit 5 is included in the voting paper classification device. A unit 8 is provided. The arithmetic processing unit 8 includes an area extraction unit 9, an evaluation character string generation unit 10, a reference character string generation unit 11, a storage unit 12, and a determination unit 13.

  The area extraction unit 9 obtains an election type area R, which is an area including information (character string) representing an election type such as “O △ Prefectural Council Election” from the image information I of the ballot paper generated by the imaging unit 5. Extract.

  The evaluation character string generation unit 10 performs character recognition processing on the election type area R in the ballot paper to be determined, thereby evaluating character as evaluation information for determining whether the election type of the ballot paper is correct A column Se is generated.

  The reference character string generation unit 11 determines whether the election type of the ballot is correct by performing character recognition processing on the election type region R in the ballot of the correct election type for learning described later. A reference character string Sr serving as reference information is generated.

  The storage unit 12 stores the reference character string Sr generated by the reference character string generation unit 11.

  Based on the comparison result between the evaluation character string Se generated from the determination target ballot and the reference character string Sr stored in the storage unit 12, the determination unit 13 has the correct election type of the determination target ballot. It is determined whether or not.

  Next, a process for determining whether or not the election type of the voting paper is correct in the voting paper classification device according to the first embodiment will be described.

  The ballot paper classification device has two types of processing modes, a “learning mode” and a “determination mode”. In the learning mode, one voting paper of the correct election type is read as a voting paper for learning, and its characteristics are stored as reference information. In the judgment mode, evaluation information is generated from each of a plurality of ballots filled in by the voter, and whether the election type of each ballot is correct based on the evaluation information and the reference information stored in the learning mode. Determine whether or not. Hereinafter, details of the processing in the learning mode and the determination mode will be described in order.

(Learning mode)
In order to start the processing in the learning mode, the user takes one ballot of the correct election type for learning and places it on the hopper 3, and instructs the ballot paper classification device to start the learning mode by an operation switch (not shown). To do.

  When the ballot paper classification device is instructed to start the learning mode, the ballot paper for learning is conveyed to the imaging unit 5 by driving each roller 4a constituting the conveyance path 4, and the process shown in the flowchart of FIG. I do.

  First, the imaging unit 5 captures the entry surface of the voting form, generates binarized image information I as shown in FIG. 4, and outputs it to the arithmetic processing unit 8 (S101).

  The region extraction unit 9 of the arithmetic processing unit 8 extracts the election type region R including the character string representing the election type from the image information I, and outputs it to the reference character string generation unit 11 (S102). In addition, the coordinates of the election type region R are stored for use in later determination modes. In the example shown in FIG. 4, a rectangular area surrounded by a broken line including the character string “◯ △ Prefectural Assembly Election Vote” is the election type area R. As an extraction method of the election type region R, a character recognition process may be performed on the entire image information I to specify a region including a character string representing the election type, or the image information I may be displayed on the monitor 7. It may be displayed and the user may designate the area.

  The reference character string generation unit 11 performs character recognition processing on the election type region R extracted by the region extraction unit 9, thereby extracting a character string included in the election type region R (S103). The reference character string Sr is stored in the storage unit 12 (S104). In the example shown in FIG. 4, a character string “◯ △ Prefectural Assembly Election Vote” is extracted and stored in the storage unit 12 as a reference character string Sr.

  Through the above processing performed in the learning mode, a character string representing the election type printed on the learning ballot is extracted and stored in the storage unit 12 as the reference character string Sr.

(Judgment mode)
When the process in the learning mode is completed, the user places a plurality of ballot sheets filled in by the voter on the hopper 3 and instructs the ballot sheet classification apparatus to start the determination mode.

  When the start of the determination mode is instructed, the ballot paper classification device drives each roller 4a constituting the conveyance path 4 to convey the ballots to be determined one by one to the imaging unit 5, and for each ballot paper The processing shown in the flowchart of FIG.

  First, the imaging unit 5 captures the entry surface of the ballot paper to generate binarized image information I as shown in FIG. 6 (S111), and the area extraction unit 9 uses the coordinates stored in the learning mode. Based on the image information I, the election type region R is extracted and output to the evaluation character string generation unit 10 (S112).

  The evaluation character string generation unit 10 performs character recognition processing on the election type region R to extract a character string included in the election type region R (S113), and sets this as the evaluation character string Se (S113). S114).

