JPH07182442A - Character recognition device and method therefor - Google Patents

Abstract

PURPOSE:To provide a character recognition device and method which can improve its operability for recognition of characters. CONSTITUTION:An original image is inputted at an image reading part 1002 and recognized at a character recognition part 4000. At the same time, a copied original image is printed at an image recording part 1004 by means of the font of a character recognition device in regard of the characters (character codes) that could be recognized at the part 4000. Meanwhile the character that could not be recognized at the part 4000 is defined as a prescribed pattern if the candidates of pattern showing the blank columns, for example, cannot be defined into one. At the same time, a group of these candidates are selectably printed. An operator carefully writes characters in the blank columns and puts a mark to a candidate for selection. Then the operator reads the output image again. The character recognition device recognizes again the written characters as the recognition result for the characters which could not be recognized at first in the original. Meanwhile the selected character is regarded as the recognition result for the candidate put with the mark.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a character recognition device and method,
More specifically, the present invention relates to a character recognition device and method for recognizing characters in a document image.

[0002]

2. Description of the Related Art Generally, in this type of apparatus, a document to be recognized is read, and individual characters in the read image are cut out to perform character recognition.

However, when recognition is not possible, or when a plurality of candidates of the same level occur and the recognition result cannot be specified, the corresponding character image of the input image is displayed and the official character code for it is displayed. I was typing on the keyboard.

[0004]

However, in order to input a proper character code for each sentence image displayed on the screen, many operations must be performed, which is very troublesome.

If a dedicated display device or keyboard is provided for that purpose, the device becomes complicated and costly.

[0006]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and an object of the present invention is to provide a character recognition apparatus and method which improve the operability of character recognition.

In order to solve this problem, for example, the character recognition device of the present invention has the following structure. That is, in a character recognition device that inputs a document image and recognizes characters in the document image, a recognition unit that recognizes characters in the input document image and a normal character pattern is generated for the recognized character, For unrecognizable characters, a printing means that generates and prints out a predetermined pattern that specifies a blank space to allow re-entry, and the entered output image is input again, and the entry in the input image is entered. The recognized character is recognized again, and the character recognized first by the recognizing means is combined with it to obtain a recognition result of the document image.

A character recognition device of another invention has the following structure.

In a character recognition device for inputting a manuscript image and recognizing characters in the manuscript image, a recognition means for recognizing characters in the inputted manuscript image and a normal character pattern are generated for the recognized characters. , If the recognition candidates cannot be narrowed down, the character is specified and a character pattern of the candidate group is generated in a format that can be selected one by one, and a printing means for printing an image based on the generated character pattern, and a selected and entered The output image is input again, and the selected character in the input image is used as the recognition result of the character that is not narrowed down by the recognition means, and the original image is recognized by combining the character recognized first by the recognition means. And resulting control means.

[0010]

In the structure or process of the present invention, for example, a pattern that clearly indicates a blank space where characters that cannot be recognized at the initial stage can be re-written is generated and printed together with the recognized characters. Let The operator sees the print result, re-fills the blank position exactly, and re-enters it. When the apparatus receives the input, the apparatus recognizes again the pattern that clearly indicates the blank, and combines it with the previously recognized result to obtain the document image recognition result.

[0011]

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. In the embodiment, an example adapted to a copying machine will be described.

FIG. 1 is a block diagram of the embodiment, and each component will be described below together with its operation.

The original 1001 passes through the CCD 102 from the lens 101 of the image reading unit 1002, enters the A / D converter 103 for A / D conversion, and the signal after A / D conversion is input to the image processing unit 3000. Image processing unit 10
In 03, shading correction 104 is performed, and the process proceeds to the mode switching circuit 105.

Here, the output destination is switched based on whether it is the image processing setting mode or the copy mode (set by the operation unit 116). In the normal copying mode, the light color density conversion circuit 106 converts the luminance data into recording data and outputs the recording data to the image editing unit 107. The image editing unit 107 performs editing processing based on the data (data processing circuit 112) processed according to the setting content of the image processing stored in the RAM 115 in the CPU circuit 113. Then, the edited image data is stored in the image recording unit 1.
Output to 004.

