JPH09238247A - Optical character reader - Google Patents

Abstract

PROBLEM TO BE SOLVED: To conduct character recognition processing efficiently by eliminating a half tone screen pattern on a document by means of a half tone screen elimination means in the pre-processing of character recognition especially with respect to the optical character reader. SOLUTION: The reader is provided with a scanner section 11, a storage means 12 storing image information read by the scanner section 11, a resolution adjustment means 13a adjusting the resolution of an image read by the scanner section 11, a comparator means 14a comparing the image data recorded in the storage means 12 with image data read from the scanner section 11, the image data whose resolution is set lower than the resolution of the image data stored in the storage means 12 are compared with the image data stored in the storage means 12 by the comparator means 14a to extract and eliminate a half tone screen picture element formed in a read area of a document.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shading removing means in preprocessing of character recognition of an optical character reader (OCR).

In recent years, OCR has become widespread, in which characters written on a form are read by a scanner or the like, the read image is subjected to recognition processing, and used for input to a computer.
With the spread of word processors and personal computers in recent years, it is possible to print on plain paper in a free form format without using a dedicated OCR form (for example, a form in which the background color of the paper described later is different from the character color). There is a demand for reading the characters written in the prepared form.

When reading a form, in addition to characters to be used as input to the computer, image data that is unnecessary for recognition processing, such as stains on the form, ruled lines and patterns, is also included. Since these unnecessary images increase the recognition processing time and cause erroneous recognition, they need to be removed before the recognition processing.

[0004]

2. Description of the Related Art FIG. 10 is a diagram for explaining a conventional optical character reader. In FIG. 10, 1 is an optical character reading device, and 2 is a computer for inputting characters read by the optical character reading device.

In the conventional optical character reader 1, as described above, the OCR form 10 uses a dedicated OCR form in which the background color of the paper is different from the character color, or is used as a recognition target on the form 10. Area A to be stored in a specific file storage device 1
The ruled lines and patterns outside the area A are not read by the scanner 11 by using the information indicating the area A stored in the specific file storage device 13 in advance.

[0006]

However, forms created by a word processor or a personal computer are sometimes shaded with dots or fine lines so that the person who is writing can easily recognize the writing area. When the characters written on the screen are read by a scanner etc., the recognition process time increases due to the increase of image elements due to the shaded patterns, and the characters written with the shaded patterns overlap. There was a problem that could not be recognized as.

As a similar technique, there is Japanese Patent Laid-Open No. 5-135203, "Document image processing system and document image processing method". In Japanese Patent Laid-Open No. 5-135203, “Document image processing system and document image processing method”, in summary, “a document image on a form is recorded as a reference image before the form is read, and the form is recorded on the form. By comparing with the document image after writing characters, the reference document image such as the shading is removed. "If the reference image is not registered, the shading can be removed. I can't do it
There is a problem that a large-capacity storage device for recording a large number of reference images is required to handle many types of forms.

In view of the above-mentioned drawbacks of the prior art, the present invention provides a pre-recognition processing means capable of efficiently recognizing characters by removing halftone patterns on a form without providing a large-capacity storage device. It is intended.

[0009]

1 to 3 are diagrams showing the principle of the present invention, FIG. 1 shows an example of the principles of the present invention, and FIG. 2 shows the present invention. 3 shows an example of the principle configuration of claim 3, and FIG. 3 shows an example of the principle configuration of claims 4 and 5 of the present invention. The above problems can be solved by the optical character reader configured as follows.

(1) An optical character reader 1, which is a scanner unit 1 for reading an image and converting an optical signal into an electric signal
1, storage means 12 for recording the image information read by the scanner section 11, and first resolution adjusting means 13a for adjusting the resolution of the image read by the scanner section 11.
And an image data recorded in the memory means 12 by the first resolution adjusting means 13a, the image data recorded in the memory means 12 being compared with image data read from the scanner section 11. The resolution is lower than that of the data, and the image data read from the scanner unit 11 and the image data recorded in the storage unit 12 are compared by the comparison unit 14a to form a network formed in the reading area of the form. It is configured to extract and remove the false image elements.

(2) The optical character reader 1, which is a scanner unit 1 for reading an image and converting an optical signal into an electric signal
1, storage means 12 for recording the image information read by the scanner section 11, and second resolution adjusting means 1 for adjusting the resolution of the image read by the scanner section 11.
3b and a comparison means 14a for comparing the image data recorded in the storage means 12 with the image data output from the second resolution adjustment means 13b, and the storage means by the second resolution adjustment means 13b. Image data having a resolution lower than that of the image data recorded in 12, and the storage means 1
By comparing the image data recorded in 2 with the comparison means 14a, the shaded image elements formed in the reading area of the form are extracted and removed.