  The determination unit 13 checks whether or not the evaluation character string Se set in step S114 matches the reference character string Sr stored in the storage unit 12 in the learning mode (S115). Determines that the election type of the ballot is correct (S116), and if the two do not match, it determines that the election type of the ballot is different (S117). In the example shown in FIGS. 4 and 6, the evaluation character string Se generated from the ballot paper to be determined matches the reference character string Sr stored in the learning mode, and therefore the election type of the ballot paper is determined to be correct. Is done.

  Through the above processing performed in the determination mode, it is determined whether or not the election type of the ballot to be determined is correct. Then, the ballots determined to be correct in the election type are classified by being conveyed to the stacker 6 corresponding to the entry contents (candidate name, etc.), and the ballots determined to be different in the election type are mixed. It is regarded as a vote and rejected.

  As described above, in the ballot paper classification device according to the first embodiment, the evaluation character string Se generated from the ballot paper to be determined and the reference character string generated from the ballot paper of the correct election type for learning Based on whether or not Sr matches, it is determined whether or not the election type of the ballot to be determined is correct. Thereby, it can be automatically determined whether or not the election type of the ballot is correct.

Embodiment 2. FIG.
Next, a ballot paper classification device according to Embodiment 2 of the present invention will be described with reference to FIGS.
Based on the character string extracted from the election type region R, the ballot paper classification device according to Embodiment 1 determines whether the election type of the ballot paper to be determined is correct. In contrast, the ballot paper classification device according to the second embodiment determines whether the election type of the ballot paper to be determined is correct based on each pixel value of the election type region R. In the following description, detailed description of the same parts as those in the first embodiment will be omitted.

  As schematically shown in FIG. 7, an arithmetic processing unit 208 is provided inside the voting paper classification apparatus according to the second embodiment. The arithmetic processing unit 208 includes an area extraction unit 9, an evaluation pixel value array generation unit 210, a mask pattern generation unit 211, a storage unit 212, and a determination unit 213.

  The area extraction unit 9 extracts the election type area R from the image information I of the ballot paper generated by the imaging unit 5 as in the first embodiment.

  The evaluation pixel value array generation unit 210 generates a two-dimensional array Pe [x, y] of evaluation pixel values serving as evaluation information by extracting each pixel value of the election type region R in the voting paper to be determined.

  The mask pattern generation unit 211 extracts each pixel value of the election type region R in the learning ballot, and generates a two-dimensional mask pattern Mr [x, y] serving as reference information based on the extracted pixel values.

  The storage unit 212 stores the mask pattern Mr generated by the mask pattern generation unit 211.

  Based on the result of applying the mask pattern Mr stored in the storage unit 212 to the evaluation pixel value array Pe generated from the ballot paper to be determined, the determination unit 213 elects the ballot paper to be determined Determine whether the type is correct.

  Next, a process for determining whether or not the election type of the voting paper is correct in the voting paper classification device according to the second embodiment will be described.

(Learning mode)
When a learning ballot is placed on the hopper 3 and the start of the learning mode is instructed, the ballot classification device performs the process shown in the flowchart of FIG.

  First, the image pickup unit 5 picks up an image of a learning ballot paper and generates binarized image information I (S201). The region extraction unit 9 extracts the election type region R from the image information I and outputs it to the mask pattern generation unit 211 (S202).

  The mask pattern generation unit 211 extracts each pixel value from the election type region R to generate a pixel value array Tmp [x, y] as shown in FIG. 9A (S203), and expands the result. Then, a mask pattern Mr [x, y] as shown in FIG. 9B is generated by performing the inversion process (S204), and the mask pattern Mr is stored in the storage unit 212 (S205). In FIGS. 9A and 9B, the black area is an area having a pixel value of 1, and the white area is an area having a pixel value of 0.

  Through the above processing performed in the learning mode, the mask pattern Mr is generated based on each pixel value of the area including the character string representing the election type printed on the learning ballot and stored in the storage unit 212. .

(Judgment mode)
When instructed to start the determination mode, the ballot paper classification device performs the process shown in the flowchart of FIG. 10 on each ballot paper to be determined.

  First, the imaging unit 5 captures the entry surface of the ballot paper and generates binarized image information I (S211), and the region extraction unit 9 extracts the election type region R from the image information I and evaluates the pixel value. The data is output to the array generation unit 210 (S212).