The image recording unit 1004 includes a control circuit such as a motor for carrying a transfer sheet, a laser recording circuit unit for writing a video signal input from the image processing unit 1003 onto a photosensitive drum, and a developing control circuit for developing. And records an image on a recording medium such as recording paper.

On the other hand, when the image processing setting mode is set, the output destination of the mode switching circuit 105 is the image recognition section 4.
000. In the image recognition unit 4000,
First, the pattern recognition unit 108 performs pattern identification. Here, an area such as a pattern for re-inputting unrecognizable characters, a plurality of candidate option patterns, and a marking for a plurality of candidate display requests is recognized.

Thereafter, the character recognition circuit 109 cuts out an image (character pattern) for each character from the input image and performs character recognition. The character recognition circuit 109 has a built-in dictionary for character recognition. Character recognition circuit 109
The result recognized by is passed to the image conversion unit 110, where the character code of the recognition output is converted into an output image. Here, a character frame for re-input is used for unrecognizable characters, and an option is added for a character having a plurality of candidates to generate an output image.

The operation unit 112 is also an image processing unit 100.
3 includes various key groups for instructing image copy operations such as image editing content, the number of copies, and scaling ratio, various LED groups, and a display unit for displaying the content at the time of operation.

FIG. 2 is a sectional view showing the structure of the image copying apparatus in the embodiment.

In FIG. 2, reference numeral 1 denotes a document feeding device serving as a document feeding means, which feeds the placed documents one by one or continuously at a predetermined position on the platen glass surface 2. Reference numeral 3 denotes a scanner including a lamp, a scanning mirror 5, and the like. The scanner 3 is placed on the platen glass surface 2 by the document feeder 1 and gives a copy instruction or a character recognition instruction from the operation unit 116.
The scanner scans in a predetermined direction (left and right direction in the drawing), and the reflected light of the document passes through the lens 8 through the scanning mirrors 5 to 7 and forms an image on the image sensor unit 9.

Reference numeral 100 is a control unit having the various circuits shown in FIG. An exposure unit 10 receives an image signal from an image signal control unit (not shown) that generates an image signal based on image data from the image editing unit 107 in the control unit 100, and outputs a light beam to the photoconductor 11. Irradiate. 12, 13
Is a developing device that visualizes the electrostatic latent image formed on the photoconductor 11 with a developer (toner) of a predetermined color. Denoted at 14 and 15 are transfer paper stacking units, in which recording media of a fixed size are stacked and housed, and the leading edge of the image is aligned with the image formed on the photoconductor 11 by being fed to the registration roller arrangement position by driving the feeding roller. The paper is re-fed in a timely manner.

A transfer / separation charger 16 transfers the toner image developed on the photoconductor 11 onto the transfer paper, and then is separated from the photoconductor 11 and fixed by the fixing unit 17 via the conveyor belt. A paper discharge roller 18 stacks and discharges the transfer-receiving paper on which the image formation is completed on the tray 20. Numeral 19 is a direction flapper to switch the carrying direction of the transfer-receiving paper for which image formation has been completed to the paper discharge port and the internal carrying direction to prepare for the multiplex / double-sided image forming process.

FIG. 3 shows the operation unit 116 used in this embodiment.
FIG.

Reference numeral 5001 is a power switch for controlling the power supply to the apparatus main body. A reset key 5002 operates as a key for returning to the standard mode during standby.
5003 is a copy start key. A clear key 5004 is used to clear a numerical value such as the number of copies.

Reference numeral 5005 denotes an ID key, which is the ID key 50.
05 enables a copy operation for a specific operator,
For operators other than the above, the copying operation can be prohibited unless the ID is input by the ID key. 5
A stop key 006 is a key used when interrupting or canceling copying. Reference numeral 5007 denotes a guide key, which is used when it is desired to know each function. An up cursor key 5008 is a key for moving the pointer up on each function setting screen. Fifty
A down cursor key 09 is a key for moving the pointer downward on each function setting screen. A right cursor key 5010 is a key for moving the pointer to the right on each function setting screen. A left cursor key 5011 is a key for moving the pointer to the left on each function setting screen. Reference numeral 5012 denotes an OK key, which is pressed when the setting is acceptable on each function setting screen.

