JP4830331B2 - Character image cutting device and program - Google Patents

Description

  The present invention relates to a character image clipping technique used for character recognition processing.

  By comparing the feature points extracted from the image data indicating characters with the feature points of characters registered in advance in the database, the characters indicated by the image data are recognized, and text data indicating the recognized characters is generated. There is technology.

In the character recognition process as described above, when image data indicates a plurality of characters, it is necessary to cut out image data indicating each character from the image data. That is, it is necessary to specify which of the continuous on-pixel groups included in the image data constitutes one character. For example, Patent Literature 1 discloses such a character image segmentation technique.
Japanese Patent Laid-Open No. 10-207985

  According to the technique disclosed in Patent Document 1, the result of character recognition processing assuming that each on-pixel group is a constituent element of one character and two or more adjacent on-pixel groups of one character By comparing the result of character recognition processing assuming that it is a component with a word registered in the word dictionary, it is specified which result is correct.

  In the case of the above prior art, since the character recognition process is performed a plurality of times and the process of matching the result of the character recognition process with a word is required, it takes time to cut out image data indicating one character. There is. In addition, it is necessary to prepare a word dictionary in advance and to secure a storage area for the word dictionary.

  In view of the circumstances described above, an object of the present invention is to provide means for cutting out each character portion from image data simply and at high speed.

In order to solve the above-described problem, the present invention provides an image data acquisition unit that acquires image data including a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane; Of the pixels indicated by the pixel data included in the image data acquired by the image data acquisition means, the pixel is represented by a predetermined grid on the plane from among the on pixels indicating attribute values exceeding a predetermined threshold value. a region in the image recognition means for recognizing a set of oN pixels as an area in the image being arranged in succession to one another in the region, before Symbol territory outside adjacent to the recognized area in the image by the area in the image recognition means a determination unit configured to determine whether there is a collection of on-pixels arranged mutually in succession, the on if a collection of pixels is present by the determining means There recognizes an image obtained by integrating a group of the ON pixels in the area in the image as an extended area in an image, a collection of on-pixels that are arranged consecutively to each other in front Symbol territory outside and adjacent to the region in an image If not, the extended area image recognition means for recognizing the in-area image as the extended area image, and the data indicating the extended area image recognized by the extended area image recognition means are image data representing one character. And a character image cutting device.

In addition, the present invention provides an image data acquisition unit that acquires image data including a collection of pixel data indicating attribute values of a plurality of pixels constituting an image arranged on a plane, and the image data acquisition unit acquires the image data. Among the pixels indicated by the pixel data included in the image data that has been processed, the pixels that are consecutive from each other in the area represented by the predetermined grid on the plane are selected from the on pixels that indicate attribute values that exceed a predetermined threshold. An intra-region image recognition unit that recognizes a collection of on-pixels arranged as an intra-region image, and performs an expansion process on the intra-region image recognized by the intra-region image recognition unit to generate an in-expansion region image an expansion image generation means, the dilated image generating means by adjacent the generated expanded area images on the pixels that are arranged consecutively to each other in front Symbol territory outside A determination unit configured to determine whether Mari is present, the determination unit by the ON pixels of the on-pixels gather the expansion region in the integrated image extended expansion area in an image in the image of when the collection is present recognized as recognized extend the expansion region in an image when the collection of oN pixels that are arranged consecutively to each other does not exist in the previous SL territory outside and adjacent to the expansion region in an image as an extended expansion zone in the image Characters comprising: in-expansion area image recognition means; and output means for outputting data indicating the in-expansion area image recognized by the in-expansion area image recognition means as image data representing one character An image cutting device is provided.

  In a preferred aspect, the character image cutout device further includes a contraction image generation unit that performs contraction processing on the image in the expansion expansion region recognized by the image recognition unit in expansion expansion region and generates an image in the expansion expansion region. The output means may be configured to output data indicating a contraction / expansion / expansion area image generated by the contraction image generation means, instead of data indicating the expansion / expansion area image.

In addition, the present invention provides an image data acquisition unit that acquires image data including a collection of pixel data indicating attribute values of a plurality of pixels constituting an image arranged on a plane, and the image data acquisition unit acquires the image data. Among the pixels indicated by the pixel data included in the image data that has been processed, the pixels that are consecutive from each other in the area represented by the predetermined grid on the plane are selected from the on pixels that indicate attribute values that exceed a predetermined threshold. An intra-region image recognition unit that recognizes a collection of on-pixels arranged as an intra-region image, and performs an expansion process on the intra-region image recognized by the intra-region image recognition unit to generate an in-expansion region image an expansion image generation means, the dilated image generating means by adjacent the generated expanded area images on the pixels that are arranged consecutively to each other in front Symbol territory outside A determination unit configured to determine whether Mari is present, the determination unit by the ON pixels of the on-pixels gather the expansion region in the integrated image extended expansion area in an image in the image of when the collection is present recognized as recognized extend the expansion region in an image when the collection of oN pixels that are arranged consecutively to each other does not exist in the previous SL territory outside and adjacent to the expansion region in an image as an extended expansion zone in the image Pixels that are ON pixels in both the in-expansion region image recognition unit, the image indicated by the image data acquired by the image data acquisition unit, and the expansion in-expansion region image recognition unit A duplicate image recognition means for recognizing a set of images as a duplicate image, and data representing the duplicate image recognized by the duplicate image recognition means as one character. Providing a character image extraction device characterized by an output means for outputting as image data.

In addition, the present invention provides an image data acquisition unit that acquires image data including a collection of pixel data indicating attribute values of a plurality of pixels constituting an image arranged on a plane, and the image data acquisition unit acquires the image data. Among the pixels indicated by the pixel data included in the image data that has been processed, the pixels that are consecutive from each other in the area represented by the predetermined grid on the plane are selected from the on pixels that indicate attribute values that exceed a predetermined threshold. a region in the image recognition means for recognizing a group of on-pixels arranged as an area in an image Te, arranged in series with each other in adjacent pre Symbol territory outside the recognition area in the image by the area in the image recognition means a determination unit configured to determine whether a collection of on-pixels are there, collecting of the oN pixels in the case where the collection of the oN pixels is present by the determining means Ri and recognizes images integrated in the area in the image as an extended area within the image, the in the case where collection of ON pixels that are arranged consecutively to each other in front Symbol territory outside and adjacent to the region in an image is not present An extension area image recognition means for recognizing an in-area image as an extension area image, and an area of a predetermined shape on the plane that circumscribes the extension area image recognized by the extension area image recognition means as an extension area An extended area recognizing means for recognizing; and an output means for outputting data indicating an image indicated by a collection of on-pixels included in the extended area recognized by the extended area recognizing means as image data representing one character. A character image cutting device is provided.

