JP6545592B2 - Image processing apparatus and image processing program - Google Patents

Description

  The present invention relates to an image processing apparatus and an image processing program, and more particularly to a technique for determining the type of a character described in a document.

  In an image forming apparatus such as a copying machine or a multi-function machine, it is necessary to determine what type of original the original read by the image input device (hereinafter referred to as “read original”) is. It is important to perform preferable image processing according to In particular, when the read document is a text document, it is important to determine whether the characters described in the document are handwritten characters. This is because, in the case of handwritten characters, it is necessary to perform expansion processing on image data of a read document in order to reduce blurring of characters and the like in the image.

  Until now, there has been proposed a technique for determining whether a character described in a document is a handwritten character (see, for example, Patent Document 1). Specifically, the certainty that the character described in the document is a handwritten character or a print character is quantified, and determination is made based on the numerical value. Then, the numerical value may be obtained by analyzing the gradation distribution of the characters in the image, one obtained by analyzing the color of the characters in the image, or an analysis based on the strokes of the characters in the image. And various numerical values such as those obtained by analyzing the position and size of characters in an image.

International Publication No. 2011/0174067

  However, in the above-mentioned judgment technique that has been proposed, it is necessary to perform numerical analysis based on various analyses, and thus complicated processing is required. Such complicated processing is time consuming and thus contrary to the recent demand for shortening the time from input to output of an image.

  Therefore, an object of the present invention is to provide an image processing apparatus and an image processing program capable of determining with high accuracy whether or not a character described in an original document is a handwritten character, in spite of simple processing. It is to provide.

  An image processing apparatus according to the present invention includes a character recognition processing unit, a recognized character counting unit, and a determination unit. The character recognition processing unit recognizes characters included in image data of a document. The recognition character counting unit counts the number of characters recognized by the character recognition processing unit to calculate the total number of recognized characters (the number of recognized characters). The determination unit determines, based on the total number (the number of recognized characters) calculated by the recognized character counting unit, whether or not the character described in the document is a handwritten character.

  The number of recognized characters tends to be significantly smaller in the manuscript of handwritten characters than in the manuscript of print characters. The reason is that the character form varies depending on the person writing the character, and therefore, the accuracy of character recognition in the character recognition processing unit is reduced. Then, determination based on such a tendency is performed by the determination unit.

  In the specific configuration of the image processing apparatus, the image processing apparatus further includes an edge extraction unit, an edge counting unit, and a first comparison unit. The edge extraction unit extracts edge pixels that constitute the edge of the image included in the image data. The edge counting unit calculates the total number of edge pixels (the number of edge pixels) by counting the number of edge pixels extracted by the edge extraction unit. The first comparison unit determines whether a first ratio, which is a ratio of the total number of edge pixels (number of edge pixels) to the total number (number of recognized characters) calculated by the recognized character counting unit, is greater than a first threshold. When the first comparison unit determines that the first ratio is greater than the first threshold, the determination unit determines that the character described in the document is a handwritten character.

  There is no significant difference in the number of edge pixels between the handwritten character manuscript and the print character manuscript. On the other hand, the number of recognized characters tends to be significantly smaller in the manuscript of handwritten characters than in the manuscript of print characters. Therefore, the first ratio tends to be significantly larger in the manuscript of handwritten characters than in the manuscript of print characters. Then, by setting the first threshold to an appropriate value based on such a tendency, the determination based on the magnitude relationship between the first ratio and the first threshold is realized.

In another specific configuration of the image processing device, the image processing device further includes a continuous character counting unit and a second comparison unit. Continuous character counting section, of the recognized characters in the character recognition processing section, by size counts the number of characters you continuously be determined and more than a predetermined one and the same, the total number of consecutive characters (Continuous Calculate the number of characters). The second comparison unit determines whether a second ratio, which is a ratio of the total number of consecutive characters (number of consecutive characters) to the total number (number of recognized characters) calculated by the recognized character counting unit, is smaller than a second threshold. When the second comparison unit determines that the second ratio is smaller than the second threshold, the determination unit determines that the character described in the document is a handwritten character.

  Since the character form (font and size) is often unified in a print character document, the number of recognized characters is a large value, and the number of consecutive characters tends to be a value close to the number of recognized characters. Therefore, the second ratio tends to be a value close to 1 in the case of printed characters. On the other hand, in manuscripts of handwritten characters, the form of the characters varies depending on the person writing the characters, and the size of the characters tends to be irregular. For this reason, the recognition character number tends to be a small value, and the continuous character number tends to be a further smaller value. Therefore, the second ratio tends to be a value close to 0 in a manuscript of handwritten characters. Then, by setting the second threshold to an appropriate value based on such a tendency, the determination based on the magnitude relationship between the second ratio and the second threshold is realized.

