JPH03214286A - Character recognizing device - Google Patents

Abstract

PURPOSE:To improve the identification factor of a separate character by registering the image data on the separate character that is erroneously identified into a recognition dictionary part. CONSTITUTION:The segmented character signals S7 which form the while of a separate character that is erroneously recognized is supplied to a dictionary production part 29. Thus an entire image of the separate character is registered into a recognition dictionary part 30. Therefore a character recognizing part 28 integrates the signals S7 forming each component part of the separate character when this character is identified. Then the part 28 can accurately identify the separate character by reference to the part 30. In such a way, the identification factor is improved for the separate character.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character recognition device suitable for use, for example, in recognizing characters in a printed document and converting them into character codes.

[Summary of the invention]

The present invention relates to a character recognition device suitable for use, for example, when recognizing characters in a printed document and converting them into character codes.
a character extraction unit that sequentially extracts extracted character signals forming each component of one character or separated characters from an original character signal corresponding to an image of an original document; and a recognition dictionary unit that stores image data corresponding to character codes. a character identification section that uses the recognition dictionary section to identify a character code corresponding to one or more of the cutout character signals; and a character identification section that recognizes the image data of the cutout character signal supplied from the character cutout section. It has a dictionary creation section that registers the separated characters in the dictionary section, and if the character identification section makes a mistake in identifying a separated character, the cutout character signal forming the entire separated character that was misidentified is transferred from the character cutting out section. By supplying it to the dictionary creation section,
By registering the image data of the incorrectly identified separated characters in the recognition dictionary, it is possible to easily increase the identification rate of separated characters.

[Conventional technology]

For example, in order to automate the process in which a worker picks up type in letterpress printing, a character recognition device is required to recognize each character in a document created by type printing or the like and convert it into a character code.

Fig. 6 shows a conventional character recognition device disclosed in Japanese Unexamined Patent Publication No. 62-74181. In Fig. 6, (1) is a document reading section;
Original character signal S1 corresponding to the shading of one page of the original from
is supplied to the character string extraction unit (2).This original character signal 51 is obtained by dividing the document into dots at a predetermined density, and representing black dots as high level "1" and white dots as low level "O". However, the density of each dot is
It may also be expressed in decimal numbers.

The character string extraction unit (2) is composed of a first stage preprocessing unit (3), a second stage preprocessing unit (4), and a third stage preprocessing unit (5). The preprocessing unit (3) removes noise and corrects the rotation of the document, and the second stage preprocessing unit (4) separates the character area AR (see Figure 7) from other areas (areas of photographs, drawings, etc.). Only the image data included in the character area AR is extracted, and the third
The character area A extracted in the pre-processing section (5)
Character string signals S4 corresponding to character strings API, AR2, . . . included in R are extracted.

In order to extract this character string signal S4, as shown in FIG. Y)
Expressed in coordinates, the value of "1" or "0" of each dot is projected onto the Y axis and the sum is calculated to generate a Y projection signal Sy. Then, when the second Y projection signal Sy is binarized at a predetermined threshold level, each section of the 200,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,0,0,0,0,0,0,0,0,0,0,0,0,0,0, level, section of level "1" section becomes a character string. A Rl, A R2
, . . , and the character string signal S4 is supplied to the subsequent character cutting section (6).

In character extraction error (6), for example, the character string signal S4 of the first character string ARi shown in FIG. 8A is projected onto the X axis to generate an X projection signal Sx, and this X projection signal SX is By binarizing with a threshold value THI of level (value 1), a coarse cutout signal DTI (Fig. 8C) is obtained, and this X projection signal Sx is converted to a medium level threshold value TH2 (Fig. 8D).
By binarizing the signal DT2 (Fig. 8E)
》obtained. Similarly, the coarse cutting signal DTI is at high level.
By individually generating the Y projection signal Sy only in the 1'' section, it is possible to generate the cutout signal in the Y direction.