Reference numeral 5013 indicates 50 on each function setting screen.
Press this key when you want to execute what was output at the bottom right of the screen at 52. A fixed size reduction key 5014 is used when reducing a fixed size to another fixed size. 501
5 is used when selecting the same size copy. Reference numeral 5016 is a standard enlargement key, which is used to enlarge a standard size to another standard size. A cassette selection key 5017 is used to select a cassette stage to be copied. A copy density adjustment key 5018 is used to reduce the density. Reference numeral 5019 denotes an AE key, which automatically adjusts the copy density with respect to the density of the original. A copy density adjustment key 5020 is used to increase the density. Reference numeral 5021 is a key for designating the operation of the sorter. 5022 is a preheat key, and the preheat mode is ON
Used for / OFF. 5023 is an interrupt key,
Press this to interrupt during copying and perform copying.
Reference numeral 5024 is a ten-key pad, which is used when inputting a numerical value.

A marker processing key 5025 is used to set trimming, masking, and partial processing (contour processing, mesh processing, shadowing processing, negative / positive processing). Reference numeral 5026 denotes a pattern processable key, which is used when a color is expressed as a pattern or a color is expressed as a density difference. Fifty
A color erasing key 27 is used to erase a specific color. An image quality key 5028 is used when the user wants to set the image quality. 5029 is a negative-positive key,
Used when performing negative / positive processing. An image create key 5030 is used when performing contour processing, shadowing processing, halftone processing, italic processing, mirror processing, and repeat processing. A trimming key 5031 is used when the area is designated and trimming is performed. Reference numeral 5032 denotes a masking key, which is used when an area is designated and masking is performed. Reference numeral 5033 denotes a partial processing key, which specifies an area and then partial processing (contour processing, halftone processing, shadowing processing, negative / positive processing). Reference numeral 5034 denotes a frame erasing key, which is used to erase the frame according to the mode. The modes are sheet frame erase (creates a frame for the sheet size), document frame erase (creates a frame according to the document size, and document size is specified), book frame erase (frames match the book spread size). There is a blank in the center. There is a book spread size specification. A binding margin key 5035 is used when a binding margin is to be created at one end of the paper.

Reference numeral 5036 is a movement key for moving an image in a desired area, which is used when it is desired to move.
The movement includes parallel movement (up and down, left and right), center movement, corner movement, and designated movement (point designation). 5037
Is a zoom key, and the copy magnification can be set from 25% to 400% in 1% increments. In addition, main scanning and sub scanning can be set independently. Note that the scaling of the image is performed by controlling the moving speed of the scanner 3 in the sub-scanning direction of the document, and in the main scanning direction of the document, image data read out from the scanner 3 is thinned or interpolation processing is performed. Done in.

Reference numeral 5038 denotes an automatic variable magnification key, which automatically enlarges or reduces the size of the copy paper. Further, the main scanning and the sub-scanning can be automatically scaled independently. Reference numeral 5039 denotes an enlargement continuous shooting key, which is used when one original is enlarged to a plurality of sheets for copying. A reduction layout key 5040 is used when a plurality of originals are enlarged or reduced to one and copied. Reference numeral 5043 denotes a continuous shooting key, which is used when the copying area on the platen glass surface is divided into right and left and automatic copying is performed to perform continuous copying (page continuous shooting, double-sided continuous shooting). Reference numeral 5044 denotes a double-sided key, which is used when it is desired to output both sides (single-sided double-sided, page continuous shooting both-sided, double-sided both-sided). A multiple key 5045 is used when it is desired to perform multiplex (multiplex, page continuous shooting multiplex). Fifty
Reference numeral 46 is a memory key, which is used when a mode using the memory is desired (memory synthesis, area synthesis, watermark synthesis). Reference numeral 5047 is a projector key, which is used when the projector is used. Reference numeral 5048 is a printer key, which is used when setting at the time of printing.
Reference numeral 5050 denotes a document mixed loading key, which is used when document sizes are mixed when a copy is made using the feeder. Reference numeral 5051 denotes a mode memory key, which is used to call the registered copy mode in order to register the copy mode set for copying. A display screen 5052 displays the state of the apparatus, the number of copies, the copy magnification, and the copy paper size, and displays the setting contents during the copy mode setting.

Reference numeral 5053 is a key for switching between the copy mode and the image processing setting mode.