  Another image processing apparatus according to the present invention includes an edge extraction unit, a straight line detection unit, a straight line interval calculation unit, and a determination unit. The edge extraction unit extracts edge pixels constituting the edge of the image included in the image data of the document. The straight line detection unit detects a straight line corresponding to the edge of the image forming the straight line included in the image data, based on the edge pixel extracted by the edge extraction unit. The straight line interval calculation unit calculates a straight line interval detected by the straight line detection unit. The determination unit determines whether the character described in the document is a handwritten character based on the interval calculated by the linear interval calculation unit.

  If the interval calculated by the linear interval calculation unit matches the interval between the ruled lines attached to the existing document, the document is highly likely to have a ruled line. Then, there is a high possibility that the characters are handwritten on the document having the ruled lines. Accordingly, the determination unit can make a determination based on the interval calculated by the linear interval calculation unit.

  More specifically, the straight line interval calculation unit extracts one of the straight lines detected by the straight line detection unit that has a length equal to or greater than a predetermined length, and calculates the interval for the extracted straight line. As a result, among the straight lines detected by the straight line detection unit, one having a high possibility of corresponding to the ruled line is extracted. Therefore, the determination accuracy in the determination unit is improved.

  In still another specific configuration of the image processing apparatus, the image processing apparatus further includes an expansion processing unit, and the expansion processing unit determines that the character described in the document is a handwritten character by the determination unit. When it is done, the expansion process is applied to the image data. As a result, even if the density of handwritten characters is low or the handwritten characters are distorted in the original, the image after processing is in a state where the characters are easy to read.

  An image processing program according to the present invention causes an image processing apparatus to execute character recognition processing, recognition character counting processing, and determination processing. In the character recognition process, characters included in image data of a document are recognized. In the recognition character counting process, the total number of recognized characters (the number of recognized characters) is calculated by counting the number of characters recognized by the character recognition process. In the determination process, it is determined whether the character described in the document is a handwritten character based on the total number (the number of recognized characters) calculated by the recognition character counting process.

  Another image processing program according to the present invention causes an image processing apparatus to execute edge extraction processing, straight line detection processing, straight line interval calculation processing, and determination processing. In the edge extraction process, edge pixels constituting an edge of an image included in image data of a document are extracted. In the straight line detection process, a straight line corresponding to an edge of an image constituting a straight line included in image data is detected based on the edge pixel extracted by the edge extraction process. In the straight line interval calculation process, the distance between straight lines detected by the straight line detection process is calculated. In the determination process, it is determined whether the character described in the document is a handwritten character based on the interval calculated by the linear interval calculation process.

  According to the image processing apparatus and the image processing program according to the present invention, it is possible to determine with high accuracy whether the character described in the document is a handwritten character or not despite the simple processing. Become.

FIG. 2 is a block diagram showing the configuration of an image forming apparatus (multifunction machine) to which the image processing apparatus of the present invention is applied, and the flow of processing executed by the image processing apparatus when a copy mode is selected. FIG. 6 is a block diagram showing the configuration of an automatic document determination unit and the flow of determination processing executed by the automatic document determination unit. It is the figure which showed typically the image before (a) process by the expansion process part, and the image after (b) process. FIG. 6 is a block diagram showing the configuration of the character determination unit and the flow of processing when the first determination process is executed by the character determination unit of the automatic document determination unit. It is the flowchart which showed the flow of the 1st decision processing. It is the block diagram which showed the structure of the character recognition process part which a character determination part has, and the flow of the recognition process performed by a character recognition process part. It is a figure of the image data used for description of the character extraction by the layout analysis part which a character recognition process part has. FIG. 7 is a block diagram showing the configuration of the character determination unit and the flow of processing when the second determination process is executed by the character determination unit. It is the flowchart which showed the flow of 2nd determination processing. It is the flowchart which showed the flow of the processing which is run with consecutive character counting section. It is the block diagram which showed the flow of the structure and process of a character determination part when 3rd determination processing is performed in a character determination part. It is the flowchart which showed the flow of the 3rd decision processing. It is a figure used for description of the process in a straight line detection part, Comprising: (a) Coordinate system before Hough transformation (xy coordinate system), Coordinate system after Hough transformation (r-theta coordinate system) And FIG. It is the figure which showed the space | interval of two adjacent straight lines calculated by the linear space | interval calculation part. FIG. 16 is a block diagram showing the configuration of the character determination unit and the flow of processing when the fourth determination process is executed by the character determination unit. It is the block which showed the flow of composition and processing of a character judging part when the 1st-4th judging processing were combined. FIG. 6 is a block diagram showing the flow of processing executed by the image processing apparatus when an image transmission mode is selected for an image forming apparatus (multifunction peripheral). FIG. 6 is a block diagram showing the configuration of another image forming apparatus (color image reading apparatus) to which the image processing apparatus of the present invention is applied, and the flow of processing executed by the image processing apparatus of the image forming apparatus.

  Hereinafter, an embodiment in which the image processing apparatus of the present invention is applied to an image forming apparatus will be specifically described with reference to the drawings.