(3) The optical character reader 1, which is a scanner unit 1 for reading an image and converting an optical signal into an electric signal
1, a storage unit 12 for recording image information read by the scanner unit 11, a density conversion unit 15 for performing density conversion of the recorded image data, and image data density-converted by the density conversion unit 15. And a comparison means 14b for comparing the image data recorded in the storage means 12 with the image data density-converted by the density conversion means 15 and the image data recorded in the storage means 12 The comparison means 14b is configured to extract and remove the shaded image elements formed in the reading area of the form.

(4) The optical character reader 1, which is a scanner unit 1 for reading an image and converting an optical signal into an electric signal
1, a storage unit 12 for recording the image information read by the scanner unit 11, and an extraction unit for extracting a predetermined isolated point from the image data recorded in the storage unit 12.
As the extraction means 16, the positional relationship between neighboring isolated points is made into a function, the position of the isolated point is recognized by using the functionalized positional relationship between the isolated points, and the image of the isolated point is obtained. An arithmetic means 14c for extracting and removing isolated points generated by shading from the data and the image data recorded in the storage means 12 is provided.

(5) An optical character reader 1, which is a scanner unit 1 for reading an image and converting an optical signal into an electric signal
1, a storage unit 12 for recording the image information read by the scanner unit 11, and an extraction unit for extracting a predetermined isolated point from the image data recorded in the storage unit 12.
As the extraction means 16, the positional relationship between neighboring isolated points is functionalized, the position of the isolated point is recognized by using the functionalized positional relationship between the isolated points, and an image of the isolated point is provided. Calculating means 14c for extracting data and extracting and removing isolated points generated by shading from the image data recorded in the storage means, and extracting and removing isolated points generated by shading by the calculating means 14c It is configured to include a complementing means 17 for complementing the loss of image data caused by the above.

That is, FIG. 1 (a) is a principle configuration diagram of claim 1 of the present invention. In the present optical character reading apparatus, first, image information is read from the form 10 in the scanner section of FIG. 1 (a). type in,
Record in storage means 12. The first resolution adjusting means 13a controls the reading resolution of the scanner section 11. The comparison unit 14a compares the image data input from the scanner unit 11 and the storage unit 12 (for example, by logical product).
Then, the result is output.

FIG. 1 (b) is a principle configurational diagram of claim 2 of the present invention, which is a second resolution adjusting means 13b for adjusting the resolution of an image read by the scanner section 11 and the storing means 1
Comparing means for comparing the image data recorded in 2 with the image data output from the second resolution adjusting means 13b.
The image data recorded in the storage means 12 is compared with the image data recorded in the storage means 12 by the second resolution adjusting means 13b and the image data recorded in the storage means 12 is compared by the comparison means 14a. The result is output (for example, by logical product).

FIG. 2 is a principle block diagram of claim 3 of the present invention, in which the image data stored in the storage means 12 and the image data density-converted by the density conversion means 15 are compared by the comparison means 14b ( For example, the result is output (by logical product).

FIG. 3 shows claims 4 and 5 of the present invention.
It is a principle block diagram of. Storage means 12 by extraction means 16
The isolated points in the image data recorded in are extracted, and the isolated points extracted from the original image data are removed and output by passing through an arithmetic means (for example, by exclusive OR) 14c. is there. In addition, the lack of the image data of the character caused by removing the isolated points is complemented.

Generally, the number of pixels constituting a pattern used for shading is smaller than the number of pixels occupied by characters,
It also has the feature that the size of the image is uniform. Therefore, if the resolution for reading an image is lowered, the shaded pattern cannot be read below a certain resolution according to the size of the shaded pattern, and the number of reading elements is drastically reduced.

Claims 1 and 2 of the present invention make use of the above characteristics, and in claim 1, the image resolution of the scanner unit 11 is changed, and
According to the second aspect, the scanner unit 11 reads and stores the data.
The halftone pattern is erased by changing the resolution of the image data recorded in (1) or by subjecting the read image data to density conversion (for example, known affine conversion). Specifically, the comparison unit 14
In a and b, the image element generated by shading is removed by taking the logical product of the image data after shading pattern deletion and the original image data.