The flowchart of FIG. 4 shows the operation of the character recognition circuit 109 of the embodiment. Hereinafter, description will be given along the flow of this flowchart. It is assumed here that the document image is vertically written, for example. Further, when the recognition target sentence is in horizontal writing, “character width” described below may be read as “character height” and processed.

First, the character direction of a document is determined based on the presence or absence of an image in the vertical direction (for example, a histogram) from the input image, and the line width h is obtained (step S0).

Next, character patterns are cut out according to the presence or absence (histogram) of an image in the horizontal direction for each obtained line, and the character width w (i) and position coordinate x (i) of each character pattern are stored. (Step S1). The character width w (i) of the character pattern cut out here is: h−Δa ≦ w (i) ≦ h + Δa (1) (where Δa: allowable range value from line width h, i: character pattern number) Then, the standard character width w is obtained by picking up only the character pattern that is and calculating the average thereof (step S2).

Next, the obtained standard character width w is compared with the character width w (i) of each cut out character pattern, and when the character width w (i) is significantly larger than the standard character width W,
That is, a character pattern satisfying w + Δb <w (i) (2) (where Δb: an allowable range from the standard character width) is searched for (step S3).

If a character pattern satisfying the condition of step S3 is present, up to a character pattern having a standard character width before and after the character pattern, that is, including a character pattern satisfying this condition, and The character pattern between the character patterns having the standard character width at the closest position is projected in the horizontal direction to create a histogram (step S4).

The position on the horizontal axis is x, the number of dots on the vertical axis is y, a predetermined threshold value z is set for the value y, and the position of the smallest character pattern number i is used as a reference (x =
0), a value x near the standard character width and smaller than the threshold z is obtained (step S5).

If the length of the projection is L, then x
If the cut width is about the standard character width W and the final cut width is about W, character cutting processing is newly executed at the calculated x (step S6).

However, in other cases, that is, when no break is found near the standard character width W in the process of obtaining x, or even when the break is reached to the end, the character width of the last character pattern is significantly less than W, new. Character cutting processing is not executed (steps S5 → S7).

Character recognition processing is executed for the finally cut out character pattern to generate a character code (step S7).

FIG. 5 is a flow chart showing the processing in the character recognition unit 109 when character recognition is performed on a document image.

First, character recognition is performed (step S600).
1), it is determined whether or not the character is fixed (step S)
6002). When it is confirmed, the character code is stored in the character recognition RAM (provided in the character recognition circuit 109) (step S6008), and the image data corresponding to the character code is generated by the image data conversion unit 110 (step S6008). S6009), and records it in the output image RAM in the image editing unit 107 (step S6010).

On the other hand, if the character is not confirmed, step S60 is performed.
If the determination is 02, the following processing is performed.

First, when there are a plurality of candidates (step S6003), pattern data of inverted characters (numerals) indicating that there are a plurality of candidates are added to the add-on data creation unit 1.
It is generated in 11 and stored in the output image RAM at the position corresponding to the unrecognized character (step S6004).
The candidate characters are converted into image data (step S6005), added with check frame data (step S6006), and recorded in the output image RAM (step S6006).

If there is no candidate (step S600)
In 3), the re-input frame data is generated by the add-on data unit 111 and recorded in the output image RAM (step S6007).

The above-described processing is repeated for the number of characters to generate output image data, which is output by the image recording unit 117.

FIG. 7 shows the transition of the character recognition processing for the "horizontal writing" document.

When the original image 8001 to be recognized is read and recognized, the image indicated by reference numeral 8002 is printed. Here, "East" indicates that it could not be recognized, and the number "1"
Encourage the selection of candidates indicated by. here,
This shows an example in which "east" and "bundle" are not candidates at the same level and are candidates for selection.

It also indicates that "sea" and "road" could be recognized, and "line" was not recognized and no candidate appeared.

The operator looks at this output document and selects any of the selection items, and carefully writes the characters for which no candidate could be generated in the blank space (reference numeral 8003 in the figure). ).

During this period, the apparatus is instructed to read the original written by the operator. However, when the reading instruction is received again, it is recognized that the original is read and the processing is performed. To do. However, in this recognition process, since it is known at which position the add-on mark or blank space was printed in the previous printing process, only that position is checked, the character is selected by the selected item, and further, the blank space is filled. The recognized character is recognized again.