In a preferred embodiment, the area in the image recognition means of the character image extraction apparatus, the extension after the recognition of the extended region is performed by the region recognizing means, before Symbol territory recognized extended regions by the extended region recognizing means And re-execution of the recognition process of the in-region image as a region, and the extended region image recognition means performs the processing of the extension in-region image with respect to the in-region image recognized by re-execution of the recognition process by the in-region image recognition unit Re-execution of recognition processing, the extension region recognition means re-executes recognition processing of the extension region on the image in the extension region recognized by re-execution of recognition processing by the image recognition means in the extension region, and the output Means for representing data representing an image indicated by a collection of on-pixels included in an extended area recognized by re-execution of recognition processing by the extended area recognizing means; It may be configured to output as image data.

In addition, the present invention provides an image data acquisition unit that acquires image data including a collection of pixel data indicating attribute values of a plurality of pixels constituting an image arranged on a plane, and the image data acquisition unit acquires the image data. Among the pixels indicated by the pixel data included in the image data that has been processed, the pixels that are consecutive from each other in the area represented by the predetermined grid on the plane are selected from the on pixels that indicate attribute values that exceed a predetermined threshold. An intra-region image recognition means for recognizing a collection of on-pixels arranged as an intra-region image, and at least one intra-region image among at least one intra-region image recognized by the intra-region image recognition means, a determination unit configured to determine whether there is a collection of on-pixels that are arranged consecutively to each other in front Symbol territory outside adjacent to the region in an image, said determining means If there are more group of the on-pixel, and the area in the image excluding means excludes from collection areas within the image that has been recognized by the region within the image the area in the image recognition unit are recognized by the region image recognition means A character image cutout device comprising: output means for outputting a collection of in-area images that are not excluded by the in-area image excluding means among the in-area images, as image data representing one character. To do.

In a preferred aspect, the in-region image excluding unit of the character image cutting device includes a group of on pixels that are adjacent to the in-region image and are continuously arranged outside the region with respect to the one in-region image. If present, if the area occupied by the image in one region is smaller than the area occupied by the set of on-pixels, the image in the one region is extracted from the set of images in the region recognized by the image recognition means in the region. If the area occupied by the image in one region is larger than the area occupied by the collection of on-pixels, the image in the one area may not be excluded from the collection of images in the region .

Further, in another preferred embodiment, the character image extraction apparatus collection of ON pixels that are arranged consecutively to each other in front Symbol territory outside adjacent to that have not been region image excluding by the region image excluding means oN pixels in which the a group of on-pixels to recognize an image integrated in the area in the image as an extended area in the image, is arranged in series with each other before Symbol territory outside and adjacent to the region in the image, if present If there is no set of images, the image processing apparatus further includes an extended area image recognition means for recognizing the image in the area as an extended area image, and the output means recognizes the extended area image recognized by the extended area image recognition means. May be output as image data representing the one character.

In addition , the present invention provides a program that causes a computer to execute the same process as the process performed by any one of the character image cutting apparatuses described above.

  According to the present invention, since an image that correctly shows one character is cut out from an image that shows characters that are written out of the cell, highly accurate character recognition is possible.

[Embodiment]
Hereinafter, preferred embodiments of the present invention will be described. FIG. 1 is a block diagram showing a configuration of a character recognition system 1 according to an embodiment described below. The character recognition system 1 includes a character image cutting device 10 that cuts and transmits image data indicating each character from image data indicating a plurality of characters, and image data indicating a plurality of characters to the character image cutting device 10. A tablet PC (Personal Computer) 11 for transmission and a character recognition device 12 for generating text data indicating a recognized character by performing character recognition processing on image data indicating each character transmitted from the character image cutting device 10 I have.

  The tablet PC 11 includes a pen tablet-type input device stacked on a liquid crystal display. When a user performs a writing operation using a pen-type stylus on a grid displayed on the liquid crystal display, the pen tablet-type input device performs the writing operation. The position where the pen pressure is applied in the operation and the magnitude of the pen pressure are measured, and image data indicating characters written by the user is generated based on the measured values.

[1. First Embodiment]
FIG. 2 shows image data cut out from the second square by the conventional technique when the user writes a character string “Masayoshi Sugawara” for the four-character square displayed on the tablet PC 11 (FIG. 2A). ) And image data cut out from the second grid by the character image cutout device 10-1 according to the first embodiment of the present invention (FIG. 2B), the character recognition device 12 performs character recognition processing. It is the figure which compared and showed the text data obtained when performing.

  As shown in FIG. 2, when a part of the “original” character is written out from the cell, according to the prior art, image data that does not include the image portion that protrudes from the cell is transmitted to the character recognition device 12. Therefore, character recognition using the image data is often not performed correctly. On the other hand, in the case of the character image cutting device 10-1, even if the image is arranged outside the cell, image data including a portion continuous with the image arranged in the cell is a character. Since it is output to the recognition device 12, the possibility that character recognition using the image data is correctly performed is higher than in the case of the prior art.

  FIG. 3 is a block diagram showing the configuration of the character image cutting device 10-1. The character image cutout device 10-1 stores a control unit 101 that controls the components of the character image cutout device 10-1, a program that instructs various processes by the control unit 101, and various types of data. A storage unit 102 used as a work area is provided.

  The control unit 101 includes an image data input unit 1011 that receives image data indicating a character string from the tablet PC 11 and region data indicating each region of the grid displayed on the tablet PC 11. The image data input unit 1011 stores the image data and area data received from the tablet PC 11 in the storage unit 102 as image data 1021 and area data 1022, respectively. The image data 1021 is a set of pixel data corresponding to pixels arranged at equal intervals on a plane, and each pixel data includes, for example, saturation data, lightness data, and hue data.

  The control unit 101 includes a binarization unit 1012 that converts image data indicated by the image data 1021 into monochrome binary image data. The binarization unit 1012 converts the pixel data included in the image data 1021 into light-on pixel data having a value “1” when the lightness data exceeds a predetermined threshold value, and off-pixel data having a value “0” for the other pixel data. By converting, binary image data is generated. The binarization unit 1012 stores the generated binary image data in the storage unit 102 as binary image data 1023.