Further, the shaded pattern is a rule such as a group of parallel lines having even intervals, a group of parallel lines intersecting at a constant angle, or an intersection of groups of parallel lines intersecting at a constant angle. Characteristic patterns are generally used. Claim 3 of the present invention utilizes the above characteristics. For example, according to claim 1, 2 or 3 of the present invention, the positional relationship of the image elements of the shaded pattern is extracted from the extracted image data. The regularity is converted into a function to remove the image element at the position where the shaded pattern should be.

Further, in claim 5 of the present invention, for example, when a part of the image elements of the halftone dot pattern detected by using the above claim 4 of the present invention overlaps with the character image, halftone dot When the image element of the pattern is removed, it is complemented that a part of the original character image is lost.

Therefore, according to the optical image reading apparatus of the present invention, there is no need to use a dedicated OCR form as in the prior art and to store the image data of the form and the form area in advance. , Even if it is image data of a character written in a shaded area, only the character image data can be extracted, the processing time of the character input processing can be significantly shortened, and the recognition error can be prevented. It greatly contributes to the prevention of unrecognizable characters.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. FIG. 1 to FIG. 3 described above are principle configuration diagrams of the present invention, FIG. 4 to FIG. 9 are diagrams showing an embodiment of the present invention, and FIG. 4 shows the second resolution adjusting means. 5 shows an example, FIG. 5 shows an example of removing the shaded pattern by adjusting the resolution, specifically, lowering the resolution, and FIG. 6 shows an example of removing the shaded pattern by density conversion. FIG. 7 schematically shows an example of an extracting means for removing the shaded pattern by functionalizing the positional relationship of the shaded pattern, and FIGS. 8 and 9 show the result of removing the image element caused by the shaded pattern. 9 schematically shows an example of complementing the missing image data.

In the present invention, the optical character reader 1
A scanner unit 11 for reading an image and converting an optical signal into an electric signal; a storage unit 12 for recording image information read by the scanner unit 11; and a first unit for adjusting the resolution of an image read by the scanner unit 11. The first resolution adjusting means 13a includes a first resolution adjusting means 13a and a comparing means 14a for comparing the image data recorded in the storing means 12 with the image data read from the scanner section 11. The resolution is lower than that of the image data recorded in 12 and the image data read from the scanner unit 11 is compared with the image data recorded in the storage unit 12 by the comparison unit 14a to read the form. Means for extracting and removing shaded image elements formed in the area, or density conversion for performing density conversion of the recorded image data. The image data density-converted by the density conversion means 15 is provided with a conversion means 15 and a comparison means 14b for comparing the image data density-converted by the density conversion means 15 with the image data recorded in the storage means 12. By comparing the data and the image data recorded in the storage means 12 by the comparison means 14b, a means for extracting and removing the shaded pixel elements formed in the reading area of the form, and the like are provided. It is a necessary means to carry out. Note that the same reference numerals indicate the same object throughout the drawings.

The configuration and operation of the optical character reader of the present invention will be described below with reference to the principle configuration diagrams of FIGS. As shown in FIGS. 1 to 3, the optical character reader 1 according to the present invention has an overall configuration example in which a scanner section 11 for reading an image and converting an optical signal into an electric signal and a scanner section 11 for reading the image are read. Storage unit 12 for recording image information, and in the example shown in FIG. 1, a first unit for adjusting the resolution of an image read by the scanner unit 11 is provided.
Resolution adjusting means 13a, and comparing means 14a for comparing the image data recorded in the storage means 12 with the image data read from the scanner section 11, or
Second resolution adjusting means 13b for adjusting the resolution of the image read by the scanner section 11, image data recorded in the storing means 12, and the second resolution adjusting means 13b.
It is provided with a comparison means 14a for comparing the image data output from the.

FIG. 4 shows an embodiment of the second resolution adjusting means 13b shown in FIG. 1, and is a block diagram of an embodiment of claim 2 of the present invention. In this figure, 110
Is a photoelectric conversion element such as a CCD used in the scanner section 11. The output of the CCD 110 is read as a digital signal and, for example, a black pixel is output as "1" and a white pixel is output as "0". In the present embodiment, the second resolution adjusting means 13b is realized by taking the logical product of the outputs of the adjacent CCDs 110, and the resolution is halved by the 2-input logical product.
In addition, the resolution can be reduced to 1/3 by the 3-input logical product. The enable lines A and B are, for example, control signal lines for selecting whether to reduce the resolution to 1/2 by the 2-input logical product or to reduce the resolution to 1/3 by the 3-input logical product. In the figure, 13b is a specific configuration example of the AND circuit, which has a configuration in which a pixel signal obtained by ANDing the outputs of the adjacent CCDs 110 is expanded to the original number of pixels by the OR circuit and output. By doing so, the original image data stored in the storage means 12 can be compared with the comparator 14a.