[1] First Embodiment [1-1] Configuration of Image Forming Device As shown in FIG. 1, the image forming device includes an image input device 1, an image processing device 2, an image output device 3, a transmitting and receiving device 5, and a storage. A unit 6 and a control unit 7 are provided. The image forming apparatus is a multifunction peripheral having various operation modes (copy mode, printer mode, facsimile transmission / reception mode, image transmission mode, etc.). FIG. 1 shows the operation of the image forming apparatus when the copy mode is selected.

«Image input device»
The image input device 1 is a device that optically reads an image of a document placed on a document table (not shown) or a document feeding device (not shown). The image input device 1 is, for example, a scanner having a CCD (Charge Coupled Device) line sensor, and separates light reflected from a document into three colors of R (red), G (green) and B (blue). And convert the light after decomposition into electric signals (hereinafter referred to as "RGB analog signals"). The RGB analog signal is input to the image processing device 2. The operation of the image input apparatus 1 is controlled by the control unit 7. Here, the control unit 7 is configured of a central processing unit (CPU) and a digital signal processor (DSP).

<< image processing device (processing when selecting copy mode) >>
When reading of a document is executed by the image input device 1, the image processing device 2 executes image processing based on the RGB analog signal generated by the image input device 1. The image processing performed at this time differs depending on the operation mode (copy mode, facsimile transmission / reception mode, image transmission mode, etc.) selected by the image forming apparatus.

  Here, with reference to FIG. 1, the process executed by the image processing apparatus 2 when the copy mode is selected will be described. Specifically, in the image processing apparatus 2, the A / D conversion unit 201, the shading correction unit 202, the input processing unit 203, the automatic document determination unit 204, the expansion processing unit 205, the area separation processing unit 206, the color correction unit 207, Each of the black generation / under color removal unit 208, the spatial filter unit 209, the output tone correction unit 210, and the halftone generation unit 211 executes processing. The control unit 7 controls operations performed by the respective units of the image processing apparatus 2 and operations such as delivery of signals and image data.

<A/D converter>
The A / D conversion unit 201 converts the RGB analog signal input to the image processing apparatus 2 into a digital signal (RGB digital signal) to generate image data composed of three colors of RGB (hereinafter referred to as “RGB image Data (referred to as data).

<Shading correction unit>
The shading correction unit 202 removes distortion generated in each of the illumination system, the imaging system, and the imaging system of the image input device 1 from the RGB image data generated by the A / D conversion unit 201.

<Input processing unit>
The input processing unit 203 performs γ correction processing on the RGB image data after the shading correction. The RGB image data after γ correction is stored in the storage unit 6. Then, the RGB image data stored in the storage unit 6 is managed by the control unit 7 as filing data.

  When the gamma-corrected RGB image data is stored in the storage unit 6, it is compressed into a JPEG code, for example, based on a JPEG compression algorithm. When image formation is performed in the image forming apparatus, the JPEG code is read from the storage unit 6 and the JPEG code is subjected to decoding processing. Hereinafter, RGB image data generated by the decoding process will be simply referred to as “RGB image data” unless otherwise specified. In FIG. 1, RGB image data directed from the input processing unit 203 to the automatic document determination unit 204 correspond to RGB image data generated by the decoding process.

<Auto Document Judgment Unit>
The automatic document determination unit 204 determines what type of document the document read by the image input device 1 (hereinafter referred to as “read document”) is based on the RGB image data input thereto. Determine. Specifically, according to the flow shown in FIG. 2, in the automatic document determination unit 204, the document determination unit 204a and the character determination unit 204b execute processing in order.

  Based on the RGB image data, the document determination unit 204a selects either a text document, a printed photo document, a document in which characters and a printed photo are mixed, or a photo document in which the gradation changes continuously. Determine if there is. As an example, a conventional discrimination technique is used for the discrimination method (see, for example, JP-A-2002-232708). The determination method is not limited to the conventional determination technique, and various modifications are possible.

  Data of the determination result in the document determination unit 204a is stored in the storage unit 6 as document determination data together with data (character determination data) of the determination result in the character determination unit 204b described later. Then, the document discrimination data is read out from the storage unit 6 as necessary, and the expansion processing unit 205, the color correction unit 207, the black generation / under color removal unit 208, the spatial filter unit 209, and the halftone generation unit 211. Used in

  When the document determination unit 204a determines that the read document is a text document, the character determination unit 204b receives the result and executes the determination process. Specifically, based on the RGB image data input to the automatic document determination unit 204, the character determination unit 204b determines whether the character described in the text document is a handwritten character. The document to be subjected to the determination process by the character determination unit 204b may include, in addition to the text document, a document or the like in which a printed picture or the like is partially mixed. Details of the determination process executed by the character determination unit 204b will be described later.

<Expansion processing unit>
When the document determination unit 204a determines that the read document is a text document, and the character determination unit 204b determines that the character described in the text document is a handwritten character, the result (document determination data ), The expansion processing unit 205 performs expansion processing on the RBG image data. Specifically, the expansion processing unit 205 performs image processing such that the character is thickened on the portion of the character included in the RBG image data. This is because in the case of handwritten characters (particularly, in the case of characters written with a pencil), there is a risk that the density of the characters may be light or the characters may be distorted.

