JP4435141B2 - Character recognition device, character recognition method, and character recognition program - Google Patents

Description

  The present invention relates to a character recognition device, a character recognition method, and a character recognition program for recognizing characters from image data when, for example, a form in which characters are entered in a shaded area is read by a scanner or the like.

  There is already a character recognition device that reads an image of a form with a scanner or the like and recognizes characters entered on the form from the image.

  In recent years, the character recognition accuracy of character recognition devices has improved due to improvements in computation speed and higher accuracy of character analysis programs. However, depending on the typefaces and typefaces of characters on the form, the character recognition accuracy may drop. Many. Hereinafter, a case where a character recognition device recognizes a character with respect to a shaded image (hereinafter referred to as “shaded image”) will be described.

  Normally, when characters are read from a shaded form with a character recognition device, character recognition processing is usually performed using an image after removing small connected components, that is, halftone dots scattered in portions other than characters. Do.

  As a conventional technique for removing halftone dots from such an image, for example, there is a technique for removing a specific spatial frequency component by performing a fast Fourier transform on a shaded form image (for example, Patent Documents). 1).

Further, as another prior art, for example, there is a technique for removing protrusions by tracking the outline of a character (see, for example, Patent Document 2).
JP 2000-293630 A JP 2003-67738 A

  In the case of the technique of the above-mentioned Patent Document 1, the amount of calculation is enormous even if it is fast Fourier transform, so that it is not suitable for business processing such as reading a form continuously. In addition, this technique has a drawback that an original character image is also removed because a specific spatial frequency component is removed.

  In the case of the technique disclosed in Patent Document 2, there is a problem that feature portions of minute characters that are separated from characters, such as muddy dots and periods, are also lost at the same time.

  The present invention has been made to solve such a problem, and can improve character recognition accuracy by removing halftone dots overlapping a character image included in a shaded image as faithfully as possible. An object is to provide a character recognition device, a character recognition method, and a character recognition program.

In order to solve the above problems, a character recognition device according to the present invention is a character recognition device for recognizing characters from a shaded image shaded by regularly dotted halftone dots, a storage device, and the shaded image. from detects the center position of the pixel connected components under size or less of the preset the dot, the positional information of the center position of the detected pixel connected components, the pre-Symbol pixel connected components were removed from the shaded image a first halftone dot elimination means for storing the first processed image in the storage device, the positional information of the any two adjacent pixel connected component of the position information read from the storage device, preset and other point position calculating means for calculating the position of the other pixel connected components adjacent to the two pixel connected components in accordance with the arrangement pattern or calculation rule of the halftone dot, the positions of the other pixel connected components the calculated, the First process A halftone dot determining unit for determining whether a pixel of an image is a white pixel or a black pixel, and a position within a predetermined diameter range that can be taken by the halftone dot, with the position determined as a black pixel by the halftone dot determining unit as a center Circumference calculation means for calculating coordinates, and whether or not the pixels of the first processed image on the circumference coordinates calculated by the circumference calculation means are divided into continuous white pixels and continuous black pixels Dividing the first processed image by connecting the dividing points of the pixels of the two colors determined to be divided by the dividing point determining means and the dividing point determining means for determining A second halftone dot removing unit that removes black pixels on the white pixel side to generate a second processed image; and a character recognition unit that performs character recognition on the second processed image. To do.

The character recognition method of the present invention is a character recognition method in a character recognition apparatus for recognizing a character from a halftone image that is regularly shaded by halftone dots. detecting the center position of the pixel connected components under size or less of the points, it stores the positional information of the center position of the detected pixel connected components, and the pre-Symbol pixel connected component first processed image is removed from the hatching image device and storing in an arbitrary position information of the two adjacent pixels connected component is read out from said storage device, said two pixels in accordance with the arrangement pattern or calculation rule preset dot of the position information Calculating a position of another pixel connected component close to the connected component, and determining whether the pixel of the first processed image at the calculated position of the other pixel connected component is a white pixel or a black pixel And a step of calculating a circumferential coordinate in a range of a predetermined diameter that can be taken by the halftone dot with the position determined as the black pixel as a center, and the first processing on the calculated circumferential coordinate A step of determining whether or not an image pixel has been divided into a continuous white pixel and a continuous black pixel, and connecting between the division points of the two color pixels determined to be divided into two. Dividing the first processed image, removing black pixels on the white pixel side of the circumference and generating a second processed image; and performing character recognition on the second processed image; It is characterized by having.