Finally, as shown in FIG. 8A, for example, for the character "te", the circumscribing frame (9) (M[) becomes high level "1", and the separated character "i"
For each separated part, the circumscribing frame (11
), (12), a basic rectangular signal is obtained by sequentially cutting out only the parts where the cutting signal becomes high level "1" from the human input character string signal S4. This becomes a cutout character signal S7.

Incidentally, the fine cutting signal DT2 shown in FIG. 8E is used when examining the finer structure of each character. In addition, for “i”, which is a separated character in Figure 8A, the circumscribing frame (11), (
12) Since there are two, it is necessary to integrate them later at the character identification stage.

Reference numeral (7) indicates a character recognition section, and this character recognition section (7) takes in the basic rectangular cut-out character signal S7 for each circumscribed frame and performs character recognition. Specifically, first, classification is performed by position, and characters existing in the upper half range (r' J, r'', ゜, etc.) and characters existing in the lower half of the character example ARi in Figure 8 are classified. character (``.''r.'', r,
Pattern matching is performed using J, etc.) as the first characteristic character, and a corresponding character code (JIS code, etc.) is assigned. If identification is not possible, the width of the circumscribing frame is set to W1 and the height is h, and classification is performed based on the aspect ratio h/w and relative size. That is, w1 side ratio h/w is 0<h
Classification is performed depending on whether it falls within the range /w<0,5 or l,5<h/w. Furthermore, let the width of the average-sized circumscribed cutlet be wR and the height be h, and the vertical ratio h/
ha and stirring ratio w/wII are each 0 and h/L<
0.5 and 0 < w / w tr < 0. 5
(Classification is performed depending on whether it falls within the D range or not. Pattern matching is performed using characters that fall within the above range as second characteristic characters.

Further, for characters that are not classified as the first or second characteristic characters, pattern matching is performed with individually stored dot patterns, and if a predetermined degree of matching is obtained, the character code is assigned.

If a character that cannot be lJIWa still remains, the operations of re-cutting the circumscribing frame to further separate it into a plurality of minute circumscribing elements and merging them with the subsequent circumscribing frame are performed. In addition, if an unrecognizable character ultimately remains, a reject code indicating that the character is an unrecognizable character is given to that character.

The character code for one page of the manuscript generated by the character identification section (7) is stored in a predetermined storage device along with information indicating the position and size of the character. Furthermore, the video signal of the character corresponding to the character code is displayed on a display unit (8) such as a cathode ray tube so that the operator can judge whether the recognition result is correct or not.
A group of characters as a recognition result is displayed on the display screen of the display unit (8) in a format corresponding to the document. In this case, a high-intensity rectangular blank is displayed in the portion of the character that cannot be recognized. Therefore, if a character to be corrected or a character that cannot be recognized exists, the operator can input the desired character into that part in the same way as a word processor.

As described above, an original character signal S1 corresponding to the shading of the original is generated, and this signal S1 is cut out using a circumscribing frame circumscribing one character to generate a cutout character signal S7, and a cutout character signal S7 corresponding to the cutout character signal S7 is generated. It can be said that the character recognition algorithm itself, which specifies the characters that are used, is basically established.

[Problem to be Solved by the Invention] However, when a character recognition device using this character recognition algorithm was actually set up in an office and used by an operator, it was found that there were various inconveniences in terms of operability.

One of its disadvantages is that it is difficult to increase the identification rate of separated characters. In other words, character recognition devices may misread only certain characters in a document;
If the specific character is a normal integrated character, the learning function uses the basic rectangular cutout character signal S7 output from the character cutout section (6) as it is to identify the character B (
By registering image data such as feature amounts and dot patterns in step 7), accurate recognition can be performed from the next time onwards.

However, for example, for separated characters such as irregular fonts, the basic rectangular cutout character signal S7 output from the character cutout B (6) is the cutout character signal of each part constituting the separated character, so the image data of the separated character is Even if you try to register it in the character recognition section (7) using the learning function, what is registered is image data for each part that makes up the separated character, and the entire separated character is regarded as one character. Image data could not be registered.