As a result, it becomes possible to obtain the result indicated by reference numeral 8004.

The character size (dot pattern size) generated at the printing stage is adjusted depending on the character size of the original 8001. Since the generation of the character pattern in the embodiment is generated by the scalable font (for example, outline font), its quality is very high.

Now, the above-mentioned processing will be carried out. The details of the re-recognition processing will be described with reference to the flowchart of FIG.

FIG. 6 is a flow chart of processing at the time of reloading. First, the specific pattern recognition unit 108 recognizes a character frame, a reverse character pattern, or the like (step S7001). Next, it is determined whether or not there is a character frame (step S7002), and if there is, an image within the character frame is extracted (step S7003), and character recognition is performed only on the image within the character frame (step S7004). The character code which is the recognition result is stored in the recognition RAM (step S7005). At this time, from the position of the character frame,
It collates which character each character corresponds to the previously undecided character (to determine the expansion position of the character image).

Then, it is also determined whether or not there is a reverse character pattern (step S7006). If there is a reverse character pattern, it is recognized which check box of the plural candidate patterns outside the box is checked (step S
7007), it recognizes from the checked position which original character character which has not been fixed last time is fixed as which candidate character character and outputs the character code (step S7008). At this time, the character code is stored in the character recognition RAM (step S7009). When these processes are completed, the character data in the character recognition RAM is expanded in the output image RAM (step S7010), and then the data is read from the output image RAM and the recognition image is output (step S70).
11).

As described above, the final output image 8004 shown in FIG. 7 is obtained.

In the above example, the final output image 8004 includes the intermediately output candidate character, but it may be deleted.

In the above-described embodiment, the re-recognition process is described as a process to be taken from the fact that the output of the first recognition result has already been performed. However, it is determined whether the first recognition process or the subsequent recognition process is performed on the operation unit. A key for instructing may be provided and this may be realized.

<Description of Other Embodiments> With the same configuration as in the first embodiment, at the time of re-inputting, regarding the character recognized last time,
An embodiment will be described in which the character code is not expanded and the clean copy output can be performed quickly.

Character recognition of the original image is performed according to the flowchart of FIG.

First, when the character cannot be recognized (step S
5001, 5002) and the image data of the output frame is recorded in the output image RAM (step S5007).

When the character is confirmed (step S5)
002) stores a character code in the character recognition RAM (step S5008), converts the character code into image data (step S5009), and outputs image RAM.
Is recorded (step S5010). After doing this for all characters, the data is read from the output image RAM and the correction image is output.

The control when the output image is modified and re-input is described with reference to FIG.

First, the image is recorded in the output image RAM (step S9001). At the same time, the specific pattern recognition (step S900) for recognizing the specific pattern (character frame) is performed.
After performing 2), the image in the character frame is extracted (step S
9003) and character recognition is performed (10004). And
Each recognized character is recorded in the character recognition RAM (step S9005), simultaneously converted into image data (step S9006), and written in the output image RAM (step S9007). The data in the output image RAM is read and output as the final character recognition image (step S9).
008). In this way, at the time of correction, by correcting only the image data of the corrected portion, the character-recognized image can be output quickly.

As described above, according to this embodiment, when the candidates cannot be narrowed down in the first recognition processing result,
A correct recognition result can be obtained by explicitly printing the unrecognizable case, printing the print result directly by the operator, and reading the print result again. Therefore, since there is no need to operate the various keys on the screen to make corrections, it is possible to significantly improve the operability.

In the embodiment, in the first recognition process, the copy image of the original image is printed and the candidate group is printed together in the printed image, but it may be printed separately. In particular, when there are many characters in the original image and there is no space for printing the candidates, it is convenient to use this method.

In the embodiment, it is adapted to a copying machine and the finally recognized result is printed out. However, for example, it is stored in a memory and used for document editing, or in some cases a floppy disk device. It may be stored in a storage medium such as.

Further, in the embodiment, an example in which it is adapted to a single copying machine has been described, but the same operation and effect can be obtained in a system including, for example, an image reader, a host computer for character recognition processing, and a printer for printing out. Can be obtained. In this case, the host computer may be loaded with the program for realizing the above processing, so that the present invention may be a single device or a combination of a plurality of devices. Furthermore, it can be easily inferred that it can be realized even if it is realized by supplying the program.