The program of the present invention is a program for operating a character recognition device that recognizes a character from a halftone image shaded by regularly dotted dots, the character recognition device comprising: a storage device; and the halftone image. from detects the center position of the pixel connected components under size or less of the preset the dot, the positional information of the center position of the detected pixel connected components, the pre-Symbol pixel connected components were removed from the shaded image a first halftone dot elimination means for storing the first processed image in the storage device, the positional information of the any two adjacent pixel connected component of the position information read from the storage device, preset and other point position calculating means for calculating the position of the other pixel connected components adjacent to the two pixel connected components in accordance with the arrangement pattern or calculation rule of the halftone dot, the positions of the other pixel connected components the calculated A halftone dot determination unit that determines whether a pixel of the first processed image is a white pixel or a black pixel, and a position that is determined as a black pixel by the halftone dot determination unit with a predetermined diameter that the halftone dot can take Circumference calculation means for calculating the circumference coordinates in a range, and pixels of the first processed image on the circumference coordinates calculated by the circumference calculation means are divided into two continuous white pixels and continuous black pixels. Dividing the first processed image by connecting between the dividing points of the pixels of the two colors determined to be divided by the dividing point determining unit; A second halftone dot removing unit that removes black pixels on the white pixel side of the circumference to generate a second processed image; and a character recognition unit that performs character recognition on the second processed image. It is characterized by.

  According to the present invention, it is possible to improve the character recognition accuracy by removing the halftone dot overlapping the character image included in the shaded image as faithfully as possible to the character.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a character recognition apparatus according to an embodiment of the present invention.

  As shown in FIG. 1, the character recognition apparatus includes an input unit 20 such as a scanner 22 that reads a form 21, and a computer 1 (hereinafter referred to as "PC1") that receives image data input from the input unit 20 and executes character recognition processing. ").).

  The PC 1 includes an input interface unit 11 (hereinafter referred to as “input I / F 11”), a first image processing unit 12 as a first halftone dot removing unit, a storage device 3, a halftone dot determination unit 13, and a circumference calculation unit 14. A segment point determination unit 15, a second image processing unit 16 as a second halftone dot removing unit, a character recognition unit 17, and an output interface unit 18 (hereinafter referred to as “output I / F 18”).

  The storage device 3 is realized by, for example, a RAM, a ROM, a hard disk drive device, and the like. The input I / F 11 and the output I / F 18 are one of the external interface devices of the PC 1, such as a USB interface, a LAN interface, and a D-sub terminal.

  The first image processing unit 12, the halftone dot determination unit 13, the circumference calculation unit 14, the segment point determination unit 15, the second image processing unit 16, the character recognition unit 17, and the like are blocks of software (program) functions. It is realized as one function of the central processing unit (hereinafter referred to as “CPU”) of the PC 1. The CPU reads the program code stored in the storage device 3 and realizes the functions of the above-described units. Each unit is not limited to software (program), and may be configured by independent hardware (functional components).

  The scanner 22 is an image input device that reads characters, figures, photographs, and the like from a form 21 on which characters to be read are written, and transfers (inputs) them as image data to the PC 1. The scanner 22 illuminates the surface of the form 21, reads the reflected light with a photoelectric conversion element such as a CCD, and converts it into digital data.

  The storage device 3 includes a processing target image storage unit 31, a halftone dot position information storage unit 32, a first halftone dot removal image storage unit 33, a second halftone dot removal image storage unit 34, a form definition storage unit 35, and dictionary data storage. Unit 36, recognition result storage unit 37, and the like.

  The processing target image storage unit 31 stores image data of the form 21 (hereinafter referred to as “shaded image”) input from the input unit 20 as a processing target image.

  The halftone dot position information storage unit 32 stores halftone dot position information detected from the halftone image.

  The first halftone dot removal image storage unit 33 stores a first halftone dot removal image as a first processed image obtained by removing halftone dots having a predetermined diameter (diameter) or less from a halftone image. A halftone dot is a set of black pixels (connected components of pixels) of about 10 pixels (0.6 mm) in the case of an image obtained by scanning the form 21 with the resolution of the scanner 22 set to about 400 dpi, for example. is there. This is an example, and the number of pixels may change depending on the resolution setting.

  The second halftone dot removal image storage unit 34 stores a second halftone dot removal image as a second processed image generated by removing halftone dots overlapping characters from the first halftone dot removal image.