Therefore, for separated characters such as irregular fonts, even if the extracted character signals are integrated in character identification B (7), the feature amount of the image data of this integrated character signal and its character identification part (
7) does not match the feature amount of the image data recorded in advance and is always erroneously recognized, making it difficult to increase the identification rate of separated characters.

In view of the above, an object of the present invention is to improve the identification rate of separated characters in a character recognition device without using a large-scale dictionary creation device.

[Means to solve the problem]

The character recognition device according to the present invention has a character cutting unit (24) that sequentially cuts out cutout character signals S7 forming each constituent part of one character or a separated character from an original character signal 51 corresponding to an image of an original document (14). .25), a recognition dictionary section (30) that records image data corresponding to the character code, and a recognition dictionary section (30) that records the character code corresponding to one or more of the extracted character signals S7. a character recognition unit (28) for identifying characters using the
5) a dictionary creation unit (29) that registers the image data of the cut-out character signal S7 supplied from the recognition dictionary unit (30), and the character identification unit (28) identifies the separated characters. If the identification is incorrect, the cut-out character signal S7 forming the entire separated character for which the identification was incorrect is sent to the character cut-out portion (
24. 25) to the dictionary creation section (29), thereby registering the image data of the incorrectly identified separated character in the recognition dictionary section (30).

[Effect]

According to the present invention, the cut-out character signal S7 forming the entire separated character that has been incorrectly identified is sent to the dictionary creation unit (29).
Therefore, the entire image of the separated character that has been misidentified is registered in the recognition dictionary section (30). Therefore, the next time a separated character that has been incorrectly identified is to be identified, the character identification unit (28) integrates the cut-out character signals S7 forming each constituent part of the separated character and uses the recognition dictionary fl'E. (30), the separator character can be accurately identified.

Therefore, the identification rate of separated characters can be increased.

〔Example〕

Hereinafter, the first embodiment of the character recognition device according to the present invention will be described.
This will be explained with reference to FIGS.

Figure 1 shows the system configuration of the character recognition device of this example. In Figure 1, (13) is a scanner consisting of a document feeder and an image league. A scanner (14) is a document set in this scanner. (13
) is the entire page of manuscript (14), for example 400 x 4
7 with a reading density of 00dpi (dots/inch)
The original character signal S1 corresponding to the shading of each dot is generated.

(15) is an image data input/output board, (16) is a host computer, (21) is a printer, and the image data input/output board (15) supplies a predetermined portion of the original character signal S1 to the host computer (16). At the same time, a printing signal output from the host computer (16) is supplied to the printer (16). (17) is a keyboard for imitating the host computer (16);
(18) is a coordinate input unit for inputting various coordinates to the host computer (16), and (19) is a character identification board, in which the host computer (16) extracts the circumscribed frame of one character from the original character signal S1. The basic rectangular cutout character signal S7 cut out using the cutout signal with the internal high level "1" is sequentially passed through the character identification board (19).
), the character identification board (19) supplies the character code C of the character corresponding to the cutout character signal S7 (or the jump code if the character cannot be recognized) to the host computer (16).

(20) shows a display device made of a cathode ray tube, and a predetermined area of the display screen of this display device (20) has r! K draft (14
) is displayed in a format corresponding to the original (14). The display screen of this display device (20) also displays one part of the original (14) as needed.
The dot pattern itself for a page or a predetermined portion can be displayed.

FIG. 2 shows a more detailed configuration of the printer (21) shown in FIG. 1, excluding the processing part. In the image data input/output board (15) shown in FIG. The image data manual section (23) has a memory that can store dot patterns for more than one page, and (23) is a backup memory that can also store dot patterns for more than one page of the original (14).
3) Store the original character signal S1 for one page of the original document (14) outputted in the image data manual section (22).

In addition, a desired part of the original character signal S1 stored in the image data manual section (22) is transferred to the backup memo 'J (23) at any time, and the desired part corresponds to a blank sheet with no characters. (e.g. zero level "0"
The original character signal S1 with this desired portion replaced is referred to as an original character signal S2, and a predetermined portion of this original character signal S2 is referred to as an original character signal S3.