FIG. 5A shows an example of image data read by the CCD 110, and there are an image element 11a generated by halftone and an image element 11b generated by an original character image. When this image is read by taking the logical product of the CCDs 110 adjacent in the left-right direction by the second resolution adjusting means 13b shown in FIG. 4, the image data of FIG. 5B is obtained. By reducing the resolution to 1/2, the image element 11a in the shaded pattern cannot be read and the image element in the character image
Only 11b is read. By inputting the image data of FIGS. 5A and 5B to the comparison unit 14a and taking the logical product of the pixels, only the image element 11b of the character image can be extracted, and the exclusive OR is taken. As a result, the image element 11a in the shaded pattern can be extracted.

Although the logic circuit is used for the resolution adjustment in the above embodiment, the resolution can be adjusted by changing the scanning speed of the CCD 110, for example. This is a structure in which the CCD 110 forms pixel data of white pixels and black pixels by accumulating photons from a form, and therefore, for example, by increasing the scanning speed, the amount of light is small and small. Is not readable (that is, the accumulated charge is small and the output current is small), and as a result, the reading resolution is reduced. First in Figure 1 (a)
The resolution adjusting means 13a corresponds to means for reducing the resolution, specifically, the reading resolution by changing the scanning speed based on such a principle. The scanning linear velocity can reduce the resolution in the X direction (main scanning) and the Y direction (sub scanning) by increasing the scanning speed in the X direction (main scanning) and the Y direction (sub scanning). it can. {Examples corresponding to claims 1 and 2} Further, in the present embodiment, the shaded pattern is dot-shaped, but if the pattern used for the shade is sufficiently thin compared to the size of the character, The same processing can be performed even with shading by fine lines.

FIG. 6 is a diagram schematically showing an embodiment of claim 3 of the present invention. FIG. 6A shows a configuration example in which the density conversion means 15 shown in FIG. 2 is performed using a known affine transformation, and 15a is an image memory.
The image data A read from the storage means 12 is stored. This is subjected to density conversion using the affine transformation unit 15b, and the result B is stored in the image memory 15c.

Since the affine transformation is a mathematical density transformation, the adjustment of the pixel density is discretely and continuously specified by numerical values (for example, 10%, 15%, ..., 100% (state in which the density is not reduced)). } Can be done. The affine transformation includes enlargement, reduction transformation, rotation, density transformation (reduction → density calculation → enlargement), and is a means for deforming a figure by changing the size of a vector forming the figure.

FIG. 6 (c) shows the above affine transformation
9 shows an example of pixel density conversion when image data density conversion is performed. As described above, the density conversion by the affine transformation is performed by reducing → density calculation → expanding. Therefore, the area indicated by the double line is reduced to, for example, 1/4, and the predetermined density calculation is performed. When the image data is recognized as black or white, the image data of the original size is enlarged to perform the density conversion as shown in FIG. 6 (c).

In general, the size and density of pixels in a shaded pattern are more uniform and fine than those in a character image. Therefore, if the density conversion means 15 reduces the ratio of the image density (roughens the density), as shown in FIG.
As shown in (b), the number of image elements decreases sharply at a certain density. By detecting such a change in the number of image elements with a predetermined threshold value, as shown in FIG. 6 (b), the difference between the pixel density a and the pixel density b is caused by the meshing of image elements. Can be considered as.

Therefore, if the comparison means 14b shown in FIG. 2 obtains the logical product of the image data B at the pixel density b and the image data A at the original pixel density a, the halftone pattern can be removed. . On the contrary, if the comparing means 14b takes the exclusive OR of the image data B and the original image data A when the pixel density is b, only the halftone dot pattern can be extracted.

In the present embodiment, the shaded pattern is dot-shaped, but if the pattern used for shading is sufficiently smaller than the size of the character, the same processing may be performed even if it is a thin line. Can be applied.