  The form definition storage unit 35 stores format control data of the form 21 as form definition information. The format control data includes the timing mark position, the size of the form 21 (in mm), the type and size of the frame in which the character to be read is written, the number of characters, the character pitch, the character attributes (address, name, etc.), The output format (text, CSV, XML, etc.), processing information (full-width / half-width conversion, order change, etc.), image output format (JPEG / TIFF), image resolution, and image clipping position are defined.

  The timing mark indicates the reading start position of the form 21. The timing mark is set as a base point (coordinates) of the shaded image read from the form 21. Other than the timing marks, for example, any one of the squares of the form 21 may be used as a base point.

  The dictionary data storage unit 36 stores dictionary data that is referred to when performing character recognition processing. The dictionary data is data in which character codes are associated with character images (character patterns).

  In the recognition result storage unit 37, a character image cut out from the second halftone dot removal image and a character code obtained by character recognition of the character image are stored in association with each other.

  The first image processing unit 12 has almost the same diameter as a standard halftone dot from the image data of a form 21 including a character in which a halftone area is formed by regularly dotted halftone dots, that is, a halftone image. The center position of the connected component of the black pixel is detected, the connected component is removed from the shaded image to generate a first halftone dot removed image as the first processed image, and the detected center position of the connected component of the black pixel is detected. The position information and the first halftone dot removal image are stored in the storage device 3. The shaded area formed in the shaded image may be part or all.

  The regularly scattered halftone dots are, for example, halftone dots that are dotted at the intersections (vertices) when a large number of squares or rhombuses are formed by drawing line segments in a grid pattern.

Halftone determining section 13, the position of any connected components of two adjacent black pixels of the first halftone removed image stored in the storage device 3, position information of the connected component of black pixels that have been removed reads information, from a preset calculation rule, to calculate the position of the connected components other black pixels adjacent to the connected components of the two black pixels, the pixel is white first halftone removed image in that position Determine whether the pixel is a black pixel.

  The calculation rule is, for example, a calculation formula for obtaining a vertex from one side of a right-angled isosceles triangle. In addition, a halftone dot arrangement pattern or the like may be stored in the storage device 3 in advance, and the position of another halftone dot may be obtained from the arrangement pattern.

  The circumference calculation unit 14 calculates the circumference coordinates in a range of a predetermined diameter that a halftone dot (a connected component of black pixels) can take, with the position determined as a black pixel by the halftone determination unit 13 as the center.

  The range of the predetermined diameter that can be taken by the halftone dots (connected components of black pixels) is the value when the average value W of the halftone dots is about 10 pixels, for example, and the initial value is about 5 pixels of the lower limit value min. The circumference is calculated by incrementing increment Δ (for example, 1 pixel) from the maximum value, for example, up to about 15 pixels of the maximum value max (min <W <max).

  That is, the range of the predetermined diameter that can be taken by the halftone dots (connected components of black pixels) refers to a range from the lower limit value to the upper limit value of the halftone dot diameter. As the lower limit value and upper limit value of the diameter of the halftone dots, statistics may be obtained by measuring from the actual form 21, or values set in the standard form of the form 21 may be used. Further, a margin may be taken for the lower limit value and the upper limit value.

  The dividing point determination unit 15 detects the colors of all the pixels of the first halftone dot removal image on the circumferential coordinate calculated by the circumferential calculation unit 14, and the detected pixels on the circumferential coordinate are continuous. It is determined whether or not the white pixel and the continuous black pixel are divided into two.

  The second image processing unit 16 divides the first halftone dot-removed image into two regions by connecting the pixel division points determined to have been divided into two by the division point determination unit 15, and A black pixel region on the white pixel side of the white dot is removed to generate a second halftone dot removal image, which is stored in the second halftone dot removal image storage unit 34.

  The character recognizing unit 17 reads the second halftone dot removed image from which the halftone dot (black pixel connected component) overlapping the character image is removed from the second halftone dot removed image storage unit 34, and reads the read second halftone dot. Character recognition processing is performed on the removed image using dictionary data in the storage device 3.

  Hereinafter, a first operation example of the character recognition apparatus will be described with reference to FIGS. In this example, an example in which an image read from the form 21 is a monochrome image will be described, but a color image may be used.

  In the case of this character recognition apparatus, when the form 21 is installed in the scanner 22 and a reading start button operation is performed, the surface of the form 21 is read by the scanner 22 and the image data is processed through the input I / F 11 of the PC 1. It is stored in the target image storage unit 31.