  An example of the expansion processing will be described with reference to FIGS. 3 (a) and 3 (b). FIGS. 3A and 3B schematically show pixels having pixel values represented by 8 bits (256 gradations), and for convenience, the pixel values are 0, 100, and It shows three regions of 255. Regions in which the pixel values are 0 and 100 correspond to, for example, a character portion. In FIGS. 3A and 3B, only pixel values of one color component are shown in order to simplify the description.

  First, for each pixel shown in FIG. 3A, the expansion processing unit 205 compares the pixel value of the pixel (target pixel) with the pixel values of eight pixels (peripheral pixels) located around the pixel. . As a result, when any pixel value of the peripheral pixels is smaller than the pixel value of the target pixel, the expansion processing unit 205 replaces the pixel value of the target pixel with the pixel value of the smaller peripheral pixel (FIG. b) see). As a result, the pixel value of the target pixel is the smallest among the pixel values and the pixel values of the peripheral pixels.

  Such expansion processing enlarges the regions where the pixel values are 0 and 100. Therefore, even if the density of the handwritten character is thin or the handwritten character is distorted in the character original, the processed image is in a state where the character is easy to read.

<Area separation processing unit>
The region separation processing unit 206, based on RGB image data (RGB image data after expansion processing when expansion processing is performed by the expansion processing unit 205), the pixels of each pixel constituting the image data are Determine the type of image area to which it belongs. Here, in the image area, types such as a black character area, a color character area, and a halftone area exist. The data of the determination result is output from the region separation processing unit 206 and stored in the storage unit 6. Then, data of the discrimination result (region separation data) is read from the storage unit 6 as needed, and used by the black generation / under color removal unit 208, the spatial filter unit 209, and the halftone generation unit 211. Be done.

<Color correction unit>
The color correction unit 207 sets RGB image data (RGB image data after expansion processing when expansion processing is performed by the expansion processing unit 205) to C (cyan), M (magenta), and Y (yellow). It is converted into image data composed of three colors (hereinafter referred to as "CMY image data"). That is, the color space is converted from the RGB space to the CMY space in a complementary color relationship. That is, the color correction unit 207 performs processing to improve color reproducibility on RGB image data. At this time, the determination result (original determination data) in the automatic document determination unit 204 is used as needed.

<Black generation / under color removal part>
The black generation / under color removal unit 208 includes the CMY image data generated by the color correction unit 207 in four colors of C (cyan), M (magenta), Y (yellow), and K (black). It is converted into image data (hereinafter referred to as "CMYK image data"). That is, the color space is converted from the CMY space to the CMYK space. At this time, determination results (original determination data and area separation data) in the original document automatic determination unit 204 and the area separation processing unit 206 are used as needed.

<Space filter section>
The spatial filter unit 209 subjects the CMYK image data generated by the black generation / undercolor removal unit 208 to processing such as enhancement processing and smoothing processing. At this time, determination results (original determination data and area separation data) in the original document automatic determination unit 204 and the area separation processing unit 206 are used as needed.

<Output tone correction unit>
The output tone correction unit 210 outputs the CMYK image data processed by the spatial filter unit 209 to a recording medium such as a sheet of paper so that the output image can have appropriate brightness. An output γ correction process is performed.

<Half tone generation unit>
When the halftone generation unit 211 outputs the CMYK image data processed by the output gradation correction unit 210 to a recording medium such as a sheet, the output image reproduces the gradation of the image of the read document. Perform tone reproduction processing so that At this time, determination results (original determination data and area separation data) in the original document automatic determination unit 204 and the area separation processing unit 206 are used as needed.

«Image output device»
The image output device 3 executes image formation on a sheet based on the CMYK image data processed by the halftone generation unit 211. The image output apparatus 3 is an apparatus for printing an image on a sheet, such as an electrophotographic printer or an inkjet printer. The sheet includes various sheet-like recording media such as plain paper, thick paper, and OHP film.

[1-2] Configuration of Character Determination Unit and Determination Processing Method (First Determination Processing)
Next, the first determination process executed by the above-described character determination unit 204b will be specifically described. The first determination process is a process of determining whether the character described in the text document is a handwritten character based on the total number of characters recognized from the RGB image data. The second and third determination processes executed by the character determination unit 204b and the other determination processes will be described in the second and subsequent embodiments.

  When the first determination process is executed by the character determination unit 204b, according to the flow shown in FIGS. 4 and 5, the character determination unit 204b includes the edge extraction unit 401, the character recognition processing unit 402, the edge counting unit 403, Each of the recognized character counting unit 404, the first comparison unit 405, and the first determination unit 406 executes a process.

<Edge extraction unit>
The edge extraction unit 401 extracts an edge of an image such as a character included in the image data based on the RGB image data input to the document automatic determination unit 204 (step S11 in FIG. 5). Specifically, the edge extraction unit 401 extracts pixels (hereinafter, referred to as “edge pixels”) that constitute an edge of an image. As an example, a conventional extraction method is used for the extraction method (for example, JP-A-2009-94903). The extraction method is not limited to the conventional extraction method, and various modifications are possible.