FIG. 7 shows an example of the extraction means 16 of FIG.
The example of the extraction means which functionalizes the positional relationship of a shaded pattern and removes this shaded pattern is shown typically, and corresponds to the Example of Claim 4 of this invention. In Figures 7 (a) and 7 (b), 16
1 is an image element based on a half-tone dot pattern, and 165 is an image element based on a character image. When the neighboring image elements 161 of each image element 161 forming the above shaded pattern are connected, the intersections of the parallel line group α (162), the parallel line group β (163) and the parallel line group γ (164). Expressed as If each parallel line group is expressed as a function, the parallel line group α (162) is y = an (a is a constant, n = 1, 2, 3, ...) The parallel line group β (163) is y = bx + n The parallel line group γ (164) can be expressed as y = −cx + n.

In FIG. 7 (a), for example, when the shaded pattern is removed by the means described in claims 1 to 3 of the present invention, the result is as shown in FIG. 7 (b). Although 161a is an image element originally shaded, it is an image element that cannot be removed by the above-described resolution conversion, density conversion, etc. because it is close to the character image.

When trying to recognize the image of FIG. 7 (b),
The image element 161a, which should not be in the original character, may be erroneously recognized or may not be recognized. Therefore,
In FIG. 7A, since the image element 161a is located at the intersection coordinates of the plurality of straight line groups represented by the function, the image element 161a can be removed.

Specifically, the calculating means 14c shown in FIG. 3 is used.
Then, by performing an exclusive OR operation,
161a can be removed. In this embodiment, the shaded pattern is dot-shaped, but the same process can be performed even if the pattern used for the shade is a thin line.

FIG. 8 and FIG. 9 schematically show an example of complementing the loss of image data caused by the removal of image elements caused by shading, which is an embodiment corresponding to claim 5 of the present invention. FIG. In FIG. 8A, 165 is an image element by a character image, 161 is an image element by a halftone dot, and both are partially overlapping. The image element 161 by shading is, for example, the means described in claim 4 of the present invention (image data of only the image element 161 of the intersection coordinates, the image element 161 of the intersection coordinates and the image element 165 by the character image). Exclusive OR operation with image data including
When it is removed by using, the missing portion 161a is generated in the image element 165 which is originally a character. {See Fig. 8 (b)} This Fig. 8 (b)
When trying to recognize the character image shown in, the missing part 1
Due to 61a, it may or may not be recognized.

According to claim 5 of the present invention, for example, FIG.
As shown in (a) and (b), the image element 1 of FIG.
Image element by extracting 65 contours 165a and shading 1
An exclusive OR is performed with the contour 165b of the image element after removing 61 (see FIG. 8 (b)). Then, the partial contour line 1 in Fig. 9 (a) 1
61b and the partial contour line 161c in FIG. 9 (b) are intersection points 165c.
And intersect at intersection 165d, shaded image element 161
1610 {see FIG. 9 (c)} can be extracted. Here, the intersection 165c and the intersection 165d are complemented by, for example, a straight line,
Compensation for the missing portion 161a A prevents erroneous recognition in the recognition processing. In the above example, the contour 165a of the image element in FIG.
Although the explanation has been given by taking the example of taking the exclusive OR with the contour 165b of the image element after removal (see FIG. 8 (b)), the contour 165a of the image element of FIG. 8 (a) is extracted and Image element 1
The logical product of the contour 165b of the image element after removing 61 (see FIG. 8 (b)) and the contour 165b and the contour 165b of the image element of FIG. 9 (a) having the intersection 165c and the intersection 165d are obtained. ) Of the image element and contour 165a
The inside of the contour line generated by complementing the intersection 165c and the intersection 165d may be filled.

In this embodiment, the shaded pattern is in the form of dots, but the same processing can be performed even if the pattern used for shading is a thin line. As described above, according to the present invention, the scanner unit, the storage unit that records the image information read by the scanner unit, the resolution adjusting unit that adjusts the resolution of the image read by the scanner unit, and the image data recorded in the storage unit. And image data read from the scanner unit, the resolution adjusting unit lowers the resolution of the image data recorded in the storage unit, and the image data read from the scanner unit and the storage unit. By comparing the image data recorded in the means with the comparing means, the shaded pixel elements formed in the reading area of the form are extracted and removed. Instead of the resolution adjusting means, density conversion means for performing density conversion of image data, or from the image data recorded in the storage means, the positional relationship between neighboring isolated points is functionalized to extract the isolated points. It has an extraction means to do. or,
It is characterized in that the lack of image data caused by the removal of image elements caused by shading is complemented.