  Then, the first image processing unit 12 reads out the shaded image (see FIG. 3) that is the image data of the form 21 stored in the processing target image storage unit 31, and detects the black pixel having almost the same diameter as the halftone dot. The center position of the connected component is detected, and the connected component of the black pixel at that position is removed from the shaded image to generate a first halftone-removed image (see FIG. 4). The position information of the position (hereinafter referred to as “center coordinates”) is stored in the halftone dot position information storage unit 32, and the first halftone dot removal image is stored in the first halftone dot removal image storage unit 33.

  That is, the first image processing unit 12 removes the halftone dots in the character peripheral portion, and stores the first halftone dot removed image from which the halftone dots in the character peripheral portion and the center coordinates of the removed halftone dots are stored in the storage device 3. Store (step S101 in FIG. 2).

  As shown in FIG. 4, in the first halftone dot removal image, halftone dots that overlap (contact with) the character are determined to be connected components of the character and are not removed.

Next, the halftone dot determination unit 13 selects the center coordinates (x of two halftone dots (black pixel connected components) adjacent to each other above and below the central coordinates of the removed halftone dots (black pixel connected components). (Coordinate, y coordinate) is read from the halftone dot position information storage unit 32, a flag indicating that the two halftone dots are processed is attached, and the y coordinate between the two halftone dots P as shown in FIG. And the coordinates of vertices (halftone dots) A and B of a right-angled isosceles triangle where the two corners of the side are 45 ° are calculated (S102). When two halftone dots (connected components of black pixels) are adjacent to the left and right, the side between the x coordinates is defined as one side.

  Then, the halftone dot determination unit 13 reads out the first halftone dot removal image from the processing target image storage unit 31, and the pixels on the first halftone dot removal image located at the respective vertices (halftone dots) A and B are detected. It is determined whether the pixel is white or black (S103).

  In this example, the position of the vertex (halftone dot) A is a white pixel since it is the position of the halftone dot already deleted in the first halftone dot removal image, and the position of the vertex (halftone dot) B is the position of the first halftone dot. On the point-removed image, it is a black pixel because it covers the character.

  As shown in FIG. 6, the circumference calculation unit 14 sets an initial value of the diameter of a calculation formula for obtaining a circle around the position of the vertex (halftone dot) B determined as a black pixel by the halftone dot determination unit 13. As a value, 5 pixels of a preset lower limit value min are substituted (S104), and circumferential coordinates are calculated (S105). The initial value is the lower limit value min in this example, but other values may be changed.

  When the circumference calculation unit 14 calculates the circumference coordinates, the segment point determination unit 15 detects the colors of all the pixels of the first halftone dot removal image on the circumference coordinates, and detects the detected circumference coordinates. It is determined whether or not the upper pixel has been divided into two: a continuous white pixel and a continuous black pixel (S106).

  As shown in FIG. 6, in this example, since the circumference 61 is sufficiently smaller than the halftone dot region of the first halftone dot removal image in the initial value, all the pixels on the circumferential coordinate are black pixels.

  As described above, when the pixel on the circumferential coordinate is not divided into the white pixel and the black pixel (No in S106), the division point determination unit 15 instructs the circumference calculation unit 14 to recalculate the circumference. .

  In the recalculation of the circumference, the circumference calculation unit 14 adds an increment Δ (for example, 1 pixel) from the value of the diameter S immediately before (S107), and whether or not the diameter S exceeds the maximum value max of the halftone dot. Is determined (S108). The maximum value max and the value of the increment Δ are preset values and can be corrected.

  As a result of this determination, when the maximum value max is not exceeded (No in S108), the circumference calculation unit 14 returns to the process in S105 and calculates the circumference again. This calculation is repeated until the halftone dot maximum value max, for example, about 15 pixels, is not divided into two. That is, the process is repeated until the circumference becomes sufficiently larger than the diameter of the removed halftone dot.

  If the circle diameter S exceeds the halftone maximum value max while repeating the circle calculation (Yes in S108), the circle calculation unit 14 returns to the process in S102 and removes the flag The center coordinates of any two halftone dots (connected components of black pixels) not marked with are read and the processing so far is repeated.

  Then, when processing is performed for all the combinations of adjacent halftone dots that have been removed (Yes in S110), the processing ends.