<Character recognition processing unit>
The character recognition processing unit 402 recognizes characters included in the image data based on the RGB image data input to the document automatic determination unit 204 (step S12 in FIG. 5). Specifically, the character recognition processing unit 402 extracts characters included in RGB image data, and determines what character the extracted characters are. More specifically, according to the flow shown in FIG. 6, in the character recognition processing unit 402, each of the signal conversion unit 402a, the binarization processing unit 402b, the layout analysis unit 402c, and the character recognition unit 402d executes processing. .

(A) Signal Converter The signal converter 402 a converts an RGB digital signal into a luminance signal. Specifically, the signal conversion unit 402a converts the pixel value (Ir, Ig, Ib) of each pixel constituting the RGB image data into the luminance value Y. For example, the following formula (1) is used for this conversion. The data obtained by the conversion is hereinafter referred to as "brightness data".

The conversion to the luminance value Y is not limited to the conversion according to the equation (1), and various modifications are possible. Further, the signal conversion unit 402a converts RGB digital signals into signals such as CIE 1976 L ^* a ^* b ^* signals (CIE: Commission Internationale de l'Eclairage, L ^* : lightness, a ^* and b ^* : chromaticity). May be Furthermore, the signal conversion unit 402a may extract only the G signal from the RGB signal, instead of converting the RGB digital signal into another signal.

(B) Binarization Processing Unit The binarization processing unit 402b performs threshold processing on the luminance data generated by the signal conversion unit 402a. For example, when the luminance value Y of each pixel constituting the luminance data is represented by 8 bits, the binarization processing unit 402b determines, for each pixel, that the luminance value Y is equal to or greater than a threshold (for example, 128). Converts the luminance value Y into 255 (white), and when the luminance value Y is smaller than the threshold value, converts the luminance value Y into 0 (black). Note that as the threshold value, an average value of luminance values Y may be used that each of a plurality of pixels including the pixel of interest (for example, 5 × 5 pixels centered on the pixel of interest) has.

(C) Layout Analysis Unit The layout analysis unit 402c performs layout analysis based on the luminance data (image data) after binarization processing, and thereby characters (including symbols and the like) included in the luminance data. Extract. Specifically, the layout analysis unit 402c first assigns different numbers i to a plurality of black portions by performing a labeling process on luminance data. Here, the black part is a series of black pixels, and the same number i is given to the pixels constituting the same black part. Thereby, each of the labeled black parts is extracted as one character.

  Thereafter, as shown in FIG. 7, the layout analysis unit 402c sets a circumscribed rectangle Ra (i) for each of the labeled black portions. At this time, the layout analysis unit 402c extracts the pixels located at the uppermost end, the lowermost end, the rightmost end, and the leftmost end from the pixels forming the black part of the number i. Find the coordinates in the Cartesian coordinate system. Then, the circumscribed rectangle Ra (i) is set based on those coordinates. Then, the number i of characters extracted from the luminance data (hereinafter referred to as “extracted characters”) and the circumscribed rectangle Ra (i) are stored in the storage unit 6 in association with each other. At this time, as data representing the circumscribed rectangle Ra (i), for example, the coordinates of the upper left vertex and the coordinates of the lower right vertex of the circumscribed rectangle Ra (i) are used.

(D) Character Recognition Unit The character recognition unit 402d determines what kind of character the extracted character obtained by the layout analysis unit 402c is. Specifically, first, the character recognition unit 402d changes the size of the extracted character so as to be the same as the size of the registered character. Here, the registered characters are print characters of a predetermined size (for example, 10 points) registered in the storage unit 6 as dictionary data. More specifically, for each of the extracted characters (number i), the character recognition unit 402d sets the width of the extracted characters so that the size of the circumscribed rectangle Ra (i) corresponds to the size of the registered character. The height is enlarged or reduced, respectively. That is, the character recognition unit 402d normalizes the size of the extracted character.

  Thereafter, the character recognition unit 402d collates the normalized extracted character with the registered character in the dictionary data. As a result, when the extracted character matches any of the registered characters, the character recognition unit 402d recognizes the extracted character as the same character as the matched registered character.

  The data of the recognition result obtained by the character recognition unit 402d obtained in this way is character recognition data together with information on the extracted character (for example, assigned number i, vertex coordinates of circumscribed rectangle Ra (i), etc.) , And from the character recognition processing unit 402 (see FIG. 4).

<Edge counting unit>
The edge counting unit 403 counts the number of edge pixels extracted by the edge extracting unit 401. Thus, the edge counting unit 403 calculates the total number of edge pixels (the number of edge pixels Ce) included in the RGB image data (step S13 in FIG. 5). Then, the calculated edge pixel number Ce is output from the edge counting unit 403 as data.

<Recognized Character Counting Unit>
The recognized character counting unit 404 counts the number of characters recognized by the character recognition processing unit 402. Thus, the recognized character counting unit 404 calculates the total number of characters recognized from the RGB image data (the number of recognized characters Cr) (step S14 in FIG. 5). Then, the calculated recognition character number Cr is output from the recognition character counting unit 404 as data.