  The control unit 101 extracts the portion of the image data included in each square area indicated by the area data 1022 from the pixel image data indicated by the binary image data 1023, and each successive image portion included in each of the image data is indicated. An in-region image recognition unit 1013 that recognizes separately is provided. Hereinafter, the process performed in the character image cutting device 10-1 regarding the image data contained in the area of the second cell will be described. The in-region image recognition unit 1013 groups on-pixel data included in a second mesh region (hereinafter referred to as “predetermined region”) into a group of on-pixel data that are continuous with each other. Hereinafter, each of the images indicated by the on-pixel data group recognized as one group by the in-region image recognition unit 1013 is referred to as an “in-region image”. The in-region image recognition unit 1013 adds an identification label to each recognized in-region image. L1 to L4 shown in FIG. 2 indicate identification labels added by the in-region image recognition unit 1013.

  The control unit 101 includes an in-region image recognition unit 1014 that extends the in-region image to a continuous part outside the predetermined region. The in-region image recognition unit 1013 delivers data indicating the in-region image to which the identification label is added (hereinafter referred to as “in-region image data”) to the extended in-region image recognition unit 1014. The extended in-region image recognition unit 1014 uses the in-region image data and the binary image data 1023 received from the in-region image recognition unit 1013, and for each of the in-region images, there is a continuous on pixel group outside the predetermined region. In the case where there are continuous on-pixel groups, the in-region image is expanded by integrating the on-pixel group into the in-region image. Specifically, as shown in FIG. 2, the in-region image of the identification label L <b> 1 is expanded by the extended region image recognition unit 1014. Hereinafter, the in-region image that has been subjected to the extension process by the in-extension region image recognition unit 1014 is referred to as an “extended region image”.

  The control unit 101 includes an image data output unit 1015 that transmits image data representing one character to the character recognition device 12. The extended area image recognition unit 1014 delivers the extended area image data indicating the extended area image to the image data output unit 1015. The image data output unit 1015 transmits the received image data in the extended area to the character recognition device 12 as image data indicating one character.

  As described above, the image data cut out by the character image cutout device 10-1 and transmitted to the character recognition device 12 shows an image including a character portion that is written out of the cell, so that the character recognition process has high accuracy. Result.

[2. Second Embodiment]
FIG. 4 shows image data cut out from the first square by the conventional technique when the user writes a character string “Masayoshi Sugawara” for the four-character square displayed on the tablet PC 11 (FIG. 4A). ) And image data cut out from the first cell by the character image cutout device 10-2 according to the second embodiment of the present invention (FIG. 4B), the character recognition device 12 performs character recognition processing. It is the figure which compared and showed the text data obtained when performing.

  As shown in FIG. 4, when a part of the character “榊” is completely out of the cell, according to the prior art, image data that does not include an image part that is out of the cell is transferred to the character recognition device 12. Therefore, character recognition using the image data is not performed correctly. On the other hand, in the case of the character image cutting device 10-2, even if the image is arranged outside the cell, image data including a portion close to the image arranged in the cell is character. Since it is output to the recognition device 12, there is a high possibility that character recognition using the image data will be performed correctly.

  FIG. 5 is a block diagram showing the configuration of the character image cutting device 10-2. Since the configuration and operation of the character image cutting device 10-2 are the same as those of the character image cutting device 10-1 in many respects, the character image cutting device 10-2 is hereinafter referred to as the character image cutting device. Only differences from 10-1 will be described. Moreover, in the figure referred in the following description, the same code | symbol as what was used in the character image cutting device 10-1 is attached | subjected to the component corresponding to the component of the character image cutting device 10-1. .

  The control unit 101 of the character image cutting device 10-2 includes an expanded image generation unit 2011 that performs an expansion process on the in-region image. The dilated image generation unit 2011 receives the intra-region image data from the intra-region image recognition unit 1013 and turns on the off-pixel data adjacent to the on-pixel data indicating each outer ring of the intra-region image indicated by the received intra-region image data. By repeating the process of rewriting the pixel data a predetermined number of times, the in-region image is expanded. Hereinafter, the in-region image subjected to the expansion process by the expanded image generation unit 2011 is referred to as an “inflated region image”.

  The control unit 101 of the character image cutting device 10-2 includes an expanded in-region image recognition unit 2012 instead of the in-expansion region image recognition unit 1014 provided in the control unit 101 of the character image cutting device 10-1. . The expansion image generation unit 2011 delivers the expansion area image data indicating the expansion area image to the expansion expansion area image recognition unit 2012.

  The expanded in-region image recognizing unit 2012 uses the in-expanded region image data and the binary image data 1023 received from the expanded image generating unit 2011, and for each of the in-expanded region images, an on-pixel group continuous outside the predetermined region is obtained. It is determined whether or not it exists, and for those having a continuous on-pixel group, the on-pixel group is expanded by integrating the on-pixel group into the in-expansion area image. Specifically, as shown in FIG. 4, the in-expanded region image recognition unit 2012 expands the in-expanded region image of the identification label L2. Hereinafter, the in-region image subjected to the expansion processing by the inflated region in-region image recognition unit 2012 is referred to as “an inflated region image”.

  The inflated region image recognition unit 2012 passes the inflated region image data indicating the inflated region image to the image data output unit 1015. The image data output unit 1015 transmits the received image data in the expanded expansion area to the character recognition device 12 as image data indicating one character.

  As described above, the image data cut out by the character image cutout device 10-2 and transmitted to the character recognition device 12 shows an image including a character portion that is written out of the cell, so that the character recognition process has high accuracy. Result.

  By the way, the expansion in-expansion area image generated by the expansion in-expansion area image recognition unit 2012 is subjected to the expansion process for the image in the grid and is not subjected to the expansion process for the image outside the grid. In general, a result with higher accuracy is obtained when character recognition is performed using an original image than when character recognition is performed using an image subjected to dilation processing. Therefore, the following modifications may be added to the character image cutting device 10-2.

  FIG. 6 is a block diagram illustrating a configuration when the control unit 101 of the character image cutting device 10-2 includes a contracted image generation unit 2013 that performs contraction processing on an image in the expanded expansion region. The contracted image generation unit 2013 is interposed between the expansion expansion region internal image recognition unit 2012 and the image data output unit 1015, receives the expansion expansion region internal image data from the expansion expansion region internal image recognition unit 2012, and receives the received expansion expansion region The process of rewriting the on-pixel data indicating the outer ring of each of the images in the expansion / expansion area indicated by the internal image data with the off-pixel data is repeated a predetermined number of times to contract the image in the expansion / expansion area. Hereinafter, the image in the expansion / expansion area subjected to the contraction process by the contraction image generation unit 2013 is referred to as “image in the contraction / expansion / expansion area”.