In the host computer (16), (24) is a central processing unit (hereinafter referred to as rCPU.), (25)
) is the main memory, (26) is a video signal RAM for the display device (20) (hereinafter referred to as VRAMJ), and (27) is a character code input and a predetermined memory corresponding to this character code. It shows a font table consisting of an Aractor ROM that outputs dot patterns of fonts, that is, fonts, and an operator inputs various commands, data, and When the coordinate data is supplied, CP
Correspondingly, U(24)' controls the overall operation of the character recognition device of this example.

In addition, the original character signals S2 and S3 can be recorded as a main memo at any time.
J (25) and V R A M (26).

In this case, cPU (24) and main memory (25
) corresponds to the character string extraction part (2) and character extraction part (6) in the example in Figure 6, and the basic rectangle extraction corresponds to the inside of the circumscribing frame O of one character that is successively outputted from the main memo') (25). The character signal S7 is sequentially supplied to the character identification board (19), and the character code C transmitted from the character identification board (19) is supplied to the address bus of the font table (27) via the main memory (25). , the font data appearing in the data path of this font table (27) is written into a predetermined area of the VRAM (26). Furthermore, the system comprising the host combinator (16), keyboard (17), coordinate system (18), and display device (20) of this example also has the function of a word processor.

In the character identification board (19), 28) indicates a character recognition section, (30) indicates a recognition dictionary section that stores font data of various fonts in correspondence with character codes (JIS codes in this example), The recognition section (28) and the recognition dictionary section (30) basically correspond to the character identification section (7) in FIG. The recognition dictionary section 30) of this example is divided into a large classification dictionary section for large classification characters and a subclassification dictionary section for small classification characters, and the main classification dictionary section has first features classified by position as described above. Relative size of characters and circumscribing frame, aspect ratio h/w
), vertical relative ratio h/h, and horizontal relative ratio W/WR of the second characteristic character, for example, 24 dots in height x 2 dots in width.
Font data normalized to 4 dots is stored respectively. Generally, the general characteristics of a character are also expressed by the dot patterns near each side of the circumscribing frame of the character, so the dot patterns near the four sides of the circumscribing frame of each character are digitized as four-side data (or peripheral data). Then, font data of characters (major classification characters) whose four side data fall within a predetermined range may be stored in the major classification dictionary section.

On the other hand, the minor classification dictionary section stores normalized font data of all other characters (minor classification characters) not included in the major classification dictionary section in correspondence with the character codes.

(29) indicates a dictionary creation section, and this dictionary creation section (29)
When the dictionary creation mode is set by the operator, determines whether the font data represented by the supplied basic rectangular cutout character signal for one character corresponds to a major classification character or a subclassification character. When corresponding to major classification characters, the font data is normalized and written to a predetermined character code area of the major classification dictionary section of the recognition dictionary section (30), and when corresponding to subclassification characters, the font data is normalized. and write it into a predetermined character code area of the subclassification dictionary section of the recognition dictionary section (30). This allows the user to easily create a recognition dictionary section (30) that can handle various fonts.

When the character recognition mode is set by the operator, the recognition section (28) of the character identification board (19) changes the size of the recognition dictionary section (30) when the supplied basic rectangular cutout character signal S7 corresponds to a major classification character. The font data of the classified character section is sequentially processed in first-in, first-out (FIFO) format.
When the font data corresponds to subcategory characters, the font data of the subclass character section is sequentially written to the second FIFO register. Also, in parallel with this operation, the recognition unit (28)
normalizes the dot pattern corresponding to the basic rectangular cutout character signal S7 and sequentially writes it into the third FIFO register. The recognition unit (28) then uses the dot pattern of the character to be recognized in the fi2 register in the third FIFO register, the series of font data in the first FIFC5 register, and the second FIFO register.
By sequentially comparing a series of font data in the FIFO register of , 10 character codes corresponding to the font data closest to the dot pattern of the character to be recognized are generated in descending order of priority, and these character codes are is written into a predetermined area of the main memory of the host computer (16).