<First comparison unit>
The first comparison unit 405 calculates the ratio Ce / Cr based on the number of edge pixels Ce output from the edge counting unit 403 and the recognition character number Cr output from the recognized character counting unit 404. After that, the first comparison unit 405 determines whether the calculated ratio Ce / Cr is larger than the threshold Th1 (step S15 in FIG. 5). Here, the threshold Th1 is set to, for example, 10000.

<First determination unit>
The first determination unit 406 determines, based on the determination result of the first comparison unit 405, whether the character described in the text document is a handwritten character or a printed character. Specifically, when the first comparison unit 405 determines that “the ratio Ce / Cr is larger than the threshold Th1”, the first determination unit 406 determines that the character described in the character original is a handwritten character. (Step S16 in FIG. 5). On the other hand, when it is determined by the first comparison unit 405 that “the ratio Ce / Cr is not larger than the threshold Th1”, the first determination unit 406 states that “the character described in the character original is a print character”. It determines (step S17 of FIG. 5).

  The edge pixel number Ce does not have a large difference between the manuscript of the handwritten character and the manuscript of the print character. On the other hand, the recognition character number Cr tends to be significantly smaller in the manuscript of the handwritten character than in the manuscript of the print character. The reason is that the character form varies depending on the person writing the character, and the accuracy of character recognition in the character recognition processing unit is reduced. Therefore, the ratio Ce / Cr tends to be significantly larger in the handwritten character original as compared to the printed character original. Then, by setting the threshold value Th1 to an appropriate value based on such a tendency, the determination based on the magnitude relationship between the ratio Ce / Cr and the threshold value Th1 is realized.

  The data (character determination data) of the determination result of the character determination unit 204b obtained in this manner is stored in the storage unit 6 as document determination data together with the data of the determination result of the document determination unit 204a described above. Then, the document discrimination data is read out from the storage unit 6 as necessary, and the expansion processing unit 205, the color correction unit 207, the black generation / under color removal unit 208, the spatial filter unit 209, and the halftone generation unit 211. Used in

  According to the first determination process described above, in spite of the simple process, it is possible to determine with high accuracy whether or not the character described in the document is a handwritten character. The first determination process can be applied not only to Japanese but also to foreign languages.

  The first determination process described above may be realized by causing the image processing apparatus to execute a determination process program. Then, such a determination processing program may be recorded on a recording medium (hard disk, memory card, etc.) in a readable state.

[2] Second Embodiment [2-1] Configuration of Character Determination Unit and Determination Processing Method (Second Determination Processing)
The second determination process executed by the character determination unit 204b will be specifically described. The second determination process is a process of determining whether the character described in the text document is a handwritten character based on the total number of characters recognized from the RGB image data.

  When the second determination process is executed by the character determination unit 204b, according to the flow shown in FIGS. 8 and 9, the character determination unit 204b recognizes the character recognition processing unit 402, the recognized character counting unit 404, and the continuous character counting unit Each of 407, the 2nd comparison part 408, and the 2nd determination part 409 performs a process. The processes (steps S21 and S22 in FIG. 9) executed by the character recognition processing unit 402 and the recognized character counting unit 404 are as described in the first embodiment.

<Consecutive character counting unit>
Continuous character counting section 407, among the recognized characters in the character recognition processing unit 402, and a predetermined number of (N. For example N = 5) is determined to be the same size the number of more continuous to Rubun shaped Count. Thus, the continuous character counting unit 407 calculates the total number of continuous characters (the number of consecutive characters Ca) (step S23 in FIG. 9).

  Specifically, the consecutive character counting unit 407 calculates the number of consecutive characters Ca in accordance with the flow shown in FIG. First, the consecutive character counting unit 407 sets all the number i, the count variable p, and the number of consecutive characters Ca to 0, and sets the parameter N to 5 (step S201).

  Here, the number i is a number assigned to the extracted character A (i) by the layout analysis unit 402c of the character recognition processing unit 402 (see FIG. 7). Specifically, when the read original is a horizontally written character original, the number i is sequentially assigned to the extracted characters arranged on the right with the extracted character located at the upper left as the first (i = 1) (i = 2, ...). Also, when the number m (i = m) is assigned to the rightmost extracted character of a line, the next number m + 1 (i = m + 1) is assigned to the extracted character located at the left end of the next lower line. On the other hand, when the read original is a vertically-written character original, the extracted character in the upper right is regarded as the first (i = 1), and the number i is sequentially assigned to the extracted characters arranged in the lower side (i = 2,. ). Also, when the number m (i = m) is assigned to the extracted character at the lower end of a certain row, the next number m + 1 (i = m + 1) is allocated to the extracted character located at the upper end of the left adjacent row.