[0069]

As described above, according to the present invention, when recognition processing is performed, a copy image corresponding to the original image is printed, but characters that cannot be recognized are re-entered in the copy image. Will be printed with blank spaces to the extent that
The operator only needs to fill in the blanks with the appropriate characters and re-enter them. Therefore, it is possible to give the operator an intuitive and very easy-to-understand environment.

According to another aspect of the invention, when the recognition process fails to narrow down to a specific character, a copy image of the original image whose candidate list cannot be printed is printed. The operator only needs to make appropriate entries in order to select those candidates and then perform the input again, so that it is possible to provide a further excellent effect from the standpoint of the user interface.

[0071]

[Brief description of drawings]

FIG. 1 is a block diagram showing a device configuration of an embodiment.

FIG. 2 is a cross-sectional structure diagram of an apparatus of an example.

FIG. 3 is an external view of an operation unit according to the embodiment.

FIG. 4 is a flowchart showing processing contents of a character recognition circuit according to an embodiment.

FIG. 5 is a flowchart showing a procedure for outputting a first recognition result according to the embodiment.

FIG. 6 is a flowchart showing the processing contents at the time of rereading in the embodiment.

FIG. 7 is a diagram showing a transition of processing according to the embodiment.

FIG. 8 is a flowchart showing an output processing procedure of a first recognition result in another embodiment.

FIG. 9 is a flowchart showing processing contents at the time of rereading in another embodiment.

[Explanation of symbols]

 1001 original 1002 image reading unit 1003 image processing unit 1004 image recording unit 113 CPU 116 operation unit

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Keizo Isemura 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Akihiko Sato 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Incorporated (72) Inventor Noriaki Matsui 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Tokuji Kaneko 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Yoshio Mizuno 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Koichi Takahashi 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Invention Person Masahiro Serizawa 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Yasushi Fukada 3-30 Shimomaruko, Ota-ku, Tokyo No. Canon within Co., Ltd. (72) inventor Yoshihiro Hosomi Ota-ku, Tokyo Shimomaruko 3-chome No. 30 No. 2 Canon within Co., Ltd.

Claims (4)

[Claims] 1. A character recognition device for inputting a manuscript image and recognizing characters in the manuscript image, a recognition means for recognizing characters in the inputted manuscript image, and a normal character pattern for the recognized character. A printing unit that generates and prints out a predetermined pattern that clearly shows a blank space that can be re-entered for unrecognized characters that have occurred, and re-input the filled-out output image. And a control unit for recognizing the entered character again and combining the character recognized first by the recognizing unit as a recognition result of the document image. 2. A character recognition method for inputting an original image and recognizing characters in the original image, wherein a recognition step of recognizing characters in the input original image and a normal character pattern for the recognized character are provided. A printing process that generates and prints out a predetermined pattern in which a blank space is specified to allow re-entry for an unrecognizable character, and the entered output image is re-input The character recognition method, further comprising the step of: recognizing the entered character again and combining the character recognized first in the recognition step to obtain the recognition result of the original image. 3. A character recognition device for inputting a manuscript image and recognizing characters in the manuscript image, a recognition means for recognizing characters in the inputted manuscript image, and a normal character pattern for the recognized character. If a recognition candidate cannot be narrowed down, the character is specified, and a character pattern of the candidate group is generated in a format that allows selective selection, and a printing means for printing an image based on the generated character pattern, and a selection method. The input output image is input again, the selected character in the input image is used as the recognition result of the character that is not narrowed down by the recognition unit, and the original image is combined with the character recognized first by the recognition unit. And a control unit that sets the recognition result as the character recognition result. 4. A character recognition method for inputting a document image and recognizing characters in the document image, comprising: a recognition step of recognizing characters in the input document image; and a normal character pattern for the recognized characters. If a recognition candidate cannot be narrowed down, the character is specified, and a character pattern of the candidate group is generated in a format that allows selective selection, and a printing process for printing an image based on the generated character pattern, and a selection process. The input output image is input again, the selected character in the input image is used as the recognition result of the character that is not narrowed down in the recognition step, and the original image is combined with the character recognized first in the recognition step. And a control step for setting the recognition result as the recognition result.