To determine the priority order, for example, for each dot of 24 x 24 dots, the dont pattern of the character to be recognized is compared with the font data read from the recognition dictionary section (30), and the values of the two are found to be different. The total sum of dots is taken as the evaluation value, and the order of priority is set in descending order of the evaluation value. If the evaluation value of the character code with the highest priority is less than a predetermined value, it is assumed that character recognition has been performed, and the character code with the highest priority is set as the character code C of the character to be recognized. Main memo! J
(25) is written in the area determined corresponding to the original (14). At the same time, the recognition unit (28) records the character code C, data indicating the size of the character, and data indicating the position of the character within the average circumscribing frame as a main memo! J
Do as written in (25). On the other hand, if the evaluation value of the character code with the highest priority exceeds the predetermined value, the recognition unit (28) considers that the character could not be recognized and writes the reject code to the main memo IJ (25). It is written in a predetermined area corresponding to the original (14).

The character recognition operation described above is executed at high speed using the Vibration method.

Figure 3 shows the display screen (2OA) of the display device (20) of this example.
This display screen (20^) is provided with a recognition result display area (31) for displaying the recognition results of one page of characters of the original (14). In this example, the original (14) is written vertically, but the recognition results are displayed horizontally. (
32) indicates a cursor for indicating the character to be corrected, and this cursor (32) can be moved over any character in the recognition result using the keyboard or the coordinate system (18). .

Example in Figure 3 When the cursor (32) is moved over the blank that indicates the character with the reject code, the display screen (2
The words and flanks of the seven target characters are displayed in areas (34) and (35) adjacent to the recognition result display area (31) of OA>, respectively. (36) indicates the function button area, and the characters "Candidate", "Code", and "Kana-Kanji" are displayed in this area (36), and these function buttons are selected by the coordinate person (18). can do. For example, when "Candidate" is selected, the display screen (2OA)
1 recognized as being closest to the character to be corrected
Patterns of 0 characters are displayed in order of priority,
If ``Code'' is selected, the character to be corrected can be specified directly by the character code, and if ``Kana-Kanji'' is selected, the character to be corrected can be manually converted into Kana/Kanji.

In addition, (37) indicates a re-recognition button, and (38) indicates a character integration button. By selecting these buttons (37) and (38) with the coordinate input unit (18), the characters to be corrected can be re-recognized, respectively. and separate characters can be integrated. Each button of the above-mentioned function button area (36) and button (37),
Specifically, the selection of (38) is made using the cross cursor (43).
This is executed by moving the coordinate character 7N18) onto the desired button and then moving the coordinate character character 7N18) to the desired button.

In this example, a peripheral image display area (33) with an area of W x W is provided near the recognition result display area (3l), and this peripheral image display area (33) displays the document ( 14) Display the dot pattern as it is in an area of 170 dots wide x 170 dots high with the character "e" above as the center. Specifically, in FIG. 2, the CPU (24) is an image data input unit (22
) is stored in the original character signal S2 for one page (in this example, the original character signal S1 itself), the original character signal S3 of a 170 x 170 dot area centered on the character to be corrected is read out, and this The original character signal S3 is enlarged (interpolated) or reduced (thinned) and written in a predetermined area of V R A M (26>. Therefore, the peripheral image display area (2OA) in the display screen (2OA) in FIG. 33) contains the manuscript (14) corresponding to the reject characters to be corrected.
The letters above. Images of, for example, eight characters around this character are displayed as they are as a donot pattern.

When the image of the character to be corrected and the surrounding characters is displayed in this way, the operator can correct the character to be corrected according to the characters before and after it without referring to the manuscript (14). This has the benefit of improving the efficiency of correction.

Furthermore, in FIG. 3, (39) indicates a cutout character pattern display area, and in this display area (39), a normalized 24×24 dot pattern corresponding to the basic rectangular cutout character signal S7 of the character to be corrected is displayed. Display the dot pattern. Furthermore, in this example, when the character integration button (38) is selected, an integrated character area (40) is secured at the left end of the display screen (2OA), and this integrated character area (40) is secured at the left end of the display screen (2OA).
) A registration button (41) and a delete button (42) are displayed at the bottom of the screen. The characters displayed in this integrated character area (40) correspond to the image data stored in the character integrated banner provided in a part of the main memo IJ (25) in FIG.