  Incidentally, whether the read original is a horizontal original or a vertical original is determined, for example, as follows. That is, as shown in FIG. 7, from the vertex coordinates of the circumscribed rectangle Ra (i), the distance h1 between two adjacent extracted characters in the horizontal direction and the distance h2 between two adjacent extracted characters in the vertical direction are calculated. And these intervals h1 and h2 are compared. When the horizontal interval h1 is smaller than the vertical interval h2, it is determined that horizontal writing is performed, and when the horizontal interval h1 is larger than the vertical interval h2, it is determined that vertical writing is performed.

  The parameter N is a reference when determining the continuity of characters. Specifically, the parameter N is a standard for determining whether or not several characters are counted as consecutive characters. The parameter N may be set to a natural number other than 5 (preferably 2 or more).

  After execution of step S201, the continuous character counting unit 407 increments the number i by 1 in step S202, and recognizes information of the extracted character A (i) corresponding to the number i at that time in step S203. Get from the data. Thereafter, in step S204, continuous character counting unit 407 determines whether or not extracted character A (i) corresponding to number i is a character recognized by character recognition processing unit 402 based on the acquired information. .

  When the determination result that "the extracted character A (i) is a recognized character" is obtained in step S204, the continuous character counting unit 407 determines whether the count variable p is equal to 0 in step S205. To judge. On the other hand, when the determination result that "the extracted character A (i) is not a recognized character" is obtained in step S204, the continuous character counting unit 407 proceeds to step S213.

  If the determination result that "the count variable p is not equal to 0" is obtained in step S205, the continuous character counting unit 407 proceeds to the next step S208. On the other hand, when the determination result that "the count variable p is equal to 0" is obtained in step S205, the continuous character counting unit 407 changes the value of the count variable p from 0 to 1 in step S206. Thereafter, the consecutive character counting unit 407 proceeds to step S213.

  In step S208, the consecutive character counting unit 407 determines that the size of the extracted character A (i) corresponding to the number i is the size of the extracted character A (i-1) corresponding to the immediately preceding number (i-1). Determine if they are the same. Specifically, continuous character counting unit 407 compares the sizes of circumscribed rectangles Ra (i) and Ra (i-1) corresponding to these extracted characters A (i) and A (i-1), respectively. Thus, it is determined whether the sizes of the extracted characters A (i) and A (i-1) are the same. At this time, the continuous character counting unit 407 may make the determination by comparing the heights of the circumscribed rectangles Ra (i) and Ra (i-1), or may determine the circumscribed rectangles Ra (i) and Ra (i-1). Judgment may be made by comparing both the height and width of

  More specifically, when comparing the heights of the circumscribed rectangles Ra (i) and Ra (i-1), the consecutive character counting unit 407 sets the heights of the circumscribed rectangles Ra (i) and Ra (i). Recognize by the number of pixels in the height direction for i-1). Also, when comparing the width of circumscribed rectangles Ra (i) and Ra (i-1), the consecutive character counting unit 407 determines the width of each of the circumscribed rectangles Ra (i) and Ra (i-1). Recognize by the number of pixels in the horizontal direction. Then, the consecutive character counting unit 407 determines whether the sizes are the same by comparing the number of pixels. At this time, the continuous character counting unit 407 determines that the sizes are the same if the difference between the two pixel numbers to be compared is within a predetermined range.

  When the determination result that "the size is the same" is obtained in step S208, the continuous character counting unit 407 increases the value of the count variable p by 1 in step S209. That is, in step S209, the number of consecutive characters is counted. Thereafter, the continuous character counting unit 407 proceeds to the next step S213. On the other hand, when the determination result that "the size is not the same" is obtained in step S208, the continuous character counting unit 407 determines whether the value of the count variable p is equal to or more than the parameter N in step S210.

  If it is determined in step S210 that "the value of the count variable p is equal to or greater than the parameter N", the continuous character counting unit 407 determines the value of the number of consecutive characters Ca by the value of the count variable p in step S211. Enlarge. On the other hand, when the determination result that “the value of the count variable p is not the parameter N or more” is obtained in step S210, the continuous character counting unit 407 does not change the value of the number of continuous characters Ca (that is, step ) And proceeds to step S212. That is, only when the number of consecutive characters is N or more, the value of the count variable p is counted as the number of consecutive characters Ca. In step S212, the continuous character counting unit 407 resets the count variable p to zero. Thereafter, the continuous character counting unit 407 proceeds to the next step S213.

  In step S213, the consecutive character counting unit 407 determines whether the number i has reached the last number E. Here, the number E corresponds to the total number of extracted characters extracted by the layout analysis unit 402c. When the determination result that "the number i has reached the last number E" is obtained in step S213, the continuous character counting unit 407 proceeds to step S214. On the other hand, when the determination result that "the number i has not reached the last number E" is obtained in step S213, the consecutive character counting unit 407 returns to step S202, and executes the processing from step S202 again. Do.