  The image within the contraction / expansion / expansion region includes an image having the same thickness as the original image with respect to the image within the mesh, and a thin image with respect to the image outside the mesh compared with the original image. In general, in character recognition processing, in order to extract feature points of an image, processing for thinning an image having a thickness is performed. Compared with, the thin point does not have a bad influence on the accuracy of character recognition. On the other hand, the image in the grid included in the image in the contraction / expansion / expansion area has a thickness that approximates the original as compared with the image in the expansion / expansion area, resulting in an image that provides higher-accuracy character recognition.

  The contracted image generation unit 2013 delivers the image data in the contraction / expansion / expansion area indicating the image in the contraction / expansion / expansion area to the image data output unit 1015. The image data output unit 1015 transmits the received image data in the contraction / expansion region to the character recognition device 12 as image data indicating one character.

  Further, in FIG. 7, the control unit 101 of the character image cutting device 10-2 includes an overlapping image recognition unit 2014 that extracts an image of an overlapping portion between the image in the expanded expansion area and the image indicated by the binary image data 1023. It is a block diagram which shows the structure at the time of making it. The overlapping image recognition unit 2014 is interposed between the expansion expansion region image recognition unit 2012 and the image data output unit 1015, receives the expansion expansion region image data from the expansion expansion region image recognition unit 2012, and receives the received expansion expansion region For each piece of pixel data included in the internal image data, a logical product between the pixel data included in the binary image data 1023 corresponding to the pixel data is calculated. A collection of pixel data whose value is the result of the logical product calculated by the overlapping image recognition unit 2014 is data indicating an overlapping image of the image in the expanded expansion area and the image indicated by the binary image data 1023.

  As described above, the overlapping image recognized by the overlapping image recognition unit 2014 is obtained by extracting a portion included in the range occupied by the image in the expanded expansion area from the original image. Therefore, compared with the image in the expanded expansion area that has been subjected to the expansion process, the image can provide character recognition with higher accuracy.

  The duplicate image recognition unit 2014 passes data indicating the duplicate image to the image data output unit 1015. The image data output unit 1015 transmits the received data indicating the duplicate image to the character recognition device 12 as image data indicating one character.

[3. Third Embodiment]
FIG. 8 shows the third embodiment of the present invention when the user writes a character string “Masayoshi Sugawara” for the four-character cell displayed on the tablet PC 11 as in the case of the second embodiment. It is the figure which showed a mode that image data was cut out from the 1st cell by this character image cutting device 10-3.

  FIG. 9 is a block diagram showing the configuration of the character image cutting device 10-3. Since the configuration and operation of the character image cutting device 10-3 are similar to those of the character image cutting device 10-2 in many respects, the character image cutting device 10-3 is hereinafter referred to as the character image cutting device. Only differences from 10-2 will be described. In the drawings referred to in the following description, the component corresponding to the component of the character image cutting device 10-1 or the character image cutting device 10-2 includes the character image cutting device 10-1 or the character image cutting device. The same code | symbol as what was used in the taking-out apparatus 10-2 is attached | subjected.

  The control unit 101 of the character image cutout device 10-3 includes an image recognition unit 3011 that separately recognizes successive image portions included in the binary image data 1023, instead of the in-region image recognition unit 1013. . The image recognition unit 3011 groups on-pixel data included in the binary image data 1023 into groups of on-pixel data that are continuous with each other, and recognizes each on-pixel data as a partial image. The image recognition unit 3011 adds an identification label to each of the partial images recognized as such. The image recognition unit 3011 delivers data indicating the partial image to which the identification label is added (hereinafter referred to as “partial image data”) to the dilated image generation unit 2011.

  The expanded image generation unit 2011 performs expansion processing on each of the partial images indicated by the received partial image data, and generates an expanded image. The control unit 101 of the character image cutout device 10-3 includes a connected expanded image recognition unit 3012 that uses partial images that are continuous with each other by expansion processing as one partial image. The dilated image generation unit 2011 delivers dilated image data indicating the dilated image to the connected dilated image recognition unit 3012. The connected expanded image recognition unit 3012 groups the expanded images indicated by the expanded image data into continuous expanded image groups, connects the grouped expanded image groups, and re-recognizes them as one image. Hereinafter, the group of expanded images connected in this manner is referred to as a “connected expanded image”.

  The control unit 101 of the character image cutout device 10-3 includes an intra-region connected dilated image recognition unit 3013 that recognizes a connected dilated image including a part or all of a predetermined region. The intra-region connected expanded image recognition unit 3013 receives the connected expanded image data indicating the connected expanded image from the connected expanded image recognition unit 3012, and among the connected expanded images indicated by the connected expanded image data, the first region indicated by the region data 1022. Identify what occupies at least part of the cell. Hereinafter, the connected expanded image identified by the intra-region connected expanded image recognition unit 3013 is referred to as an “in-region connected expanded image”.

  The intra-region linked dilated image recognition unit 3013 delivers intra-region linked dilated image data indicating the intra-region linked dilated image to the image data output unit 1015. The image data output unit 1015 transmits the received intra-region expanded image data to the character recognition device 12 as image data indicating one character.

  As described above, the image data cut out by the character image cutout device 10-3 and transmitted to the character recognition device 12 is written out of the grid as the image data cut out by the character image cutout device 10-2. Since an image including a character portion is shown, a highly accurate result is obtained in the character recognition process.

  By the way, the intra-region linked dilated image recognized by the intra-region linked dilated image recognition unit 3013 is an image obtained by performing dilation processing on the original image in both the inside and outside of the mesh. Therefore, in order to generate an image that provides higher character recognition accuracy, the following modification may be added to the character image cutting device 10-3.

  FIG. 10 is a block diagram illustrating a configuration in the case where the control unit 101 of the character image cutting device 10-3 includes a contracted image generation unit 2013. The contracted image generation unit 2013 is inserted between the intra-region connected expanded image recognition unit 3013 and the image data output unit 1015, and is connected within the region indicated by the intra-region connected expanded image recognition unit 3013. A contraction process is performed on the expanded image to generate a connected expanded image within the contracted region. The contracted intra-region connected dilated image has a thickness that approximates that of the original image compared to the intra-region concatenated dilated image, resulting in higher-accuracy character recognition.

  The contracted image generation unit 2013 delivers the in-contracted region connected expanded image data indicating the in-contracted region connected expanded image to the image data output unit 1015. The image data output unit 1015 transmits the received in-shrinkage region connected expansion image data to the character recognition device 12 as image data indicating one character.