Can you correct the recognition conduit using the correction editor function of the character recognition device in this example? To explain the operation when performing integrated registration of separated characters when writing, since the manuscript (14) in this example is written vertically, the target separated characters are characters that would be determined as separated characters when written vertically. It is assumed that the character ``e'' is used in this manuscript (14), and that the font of the character ``e'' used in this manuscript (14) is an irregular font that is not registered in the recognition dictionary section (30) in FIG.

In this case, the character ``e'' cannot be recognized as a character as a whole, so the character ``e'' is separated into two parts, the upper part becomes 口・'', and the lower part becomes The display screen (2OA) should be considered to recognize the constituent parts as unrecognizable characters (characters with reject codes).
In the recognition result display area (31) inside, as shown in Figure 3, the position where "e" should originally be displayed (44A>,
(44B) are displayed with "." and a blank, respectively. Next, the operator moves the cursor (32) to position (44).
By moving the character over the blank in B) and operating the coordinate manual switch, the character corresponding to the blank is designated as the character to be corrected, and the character is displayed in the peripheral image display area (33) and the cut-out character pattern display area ( 39), an image of the periphery of the character on the document (14) corresponding to the blank and an image of the character itself are displayed.

These images let the operator know that the separated character "e" in the original (14) was mistakenly recognized as being separated into the first block (45A) and the second block (4513). .

Therefore, in order for the operator to integrate the image data of the separated character "e" in the irregular font and register it in the recognition dictionary section (30), first the position (44A) of the recognition result display area (31) is
) to be corrected and press the character integration button (38
) is selected, the original pattern (46A) of that "." is displayed in the integrated character area (40) on the display screen (2OA). Next, when the operator specifies the blank at position (44B) as the correction target and selects the character integration button (38), the pattern (46A) in the integrated character area (40)
The lower part of the blank is the original pattern (46B>
are combined and displayed. Then, by moving the cursor (43), selecting the registration button (41) in the integrated character area (40), and inputting the corresponding JIS code, etc., the operator can select the entire separated character 1e. The cutout character signal S7 that forms is the main memo! J (25) to the dictionary creation unit (29), and the entire image data (normalized data, major classification data, subclassification data, etc.) of the separated character set is supplied to the recognition dictionary unit (30). ) will be registered. In addition, the delete button (
42>, the integrated character area (40
) can be progressively deleted.

Next, the steps shown in Figure 4 (
To explain in detail with reference to 101) to (106),
Coordinate system of manuscript (14), main memo of separated character "E"'
Main memo 'I' to integrate the memory map of the image at the input stage to IJ (25) and its separated character 'e'
The integrated blocks (48) stored in the character integration buffer area of J (25> are shown as shown in FIG. As shown in Figure 5, the circumscribing frames of the constituent parts are the first block (47A) and the second block (47B).
), and the block (47A
), (47B) and (48) starting point Q,, Q2
, R coordinates (x+, y+), (X2, y2) respectively
and (X3, ys), block (47A),
(47B) and (48) (length in the X direction, length in the Y direction) are (Wl, h l) and (W2.
h2) and (W3, h3).

Step (101) The operator instructs the CPU (24) to select the first block (47^) and register it in the character integration buffer area of the main memo IJ (25).

One step (102) C P 'IJ (24) is its first block (47A
) is regarded as the integrated block (48). That is, (X3. Y3. W3. h3) = (X+, 3
'l+w,,h+)++-(x) is established, and the dot pattern image of this integrated block (48) is displayed on the integrated character area 40) of the display screen (2OA).

-Step (103) The operator selects the second block (47B) and writes the main memo! Instructs CPU (24) to additionally register in the character integration buffer area of J (25).

Step (104) CPU (24) is the rectangular first block (47A)
and the second rectangular block (47B).
1 block (47A). (47B) A rectangular integration block (48) circumscribing is synthesized. That is, first, the coordinates (X3, Y3) of the starting point R of the integrated block (48) are determined by the following equation.