  As a result of repeating the circumferential coordinate calculation, when the circumference 61 is divided into two continuous white pixels 61a and continuous black pixels 61b as shown in FIG. It is determined that the pixel is divided into two continuous black pixels (Yes in S106). In the drawing, the color of the circle indicated by the dotted line is reversed between the white pixel and the black pixel.

In this case, as shown in FIG. 8, the second image processing unit 16 divides the first halftone dot removed image 60 into two regions by connecting the dividing points 62 of the white pixels and the black pixels with a straight line 63, The black pixel region 64 on the white pixel side of the circumference is removed to generate a second halftone dot removal image, and the generated second halftone dot removal image is stored in the second halftone dot removal image storage unit 34.
In other words, the second image processing unit 16 replaces (fills) the area surrounded by the straight line 63 and the circumferential white pixel 61a with the white pixel. That is, the second image processing unit 16 erases the black pixel region outside the character divided by the straight line 63 connecting the segment points 62 (S109).

  As a result, as shown in FIG. 9, the halftone dots 66 protruding from the character image 65 in the first halftone dot removal image are erased, and a smooth character image 67 faithful to the character line is obtained.

  Thereafter, the character recognizing unit 17 reads out the second halftone dot removed image from which the halftone dots protruding from the outer edge of the character are removed from the second halftone dot removed image storage unit 34, and applies the second halftone dot removed image to the second halftone dot removed image. Character recognition processing is performed using the dictionary data in the dictionary data storage unit 36, and the result of the character recognition processing is stored in the storage device 3.

  Character recognition processing includes processing for cutting out a character image from the second processed image according to the form definition, matching the cut out character image with dictionary data in the dictionary data storage unit 36, selecting a character code having a high degree of matching, and selecting the character A process of storing the code and the character image in the recognition result storage unit 37 is included.

  Subsequently, another operation example of the character recognition apparatus will be described with reference to FIGS. 10 to 13. First, a second operation example will be described with reference to FIG. In the above operation example, shading in which halftone dots are diagonally arranged is removed. However, in the second operation example, in the processing of S101 shown in FIG. 2, the first image processing unit 12 has a predetermined diameter or less. Whether the halftone pattern is a square pattern or a rhombus pattern is identified from the distribution (arrangement or arrangement) of the center coordinates of the halftone dots when the halftone dots are removed.

A square pattern, the processing of the next S102, the halftone determining section 13, any two halftone (black pixels vertically adjacent to one of the center coordinates of the removed dot (connected component of black pixels) Read out the center coordinates (x coordinate, y coordinate) of the halftone dots) from the halftone dot position information storage unit 32, and attach a flag indicating that the two halftone dots (black pixel connected components) have been processed, As shown in FIG. 10, two squares having one side between the y-coordinates of two halftone dots P that are adjacent vertically are created on the left and right sides of the sides, and the respective vertices (halftone dots) A, B, C, D The coordinates of are calculated.

  The subsequent processing is the same as the processing in the flowchart of FIG. According to this second operation example, it is possible to deal with the case of halftone dots (square pattern) in which halftone dots are arranged vertically and horizontally, and halftone dots in contact with characters can be removed.

  A third operation example will be described with reference to FIG. In this example, in the processing of S101 in FIG. 2, the first image processing unit 12 determines the mutual coordinates from the center coordinates of any two adjacent halftone dots P in the removed halftone dots (black pixel connected components). Calculate the distance and direction between them.

  Subsequently, as shown in FIG. 11, the first image processing unit 12 extends the side connecting the halftone dots P according to the calculated distance and direction between the halftone dots P, and the positions of the halftone dots (center coordinates) on the entire surface of the image. ) Is calculated.

  The subsequent processing is the same as the processing in the flowchart of FIG. 2, and when the calculated halftone dot position (center coordinate) is a black pixel, the circumference is calculated around that position and overlapped with the character. Erase halftone dots. According to the third operation example, halftone dots that are apart from the erased halftone dots and are connected to the characters at the opposite edge of the character image can be erased.

  A fourth operation example will be described with reference to FIGS. When the resolution of the scanner 22 is low or the like, as shown in FIG. 12, an image of a minute connected component that is larger than the halftone dot and almost the same as the halftone dot, such as a period or a cloud point, is overlapped with the halftone dot. May be disturbed.

  In such a case, in the above-described embodiment, the period is cut at the position of the line segment X, so that the character shape is worse than the original character image (period) as shown in FIG.