  FIG. 11 is a block diagram illustrating a configuration when the control unit 101 of the character image cutting device 10-3 includes the duplicate image recognition unit 2014. The overlapping image recognition unit 2014 is interposed between the intra-region connected expanded image recognition unit 3013 and the image data output unit 1015, and is connected within the region indicated by the intra-region connected expanded image recognition unit 3013. An image of an overlapping portion between the dilated image and the image indicated by the binary image data 1023 is extracted.

  As described above, the overlapping image recognized by the overlapping image recognition unit 2014 is obtained by extracting a portion included in the range occupied by the intra-region connected expansion image from the original image. Therefore, it is an image that provides character recognition with higher accuracy compared to the intra-region connected expanded image subjected to the expansion processing.

  The duplicate image recognition unit 2014 passes data indicating the duplicate image to the image data output unit 1015. The image data output unit 1015 transmits the received data indicating the duplicate image to the character recognition device 12 as image data indicating one character.

[4. Fourth Embodiment]
FIG. 12 shows the case where the user writes the character string “Masayoshi Sugawara” for the four-character cell displayed on the tablet PC 11 as in the case of the second and third embodiments. It is the figure which showed a mode that image data was cut out from the 1st cell by the character image cutting device 10-4 concerning 4th Embodiment.

  FIG. 13 is a block diagram showing the configuration of the character image cutting device 10-4. Since the configuration and operation of the character image cutting device 10-4 are similar to those of the character image cutting device 10-3 in many respects, the character image cutting device 10-4 is hereinafter referred to as the character image cutting device. Only differences from 10-3 will be described. Further, in the drawings to be referred to in the following description, the component corresponding to the component of the character image cutting device 10-1 to the character image cutting device 10-3 has a character image cutting device 10-1 to a character image cutting device. The same code | symbol as what was used in the taking-out apparatus 10-3 is attached | subjected.

  The control unit 101 of the character image cutting device 10-4 is indicated by the area data 1022 among the image data indicated by the binary image data 1023 in addition to all the components included in the character image cutting device 10-3. An in-region image recognition unit 1013 is provided which takes out image data portions included in the cells and separates the consecutive image portions included therein and recognizes them as in-region images. The intra-region image recognition unit 1013 delivers the intra-region image data indicating the recognized intra-region images and the identification labels attached to them to the intra-region connected expanded image recognition unit 3013.

  The intra-region connected expanded image recognition unit 3013 of the character image cutting device 10-4 receives the connected expanded image data indicating the connected expanded image from the connected expanded image recognition unit 3012, and the intra-region image data from the intra-region image recognition unit 1013. Receive. For each of the in-region images indicated by the in-region image data, the in-region connected expanded image recognition unit 3013 selects any on-pixel included in the in-region image as a starting point. The intra-region connected dilated image recognition unit 3013 identifies an on-pixel at a position corresponding to the previously selected on-pixel included in the linked dilated image indicated by the connected dilated image data, and continues to the specified on-pixel. The on-pixel group is recognized as an intra-region connected expanded image.

  For example, according to the example shown in FIG. 12, the image of “榊” shown by the connected expanded image is an image in the region where the left part (L11) constituting the character is attached with the identification label L1 or L2. It is recognized as an intra-region connected expanded image by the extraction process starting from the included on-pixel. In addition, the right part (L12) of the “榊” image indicated by the connected dilated image is recognized as an intra-region dilated image by the extraction process starting from the on-pixel included in the intra-region image with the identification label L3. Is done.

  When the intra-region linked dilated image recognition unit 3013 recognizes the intra-region linked dilated image as described above, the intra-region linked dilated image data is transferred to the image data output unit 1015 indicating the recognized intra-region linked dilated image. The image data output unit 1015 transmits the received intra-region expanded image data to the character recognition device 12 as image data indicating one character.

  As described above, the image data cut out by the character image cutout device 10-4 and transmitted to the character recognition device 12 becomes the same image data as the image data cut out by the character image cutout device 10-3 and deviates from the grid. Since an image including a written character portion is shown, a highly accurate result is obtained in the character recognition process.

  By the way, the intra-region connected expanded image recognized by the intra-region connected expanded image recognition unit 3013 of the character image cutting device 10-4 is recognized by the character image cutting device 10-4 of the character image cutting device 10-3. Therefore, the character image cutting device 10-4 may be modified in the same manner as described above with respect to the character image cutting device 10-3.

  That is, a contracted area obtained by interpolating the contracted image generating unit 2013 between the intra-area connected expanded image recognition unit 3013 and the image data output unit 1015 of the character image cutting device 10-4 and performing contraction processing on the intra-area connected expanded image. It is also possible to generate an inner connected expanded image and output from the image data output unit 1015 the intra-contracted region expanded image data indicating the contracted region connected expanded image. In addition, an overlapping image recognition unit 2014 is inserted between the intra-region connected expanded image recognition unit 3013 and the image data output unit 1015 of the character image cutting device 10-4, and is indicated by the intra-region connected expanded image and the binary image data 1023. It is also possible to recognize an overlapping image with the image to be output and output data indicating the overlapping image from the image data output unit 1015.

[5. Fifth Embodiment]
FIG. 14 shows image data cut out from the fourth square by the conventional technique when the user writes a character string “Masayoshi Sugawara” for the four squares displayed on the tablet PC 11 (FIG. 14A). ) And image data cut out from the fourth cell by the character image cutout device 10-5 according to the fifth embodiment of the present invention (FIG. 14B), the character recognition device 12 performs character recognition processing. It is the figure which compared and showed the text data obtained when performing.

  FIG. 15 is a block diagram showing the configuration of the character image cutting device 10-5. Since the configuration and operation of the character image cutting device 10-5 are similar to those of the character image cutting device 10-1 in many respects, the character image cutting device 10-5 is hereinafter referred to as the character image cutting device. Only differences from 10-1 will be described. Further, in the drawings to be referred to in the following description, the component corresponding to the component of the character image cutting device 10-1 to the character image cutting device 10-4 has a character image cutting device 10-1 to a character image cutting device. The same reference numerals as those used in the dispensing device 10-4 are attached.

  The control unit 101 of the character image cutout device 10-5 includes an extended area recognition unit 5011 that extends a processing target area from a grid to a rectangular area that circumscribes the extended area image. The extended region image recognition unit 1014 included in the control unit 101 of the character image cutting device 10-5 receives and receives the region image data indicating the region image arranged in the cell from the region image recognition unit 1013. When the extension region image data is generated using the region image data and the binary image data 1023, the generated extension region image data is delivered to the extension region recognition unit 5011.