[0043]

As described above in detail, according to the present invention, it is possible to extract only character image data even if it is image data of a character written in a shaded area, and recognition is performed. This greatly reduces the processing time, and contributes greatly to the prevention of erroneous recognition and the occurrence of unrecognizable characters.

[Brief description of drawings]

FIG. 1 is a diagram showing the principle configuration of the present invention (part 1).

FIG. 2 is a block diagram showing the principle of the present invention (part 2);

FIG. 3 is a diagram illustrating the principle of the present invention (part 3);

FIG. 4 shows an embodiment of the present invention (part 1).

FIG. 5 shows an embodiment of the present invention (part 2).

FIG. 6 shows an embodiment of the present invention (part 3).

FIG. 7 shows an embodiment of the present invention (part 4).

FIG. 8 shows an embodiment of the present invention (part 5).

FIG. 9 shows an embodiment of the present invention (part 6).

FIG. 10 is a diagram illustrating a conventional optical character reading device.

[Explanation of symbols]

1 Optical character reader 10 Form 11 Scanner 110 Photoelectric conversion element such as CCD 11a Image element generated by meshing 11b Image element generated by character image 12 Storage means 13a First resolution adjusting means 13b Second resolution adjusting means 14a, 14b Comparator 14c Calculation means 15 Density conversion means 16 Extraction means 15a, 15b Image memory 161 Intersection points, intersection image, image element 162 Parallel line group α 163 Parallel line group β 164 Parallel line group γ 165 Character image, image element 161a Missing area, image element 161b, 161c Partial contour line 165a Contour line of character image 165b Contour line of missing character image 165c, 165c Intersection point of partial contour line 17 Complementary means a, b Pixel density

Claims (5)

[Claims] 1. An optical character reading device, comprising: a scanner unit for reading an image and converting an optical signal into an electric signal; a storage unit for recording image information read by the scanner unit; and an image read by the scanner unit. A first resolution adjusting means for adjusting the resolution of the image data and a comparing means for comparing the image data recorded in the storage means with the image data read from the scanner section. The resolution is made lower than that of the recorded image data, and the image data read from the scanner unit and the image data recorded in the storage unit are compared by the comparison unit to form in the reading area of the form. An optical character reader characterized by extracting and removing shaded image elements. 2. An optical character reader, comprising a scanner section for reading an image and converting an optical signal into an electric signal, a storage section for recording image information read by the scanner section, and a scanner section for reading the image information. A second resolution adjusting means for adjusting the resolution of the image; a comparing means for comparing the image data recorded in the storage means with the image data output from the second resolution adjusting means; The image data recorded in the storage unit is compared with the image data whose resolution is lower than that of the image data recorded in the storage unit by the resolution adjusting unit, and is formed in the reading area of the form. An optical character reader characterized by extracting and removing the shaded image elements. 3. An optical character reader, comprising a scanner section for reading an image and converting an optical signal into an electric signal, a storage section for recording image information read by the scanner section, and the recorded image. A density conversion unit that performs density conversion of data, a comparison unit that compares the image data density-converted by the conversion unit, and the image data recorded in the storage unit are provided, and the density conversion is performed by the density conversion unit. An optical character characterized by extracting and removing a shaded image element formed in a reading area of a form by comparing the image data with the image data recorded in the storage means by the comparison means. Reader. 4. An optical character reader, comprising a scanner section for reading an image and converting an optical signal into an electric signal, a storage section for recording image information read by the scanner section, and a storage section for storing the image information. From the image data, an extracting unit for extracting a predetermined isolated point is provided, and by the extracting unit, a positional relationship between neighboring isolated points is functionalized, and the functionalized positional relationship between isolated points is used,
The present invention is characterized by further comprising arithmetic means for recognizing the position of the isolated point, and extracting and removing the isolated point generated by shading from the image data of the isolated point and the image data recorded in the storage means. Optical character reading device. 5. An optical character reading device, comprising: a scanner unit for reading an image and converting an optical signal into an electric signal; a storage unit for recording image information read by the scanner unit; and a storage unit for storing the image information. From the image data, an extracting unit for extracting a predetermined isolated point is provided, and as the extracting unit, the positional relationship between neighboring isolated points is functionalized, and the functionalized positional relationship between isolated points is used,
An arithmetic means for recognizing the position of the isolated point and extracting and removing an isolated point generated by shading from the image data of the isolated point and the image data recorded in the storage means, and the arithmetic means. An optical character reading device comprising: a complementing unit that complements a loss of image data caused by extracting and removing isolated points caused by shading.