  Therefore, in this fourth operation example, the connected component to be removed and the size of the halftone dot are compared to check each other's difference (area difference, etc.). If there is not much difference, the erasure process in S109 is skipped.

  As a result, for small connected components, it is often the case that image disturbance is less when the overlapping halftone dots are not deleted. Therefore, the difference between the size of the connected components and the halftone dots is less than a predetermined value. The character recognition accuracy can be improved by including a condition that “the deletion process is not performed in some cases”.

  Thus, according to the character recognition device of this embodiment, first, a halftone dot that is not related to the character itself is removed from the image to generate a first halftone dot removal image, and then the removed halftone dot is created. From this position, the center position of the halftone dot included in the character or contact with the character is specified, and the distribution of black pixels and white pixels on the circumference calculated from the center position is examined. Then, when the circumference is divided into two continuous black pixels and continuous white pixels, the first halftone dot removal image is divided by connecting the divided division points, and the white pixels of the circumference are divided. A black pixel located on the side is removed to generate a second halftone dot removed image, and character recognition is performed using the second halftone dot removed image. As a result, it is possible to remove the shaded portion from the shaded image of the form 21 at high speed and with high accuracy to improve the character recognition accuracy.

  That is, according to the character recognition device of this embodiment, it is possible to remove halftone dots that overlap the character image included in the shaded image without damaging the character shape as much as possible. That is, the halftone dots that overlap the character image can be removed as faithfully as possible to improve the character recognition accuracy.

  In addition, this invention is not limited only to the said embodiment, You may deform | transform a component in the range which does not deviate from the summary in an implementation stage. Moreover, various inventions can be configured by appropriately combining a plurality of components disclosed in the embodiment.

  For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

It is a figure which shows the structure of the character recognition apparatus of one Embodiment of this invention. It is a flowchart which shows operation | movement of a character recognition apparatus. It is a figure which shows an example of the shaded image in which a halftone dot is at the apex of a rhombus. It is a figure which shows the 1st halftone dot removal image produced | generated by removing a halftone dot from a halftone image. It is a figure for demonstrating the operation | movement which pinpoints the position of the halftone dot concerning the character of a 1st halftone dot removal image from the position of the arbitrary adjacent halftone dots removed. It is a figure for demonstrating the operation | movement which calculates a circumferential coordinate from the center of the halftone dot of the removal target concerning a character. It is a figure which shows the state by which the circumference was divided into 2 by the white pixel and the black pixel, when the diameter of a circle was gradually enlarged every increment. It is a figure which shows a mode that the area | region of the black pixel of the 1st halftone dot removal image in the side of the white pixel of the circumference was connected by connecting the division points in the position divided into two. It is a figure which compares and shows the character image of a 1st halftone dot removal image, and the character image of a 2nd halftone dot removal image. FIG. 10 is a diagram for explaining an operation for obtaining the center coordinates of a removed halftone dot when the halftone dot is at a square vertex in the character recognition device. It is a figure for demonstrating the operation | movement which calculates | requires the center position of a halftone dot in the position of the edge of the separated character image when a halftone dot exists in the vertex of a rhombus. It is a figure which shows the example in which the connection component (a part of character) of the magnitude | size substantially the same as a halftone dot exists. It is a figure which shows a mode when the connection component of FIG. 12 is removed.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Computer (PC), 3 ... Memory | storage device, 11 ... Input interface part (input I / F), 12 ... 1st image processing part, 13 ... Halftone dot determination part, 14 ... Circumference calculation part, 15 ... Division point Determining unit, 16 ... second image processing unit, 17 ... character recognition unit, 18 ... output interface unit (output I / F), 20 ... input unit, 21 ... form, 22 ... scanner, 31 ... processing target image storage unit, 32 ... halftone dot position information storage unit, 33 ... first halftone dot removal image storage unit, 34 ... second halftone dot removal image storage unit, 35 ... form definition storage unit, 36 ... dictionary data storage unit, 37 ... recognition result storage Department.