  The extension area recognition unit 5011 identifies a rectangular area that circumscribes the collection of images in the extension area indicated by the image data in the extension area received from the image recognition section 1014 in the extension area. Hereinafter, the area thus identified is referred to as an “extended area”. In FIG. 14, a region A0 indicates a fourth cell region indicated by the region data 1022, and a region A1 indicates an extended region. The extended area recognition unit 5011 delivers the extended area data indicating the extended area to the in-area image recognition unit 1013.

  When the in-region image recognition unit 1013 receives the extension region data from the extension region recognition unit 5011, the in-region image recognition unit 1013 replaces the region of the mesh indicated by the region data 1022 among the on-pixel data included in the binary image data 1023. The image recognition process is executed again. As a result, in FIG. 14, the in-region image with the new identification label L4 is recognized by the in-region image recognition unit 1013 in addition to the already recognized in-region images with the identification labels L1 to L3. . The intra-region image recognition unit 1013 delivers the intra-region image data indicating the recognized intra-region image to the extended region image recognition unit 1014.

  When the intra-region image recognition unit 1014 receives the intra-region image data generated for the extension region from the intra-region image recognition unit 1013, the intra-region image recognition unit 1014 re-executes the expansion processing of the intra-region image indicated by the received intra-region image data. As a result, in FIG. 14, the in-region image with the identification label L4 is expanded. The extension area image recognition unit 1014 delivers the extension area image data indicating the recognized extension area image to the image data output unit 1015. The image data output unit 1015 transmits the received image data in the extended area to the character recognition device 12 as image data indicating one character.

  In the above description, the extension region recognition unit 5011 performs the extension region recognition processing, the region image recognition unit 1013 recognizes the region image, and the extension region image recognition unit 1014 recognizes the extension region image once. However, the number of repetitions may be a predetermined number of times. Further, these processes may be repeated until the extension of the in-region image by the extension-in-region image recognition unit 1014 or the extension of the region by the extension region recognition unit 5011 is not performed.

  As described above, the image data cut out by the character image cutting device 10-5 and transmitted to the character recognition device 12 is out of the grid, for example, not cut out by the expansion processing in the character image cutting device 10-1. Since an image including a character portion that is separately written is shown, a highly accurate result is obtained in the character recognition processing.

  By the way, the repeated execution of recognition processing by area expansion in the character image cutting device 10-5 may be combined with the expansion processing of the character image cutting device 10-2 to the character image cutting device 10-4.

[6. Sixth Embodiment]
FIG. 16 shows a case where the character image cutting device 10-6 according to the sixth embodiment of the present invention is used when the user writes a character string “Masayoshi Sugawara” with respect to the four-character grid displayed on the tablet PC 11. It is the figure which showed a mode that image data was cut out from the 1st cell. In this example, a part of the characters to be written in the second cell is written in the first cell, so that the character recognition process is performed using the image included in the first cell. In this case, high accuracy of character recognition cannot be expected. On the other hand, the character image cutout device 10-6 cuts out an image that brings about highly accurate character recognition by removing an image portion that protrudes from a neighboring cell.

  FIG. 17 is a block diagram showing the configuration of the character image cutting device 10-6. Since the configuration and operation of the character image cutting device 10-6 are similar to those of the character image cutting device 10-1 in many respects, the character image cutting device 10-6 is hereinafter referred to as the character image cutting device. Only differences from 10-1 will be described. Further, in the drawings to be referred to in the following description, the component corresponding to the component of the character image cutting device 10-1 to the character image cutting device 10-5 has the character image cutting device 10-1 to the character image cutting device. The same code | symbol as what was used in the taking-out apparatus 10-5 is attached | subjected.

  The control unit 101 of the character image cutout device 10-6 replaces the extended region image recognition unit 1014 included in the control unit 101 of the character image cutout device 10-1, and turns on the continuous images outside the grid in the region image. An intra-region image excluding unit 6011 for excluding from the group of intra-region images that are handed over to the image data output unit 1015 an intra-region image that is supposed to indicate a portion protruding from a neighboring cell from among those having a pixel group. Yes.

  The intra-region image recognition unit 1013 delivers the intra-region image data to the intra-region image exclusion unit 6011. The in-region image exclusion unit 6011 uses the received in-region image data and binary image data 1023 to determine whether or not there is a continuous on-pixel group for each of the in-region images. For an on-pixel group (in-region image with identification labels L1 and L5 in FIG. 16), the number of on-pixels (or its area) included in the in-region image within the cell and the in-region image Are compared with the number (or area) of on-pixels included in a group of on-pixels outside the grid. When the number of on-pixels in the cell is smaller than the number of on-pixels outside the cell (intra-region image with the identification label L5 in FIG. 16), the in-region image excluding unit 6011 displays the in-region image as an image. Excluded from the collection of in-region images to be output to the data output unit 1015.

  That is, the in-region image excluding unit 6011 has a larger area than an in-region image that does not have an on-pixel group continuous outside the grid and an on-pixel group that has an on-pixel group continuous outside the grid but outside the grid. The in-region image data indicating the in-region image occupying a large area is delivered to the image data output unit 1015. The image data output unit 1015 transmits the received in-region image data to the character recognition device 12 as image data indicating one character.

  As described above, the image data cut out by the character image cutout device 10-6 and transmitted to the character recognition device 12 shows an image in which the character portion protruding from the neighboring cell is excluded, so that it is highly accurate in character recognition processing. Will bring about the result.

  By the way, the in-region image exclusion processing in the character image cutting device 10-6 may be combined with the expansion processing of the character image cutting device 10-1 to the character image cutting device 10-5. For example, FIG. 18 is a block diagram showing a configuration when the character image cutting device 10-6 is combined with the in-region image expansion processing in the character image cutting device 10-1.

  In the character image cutout device 10-6 configured as shown in FIG. 18, an extended in-region image recognition unit 1014 is interposed between the in-region image excluding unit 6011 and the image data output unit 1015. In this case, the extended in-region image recognition unit 1014 receives the in-region image data indicating the in-region image that has not been excluded from the in-region image excluding unit 6011, and expands the in-region image indicated by the received in-region image data. Processing is performed to recognize the image in the extended area. The extension area image recognition unit 1014 delivers the extension area image data indicating the recognized extension area image to the image data output unit 1015. The image data output unit 1015 transmits the received image data in the extended area to the character recognition device 12 as image data indicating one character.