X 3 = min( x l. X 2)
= = (2A)y:+=m+n(y+, y2
)...12B) It is assumed that the minimum unit of coordinates (X, Y) in this example is 1×1 dot.

Next, in blocks (47A) and (47B), the origin (
If the coordinates of the farthest part in the X and Y directions from 0.0> are xsE and y3E, respectively, then these coordinates (x3E
,ysE) can be expressed as follows.

x3E=max(x, +w+ -1, X2+W2-1
)-=(3A>37+E=maX(3/++ h+
1, V2,000h , -1) - - (3El) Therefore,
The coordinates of the starting point R of equations (2A) and (2B) (X:l+
ys) of the integrated block (48).
The length W3 in the direction and the length h in the Y direction can be expressed by the following formula.

W,=X3E X:l+1. =
=<4A)h:r=V3E Y321
...(4th day) - Step (105) - CPU (24> is the main memory (25) corresponding to the integrated block (48>) obtained in step (104)
The cutout character signal S7 stored in
By supplying the separated character "e" to the dictionary creation section (29) together with the S code, the separated character "e" is registered in the recognition dictionary section (30), and then the process moves to the final step (106).

In addition, if the separated character has three or more constituent parts, (
For example, the character "three"), a step of integrating a third blog, etc. is required.

According to this example as mentioned above, the main memo! J (25)
An integrated block (48) is constructed by integrating the blocks circumscribed to each constituent part of the desired character in the character integration batho area of By supplying the image data to the dictionary creation unit (29), the image data can be registered in the recognition dictionary unit (30) through a learning action, so no matter how complex the separated characters are, the image data can be input to the recognition dictionary. Additional registration can be made to jl'l3 (30). Therefore, since the separated character can be accurately recognized from the next recognition, there is an advantage that the identification rate of the separated character can be increased.

Furthermore, in this example, since the correction editor for correcting the recognition result performs the above-mentioned character integration, there is no need to use a large-scale dictionary creation device.

It should be noted that the present invention is not limited to the above-described embodiments, and it goes without saying that various configurations may be adopted without departing from the gist of the present invention.

〔Effect of the invention〕

According to the present invention, there is an advantage that the identification efficiency of the separated prison can be easily improved.

[Brief explanation of drawings]

FIG. 1 is a front view including a partial perspective view showing the stem configuration of a character recognition R device according to an embodiment of the present invention, and FIG. Structure including partial perspective view
3: Figure 3 is a front view showing a configuration example of the display screen (2OA) of one embodiment, FIG. 4 is a flowchart showing the registration operation of separated characters in one embodiment, and FIG. 5 is one embodiment. 6 is a block diagram showing the overall configuration of a conventional character recognition device, and FIGS. 7 and 8 are diagrams showing conventional character strings and cutting out original rectangles, respectively. FIG. 3 is a diagram for explaining the operation. (13) is a scanner, (14) is a document, (15) is an image data input/output board, (16) is a host computer, (19) is a printer identification board, (20) is a display device, (24) is a central processing unit, (25) main memory, (28) recognition unit, (29) recognition dictionary unit,
(38) is a character integration button, and (48) is an integration block. Agent Hidemori Matsukuma Figure 4 Figure 5

Claims (1)

[Claims] a character extraction unit that sequentially extracts extracted character signals forming each component of one character or separated characters from an original character signal corresponding to an image of an original document; and a recognition dictionary unit that stores image data corresponding to character codes. a character identification unit that uses the recognition dictionary unit to identify a character code corresponding to one or more of the cutout character signals; and a character identification unit that recognizes the image data of the cutout character signal supplied from the character cutout unit. and a dictionary creation section that registers the separated characters in the dictionary section, and when the character identification section makes a mistake in identifying a separated character, the cutout character signal forming the entire separated character that was misidentified is sent from the character cutting out section to the above. By supplying it to the dictionary creation section,
A character recognition device characterized in that image data of the separated character that has been incorrectly identified is registered in the recognition dictionary section.