Claims (7)

In a character recognition device for recognizing characters from a shaded image shaded by regularly dotted dots,
A storage device;
From the shaded image, and detects the center position of the pixel connected components under size or less of the preset said dot, said hatching and positional information of the center position of the detected pixel connected components, the pre-Symbol pixel connected components First halftone dot removing means for storing the first processed image removed from the image in the storage device;
Other adjacent to any of the position information of the two adjacent pixels connected component is read out from the storage device, the two pixel connected components in accordance with the arrangement pattern or calculation rule preset dot of the position information Other point position calculating means for calculating the position of the pixel connected component of
Halftone dot determining means for determining whether the pixel of the first processed image at the position of the calculated other pixel connected component is a white pixel or a black pixel;
Centering on a position determined as a black pixel by the halftone dot judging means, a circumference calculating means for calculating a circumferential coordinate in a range of a predetermined diameter that the halftone dot can take,
Demarcation point determination means for determining whether or not the pixels of the first processed image on the circumferential coordinates calculated by the circumference calculation means are divided into a continuous white pixel and a continuous black pixel;
The first processed image is segmented by connecting segment points of pixels of two colors determined to be divided by the segment point determination means, and black pixels on the white pixel side of the circumference are removed. A second halftone dot removing unit for generating a second processed image;
A character recognition device comprising character recognition means for performing character recognition on the second processed image. The other point position calculating means includes:
Creating a right-angled isosceles triangle with one side between the x-coordinates or y-coordinates of two adjacent pixel connected components that have been deleted and an angle at both ends of 45 °, and calculating the vertex coordinates The character recognition device according to claim 1. The other point position calculating means includes:
2. The character recognition apparatus according to claim 1, wherein a square having one side between the x coordinates or the y coordinates of two adjacent pixel connected components in the erased area is created, and the coordinates of the vertex are calculated. The other point position calculating means includes:
Calculate the distance and direction between each other from the center coordinates of any two adjacent pixel connected components that have been removed, and extend the line segment connecting each other according to the calculated distance and direction between the pixel connected components The character recognition apparatus according to claim 1, wherein center coordinates of pixel connected components of the entire image are calculated. The second halftone dot removing means includes
2. The character recognition apparatus according to claim 1, wherein the pixel connection component to be removed is compared with the size of the halftone dot, and the pixel connection component whose difference is less than a predetermined value is not deleted. In a character recognition method in a character recognition device for recognizing characters from a shaded image shaded by regularly dotted halftone dots,
From the shaded image, and detects the center position of the pixel connected components under size or less of the preset said dot, said hatching and positional information of the center position of the detected pixel connected components, the pre-Symbol pixel connected components Storing the first processed image removed from the image in a storage device;
Other adjacent to any of the position information of the two adjacent pixels connected component is read out from the storage device, the two pixel connected components in accordance with the arrangement pattern or calculation rule preset dot of the position information Calculating the position of the pixel connected component of
Determining whether the pixel of the first processed image at the position of the calculated other pixel connected component is a white pixel or a black pixel;
Centering on the position determined to be the black pixel, calculating a circumferential coordinate in a range of a predetermined diameter that the halftone dot can take;
Determining whether the pixels of the first processed image on the calculated circumferential coordinate are bisected into a continuous white pixel and a continuous black pixel;
The first processing image is divided by connecting the dividing points of the two color pixels determined to be divided into two, and the black pixel on the white pixel side of the circumference is removed to perform the second processing. Generating an image; and
And a step of performing character recognition on the second processed image. In a program for operating a character recognition device for recognizing characters from a shaded image shaded by regularly dotted halftone dots,
The character recognition device;
A storage device;
From the shaded image, and detects the center position of the pixel connected components under size or less of the preset said dot, said hatching and positional information of the center position of the detected pixel connected components, the pre-Symbol pixel connected components First halftone dot removing means for storing the first processed image removed from the image in the storage device;
Other adjacent to any of the position information of the two adjacent pixels connected component is read out from the storage device, the two pixel connected components in accordance with the arrangement pattern or calculation rule preset dot of the position information Other point position calculating means for calculating the position of the pixel connected component of
Halftone dot determining means for determining whether the pixel of the first processed image at the position of the calculated other pixel connected component is a white pixel or a black pixel;
Centering on a position determined as a black pixel by the halftone dot judging means, a circumference calculating means for calculating a circumferential coordinate in a range of a predetermined diameter that the halftone dot can take,
Demarcation point determination means for determining whether or not the pixels of the first processed image on the circumferential coordinates calculated by the circumference calculation means are divided into a continuous white pixel and a continuous black pixel;
The first processed image is segmented by connecting segment points of pixels of two colors determined to be divided by the segment point determination means, and black pixels on the white pixel side of the circumference are removed. A second halftone dot removing unit for generating a second processed image;
A character recognition program that functions as character recognition means for performing character recognition on the second processed image.

Images