  In this way, the image data transmitted from the image data output unit 1015 is higher in character recognition processing because it indicates an image in which the character portion that protrudes from the neighboring cells is excluded and the character portion that protrudes outside the cells is captured. The result will be accuracy.

  Further, instead of the in-region image excluding unit 6011 of the character image cutting device 10-6 excluding the in-region image on the basis of the number of on-pixels arranged inside and outside the cell, the in-region image excluding unit 6011, for example, Two sets of image data, that is, image data including an image in the extended area extending inside and outside and excluded image data, may be transmitted to the character recognition device 12 via the image data output unit 1015. In that case, the character recognition device 12 performs character recognition processing on each of the image data, and compares the recognition result with text data or the like indicating the registered word, thereby adopting a recognition result that seems to be correct. It may be.

  By the way, each of the character image cutout devices 10 described above may be realized by dedicated hardware, or may be realized by causing a general-purpose PC to execute processing according to an application program.

  Moreover, in embodiment mentioned above, although the character recognition system 1 demonstrated as what was implement | achieved by mutually connecting the character image cutting device 10, the tablet PC 11, and the character recognition device 12, those arrangements are changed arbitrarily. Is possible. For example, the character recognition device 12 may be arranged in the same casing as the character image cutting device 10, or the character image cutting device 10 may be realized by the tablet PC 11. Further, the character image cutting device 10 and the tablet PC 11 and the character image cutting device 10 and the character recognition device 12 may be connected to each other via a network.

  Further, in the above-described embodiment, it has been described that image data indicating a character string is input from the tablet PC 11 to the character image cutting device 10, but for example, instead of the tablet PC 11, an image is optically read. A scanner device that generates image data is connected to the character image cutting device 10, the character string written on the paper is read by the scanner device, and the image data generated by the scanner device is input to the character image cutting device 10. You may do it. In that case, for example, a figure showing a square is colored with, for example, vermilion, and after the figure showing the square is extracted by a color filter in the character image cutting device 10, a figure recognition process is performed to identify the area of the square. do it. Or you may make it specify the area | region of a grid only by a figure recognition process.

  Further, in the above-described embodiment, the case where an image for one character is cut out from a handwritten character string has been described. However, the character recognition system according to the present invention can also be used for printed characters that are shifted from the grid. Needless to say.

It is the block diagram which showed the structure of the character recognition system concerning embodiment of this invention. It is the figure which showed a mode that image data was cut out by the character image cutting device concerning 1st Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 1st Embodiment of this invention. It is the figure which showed a mode that image data was cut out by the character image cutting device concerning 2nd Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 2nd Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 2nd Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 2nd Embodiment of this invention. It is the figure which showed a mode that image data was cut out by the character image cutting device concerning 3rd Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 3rd Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 3rd Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 3rd Embodiment of this invention. It is the figure which showed a mode that image data was cut out by the character image cutting device concerning 4th Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 4th Embodiment of this invention. It is the figure which showed a mode that image data was cut out by the character image cutting-out apparatus concerning 5th Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 5th Embodiment of this invention. It is the figure which showed a mode that image data was cut out by the character image cutting-out apparatus concerning 6th Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 6th Embodiment of this invention. It is the block diagram which showed the structure of the character image cutting device concerning 6th Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Character recognition system, 10 ... Character image cutting device, 11 ... Tablet PC, 12 ... Character recognition device, 101 ... Control part, 102 ... Memory | storage part, 1011 ... Image data input part, 1012 ... Binarization part, 1013 ... Image recognition unit in region, 1014 ... Image recognition unit in extended region, 1015 ... Image data output unit, 1021 ... Image data, 1022 ... Region data, 1023 ... Binary image data, 2011 ... Expanded image generation unit, 2012 ... Extension Expansion region image recognition unit, 2013 ... Shrinkage image generation unit, 2014 ... Duplicate image recognition unit, 3011 ... Image recognition unit, 3012 ... Concatenated expansion image recognition unit, 3013 ... Intra-region connection expansion image recognition unit, 5011 ... Expansion region recognition Part, 6011...

Claims (11)