  The evaluation pixel value array generation unit 210 extracts each pixel value of the election type region R (S213), and generates an evaluation pixel value array Pe [x, y] (S214).

  The determination unit 213 applies the mask pattern Mr stored in the storage unit 212 to the array Pe of evaluation pixel values generated in step S214 (S215). That is, the logical product of the corresponding pixel values of the evaluation pixel value array Pe [x, y] and the mask pattern Mr [x, y] is calculated. Then, it is checked whether or not a predetermined number of pixels having a value of 1 remain (S216).

  As shown in FIG. 11A, if the election type of the ballot for determination and the ballot for learning match, the result of applying the mask pattern Mr to the array Pe of evaluation pixel values is a value. However, there is almost no remaining pixel. On the other hand, as shown in FIG. 11 (b), when the election types of the voting paper to be judged and the voting paper for learning are different, the mask pattern Mr is applied to the evaluation pixel value array Pe. As a result, many pixels having a value of 1 remain. Therefore, the determination unit 213 applies the mask pattern Mr to the evaluation pixel value array Pe, and if the predetermined number or more of the pixels having the value 1 do not remain, the election type of the ballot to be determined is correct. A determination is made (S217), and if a predetermined number or more remain, it is determined that the election type is different (S218).

  As described above, in the voting paper classification device according to the second embodiment, the evaluation pixel value array Pe generated from the voting paper to be determined is generated from the voting paper of the correct election type for learning. Based on the result of applying the mask pattern Mr, it is determined whether or not the election type of the ballot to be determined is correct. Thereby, it can be automatically determined whether or not the election type of the ballot is correct.

  In the first and second embodiments, the example in which the election type region R is extracted from the image information I on the entry surface of the ballot is shown. However, if the information indicating the election type is on the back side of the entry surface, the entry is made. The election type region R may be extracted from the image information on the back side of the face. A plurality of election type regions R may be extracted.

  In the second embodiment, the evaluation pixel value array Pe is generated based on the binarized image information I. However, the image information I may be in gray scale. In that case, in step S216, it is determined whether or not the election type of the voting paper to be determined is correct based on whether or not a predetermined number or more of pixel values remain.

  5 imaging unit (imaging means), 9 area extraction section (area extraction means), 10 evaluation character string generation section (evaluation information generation means), 11 reference character string generation section (reference information generation means), 12 storage section (storage means) ), 13 determination unit (determination means), R election type area, Se evaluation character string (evaluation information, first character string), Sr reference character string (reference information, second character string), 210 evaluation pixel value array Generation unit (evaluation information generation unit), 211 mask pattern generation unit (reference information generation unit), 212 storage unit (storage unit), 213 determination unit (determination unit), Pe array of evaluation pixel values (evaluation information, pixel value Set), Mr mask pattern (reference information, mask pattern).

Claims (3)

In the ballot classification device that classifies the ballot based on the contents of the entry,
An image pickup means for picking up an image of a ballot and generating image information;
An area extracting means for extracting an election type area including information representing an election type from the image information generated by the imaging means;
Evaluation information generating means for generating evaluation information from the election type area extracted by the area extracting means;
Reference information generating means for generating reference information from the election type area in the image information of the correct election type ballot given in advance;
Storage means for storing the reference information generated by the reference information generation means;
Whether the election type of the ballot to be judged is correct based on the evaluation information generated from the election type area in the image information of the ballot to be judged and the reference information stored in the storage means A voting paper classification device, comprising: a determination unit that determines whether or not. The evaluation information is a first character string extracted from the election type area in the image information of the ballot paper to be determined,
The reference information is a second character string extracted from the election type area in the image information of a correct election type ballot given in advance,
The determination means determines whether or not the election type of the ballot to be determined is correct based on whether or not the first character string matches the second character string. The ballot paper classification device according to claim 1. The evaluation information is a set of pixel values of the election type area in the image information of the ballot paper to be determined,
The reference information is a mask pattern created based on a set of pixel values of the election type area in image information of a correct election type ballot given in advance,
The determination unit determines whether the election type of the vote ballot to be determined is correct based on a result of applying the mask pattern to the set of pixel values generated from the ballot paper to be determined. The ballot paper classification device according to claim 1, wherein:

Images