Image data acquisition means for acquiring image data consisting of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Of the pixels indicated by the pixel data included in the image data acquired by the image data acquisition means, the pixel is represented by a predetermined grid on the plane from among the on pixels that indicate an attribute value exceeding a predetermined threshold. In-region image recognition means for recognizing a collection of on-pixels arranged continuously in a region as an in-region image;
Determining means for determining whether there is a collection of on-pixels adjacent to the in-area image recognized by the in-area image recognition means and continuously arranged outside the area;
If the determination means includes a set of on-pixels, an image obtained by integrating the set of on-pixels into the image in the region is recognized as an image in the extension region, and is adjacent to the image in the region and mutually outside the region. Extended region image recognition means for recognizing the image in the region as an image in the extended region when there is no group of continuously arranged on pixels;
A character image cutting device comprising: output means for outputting data indicating an image in the extended area recognized by the image recognition means in the extended area as image data representing one character. Image data acquisition means for acquiring image data consisting of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Of the pixels indicated by the pixel data included in the image data acquired by the image data acquisition means, the pixel is represented by a predetermined grid on the plane from among the on pixels that indicate an attribute value exceeding a predetermined threshold. In-region image recognition means for recognizing a collection of on-pixels arranged continuously in a region as an in-region image;
Expansion image generation means for performing expansion processing on the in-region image recognized by the in-region image recognition means, and generating an in-expansion area image;
Determining means for determining whether there is a collection of on-pixels adjacent to the in-expansion area image generated by the inflated image generation means and continuously arranged outside the area;
When the determination unit includes the collection of on-pixels, an image obtained by integrating the collection of on-pixels with the image in the expansion area is recognized as an image in the expansion area, and is adjacent to the image in the expansion area. An expansion in-expansion region image recognition means for recognizing the in-expansion region image as an expansion in-expansion region image when there is no collection of on-pixels arranged continuously outside.
A character image cutout device comprising: output means for outputting data indicating an image in the expanded expansion area recognized by the image recognition means in the expanded expansion area as image data representing one character. Further comprising contraction image generation means for performing contraction processing on the image in the expansion expansion area recognized by the image recognition means in the expansion expansion area and generating an image in the contraction expansion expansion area,
3. The character according to claim 2, wherein the output unit outputs data indicating the image in the contraction / expansion / expansion region generated by the contraction image generation unit, instead of data indicating the image in the expansion / expansion region. Image cutting device. Image data acquisition means for acquiring image data consisting of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Of the pixels indicated by the pixel data included in the image data acquired by the image data acquisition means, the pixel is represented by a predetermined grid on the plane from among the on pixels that indicate an attribute value exceeding a predetermined threshold. In-region image recognition means for recognizing a collection of on-pixels arranged continuously in a region as an in-region image;
Expansion image generation means for performing expansion processing on the in-region image recognized by the in-region image recognition means, and generating an in-expansion area image;
Determining means for determining whether there is a collection of on-pixels adjacent to the in-expansion area image generated by the inflated image generation means and continuously arranged outside the area;
When the determination unit includes the collection of on-pixels, an image obtained by integrating the collection of on-pixels with the image in the expansion area is recognized as an image in the expansion area, and is adjacent to the image in the expansion area. An expansion in-expansion region image recognition means for recognizing the in-expansion region image as an expansion in-expansion region image when there is no collection of on-pixels arranged continuously outside.
A set of pixels that are on-pixels is recognized as a duplicate image in both the image indicated by the image data acquired by the image data acquisition unit and the image in the expansion expansion region recognized by the image recognition unit in the expansion expansion region. Duplicate image recognition means;
A character image cutting device comprising: output means for outputting data indicating a duplicate image recognized by the duplicate image recognition means as image data representing one character. Image data acquisition means for acquiring image data consisting of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Of the pixels indicated by the pixel data included in the image data acquired by the image data acquisition means, the pixel is represented by a predetermined grid on the plane from among the on pixels that indicate an attribute value exceeding a predetermined threshold. In-region image recognition means for recognizing a collection of on-pixels arranged continuously in a region as an in-region image;
Determining means for determining whether there is a collection of on-pixels adjacent to the in-area image recognized by the in-area image recognition means and continuously arranged outside the area;
If the determination means includes a set of on-pixels, an image obtained by integrating the set of on-pixels into the image in the region is recognized as an image in the extension region, and is adjacent to the image in the region and mutually outside the region. Extended region image recognition means for recognizing the image in the region as an image in the extended region when there is no group of continuously arranged on pixels;
Extended area recognition means for recognizing an area of a predetermined shape on the plane circumscribing an image in the extension area recognized by the image recognition means in the extension area as an extension area;
Output means for outputting data indicating an image indicated by a collection of on-pixels included in the extension area recognized by the extension area recognition means as image data representing one character. Out device. The in-region image recognition unit re-executes the in-region image recognition processing with the extension region recognized by the extension region recognition unit as the region after the extension region is recognized by the extension region recognition unit. ,
The extended area image recognition means re-executes the recognition process of the extended area image for the image in the area recognized by re-execution of the recognition process by the intra-area image recognition means,
The extension area recognition means re-executes the extension area recognition process on the extension area image recognized by re-execution of the recognition process by the extension area image recognition means,
The output means outputs data indicating an image indicated by a collection of on-pixels included in an extended area recognized by re-execution of recognition processing by the extended area recognition means as image data representing the one character. The character image cutting device according to claim 5. Image data acquisition means for acquiring image data consisting of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Of the pixels indicated by the pixel data included in the image data acquired by the image data acquisition means, the pixel is represented by a predetermined grid on the plane from among the on pixels that indicate an attribute value exceeding a predetermined threshold. In-region image recognition means for recognizing a collection of on-pixels arranged continuously in a region as an in-region image;
Regarding at least one intra-region image among one or more intra-region images recognized by the intra-region image recognition means, a collection of on-pixels arranged adjacent to the intra-region image and continuously outside the region. Determination means for determining whether or not it exists;
An intra-region image excluding unit for excluding the intra-region image from the intra-region image collection recognized by the intra-region image recognition unit when the determination unit includes the set of on-pixels;
Output means for outputting, as image data representing one character, a collection of intra-area images that are not excluded by the intra-area image exclusion means among the intra-area images recognized by the intra-area image recognition means. A character image cutting device. The area in the image excluding means relates the first region in an image, if the set of ON pixels that are arranged consecutively to each other in front Symbol territory outside and adjacent to the region in an image is present, the collection of the ON pixels When the area occupied by the image in one region is smaller than the area occupied, the image in the one region is excluded from the collection of images in the region recognized by the image recognition means in the region, and the collection of the on pixels occupies. The character image cutting device according to claim 7 , wherein when the area occupied by the image in one area is larger than the area, the image in the one area is not excluded from the collection of images in the area . When there is a collection of on-pixels adjacent to the in-area image that has not been excluded by the in-area image excluding means and continuously arranged outside the area, the on-pixel collection is used as the in-area image. If the integrated image is recognized as an image in the extension area and there is no collection of on pixels that are adjacent to the image in the area and are continuously arranged outside the area, the image in the area is determined to be an image in the extension area. Further comprising an image recognition means in the extended area that recognizes as
The character image cutout according to claim 8 , wherein the output means outputs data indicating the image in the extended area recognized by the image recognition means in the extended area as image data representing the one character. apparatus. Processing for obtaining image data composed of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Among the pixels indicated by the pixel data included in the acquired image data, among the on pixels indicating attribute values exceeding a predetermined threshold value, the pixels are continuous with each other within the area represented by the predetermined grid on the plane. Processing for recognizing a collection of on-pixels arranged as an in-region image,
A process of determining whether there is a collection of on-pixels adjacent to the recognized in-region image and continuously arranged outside the region;
Recognizing an image obtained by integrating the collection of the ON pixels in the area in the image in the case where collection of pre SL on pixel exists as an extended area within the image, and adjacent to the area in the image to one another continuously outside the said region A process of recognizing the image in the area as an image in the extension area when there is no collection of arranged on pixels,
A program for causing a computer to execute a process of outputting data indicating a recognized image in an extended area as image data representing one character. Processing for obtaining image data composed of a collection of pixel data indicating attribute values of each of a plurality of pixels constituting an image arranged on a plane;
Among the pixels indicated by the pixel data included in the acquired image data, among the on pixels indicating attribute values exceeding a predetermined threshold value, the pixels are continuous with each other within the area represented by the predetermined grid on the plane. Processing for recognizing a collection of on-pixels arranged as an in-region image,
For at least one in-region image among one or more recognized in-region images, it is determined whether there is a collection of on-pixels that are adjacent to the in-region image and are continuously arranged outside the region. Processing to
If a collection of pre-Symbol ON pixels are present, the process of excluding the region image from the collection of recognized areas in the image,
A program that causes a computer to execute a process of outputting a collection of intra-region images that are not excluded among recognized intra-region images as image data representing one character.

Images