  In step S214, the continuous character counting unit 407 determines whether the value of the count variable p is equal to or greater than the parameter N. When the determination result that "the value of the count variable p is equal to or greater than the parameter N" is obtained in step S214, the continuous character counting unit 407 determines the value of the number of consecutive characters Ca as the count variable p in step S215. Increase the value, and then finish character counting. On the other hand, when the determination result that "the value of count variable p is not the parameter N or more" is obtained in step S214, the consecutive character counting unit 407 does not change the value of the number of consecutive characters Ca, that is (step The character count process is ended without executing S215).

  That is, while counting the number of consecutive characters (when the count variable p has a value of 1 or more), the number of characters being counted is, even when the number i reaches the final number E. If the number is N or more, it is counted as the number of consecutive characters Ca.

  In the above counting process, when N or more consecutive characters targeted for counting are considered as one character group, all characters belonging to the same character group are required to have the same size. The size may be different. Then, according to the above-described counting process, the total number of characters (the number of consecutive characters Ca) belonging to the character groups is calculated for all the character groups. The calculated number of consecutive characters Ca is output from the continuous character counting unit 407 as data.

<Second comparison unit>
The second comparison unit 408 calculates the ratio Ca / Cr based on the number of consecutive characters Ca output from the consecutive character counting unit 407 and the number of recognized characters Cr output from the recognized character counting unit 404. After that, the second comparing unit 408 determines whether the calculated ratio Ca / Cr is smaller than the threshold Th2 (step S24 in FIG. 9). Here, the threshold value Th2 is set to, for example, 0.4.

<Second determination unit>
The second determination unit 409 determines, based on the determination result of the second comparison unit 408, whether the character described in the text document is a handwritten character or a printed character. Specifically, when the second comparison unit 408 determines that “the ratio Ca / Cr is smaller than the threshold Th2”, the second determination unit 409 determines that the characters described in the character original are handwritten characters. (Step S25 in FIG. 9). On the other hand, when it is determined in the second comparison unit 408 that “the ratio Ca / Cr is not smaller than the threshold Th2”, the second determination unit 409 states that “the character described in the character original is a print character”. It determines (step S26 of FIG. 9).

  Since the character form (font and size) is often unified in a print character document, the recognition character number Cr tends to be a large value, and the continuous character number Ca tends to be a value close to the recognition character number Cr. Therefore, the ratio Ca / Cr tends to be a value close to 1 in a print character document. On the other hand, in manuscripts of handwritten characters, the form of the characters varies depending on the person writing the characters, and the size of the characters tends to be irregular. Therefore, the recognition character number Cr tends to be a small value, and the continuous character number Ca is likely to be a further smaller value (value close to 0). Therefore, in the manuscript of handwritten characters, the ratio Ca / Cr tends to be a value close to zero. Then, by setting the threshold value Th2 to an appropriate value based on such a tendency, the determination based on the magnitude relationship between the ratio Ca / Cr and the threshold value Th2 is realized.

  According to the second determination process described above, in spite of the simple process, it is possible to determine with high accuracy whether or not the character described in the document is a handwritten character. The second determination process can be applied not only to Japanese but also to foreign languages.

  The second determination process described above may be realized by causing the image processing apparatus to execute a determination process program. Then, such a determination processing program may be recorded on a recording medium (hard disk, memory card, etc.) in a readable state.

[3] Third Embodiment [3-1] Configuration of Character Determination Unit and Determination Method (Third Determination Process)
The third determination process executed by the character determination unit 204b will be specifically described. In the third determination process, an image (hereinafter referred to as a "linear image") constituting a straight line included in RGB image data is detected, and a character described in the character original is handwritten based on the linear image. It is processing to determine whether or not it is a character.

  When the third determination process is executed by the character determination unit 204b, according to the flow shown in FIGS. 11 and 12, the edge determination unit 401, the straight line detection unit 410, the linear interval calculation unit 411, and the like in the character determination unit 204b. Each of the third comparison unit 412 and the third determination unit 413 executes a process. The process (step S31 in FIG. 12) executed by the edge extraction unit 401 is as described in the first embodiment.

<Line detection unit>
The straight line detection unit 410 detects a straight line L (j) corresponding to the edge of the straight line image included in the RGB image data based on the edge pixel extracted by the edge extraction unit 401 (step S32 in FIG. 12). The Hough transform is used to detect such a straight line L (j).

  Specifically, the straight line detection unit 410 sets polar coordinates expressed by the following equation (2) based on the coordinates (x0, y0) of each edge pixel in the two-dimensional orthogonal coordinate system (xy coordinate system). Generate a curve in a system (r-θ coordinate system). Here, the variables r and θ are defined as shown in FIG. 13A in the xy coordinate system, and the straight line in the xy coordinate system is expressed by the following equation (3) Ru. That is, the coordinates (r, θ) on the curve expressed by the equation (2) represent all straight lines having different slopes passing through the coordinates (x0, y0) through the equation (3).

  Next, the straight line detection unit 410 extracts the intersections of the curves of Formula (2) generated respectively from the coordinates of all edge pixels. Here, as shown in FIG. 13B, if n curves intersect at one point (r0, θ0), n edge pixels respectively corresponding to those curves have the following formula ( It will be located on one straight line expressed by 4).

  Thus, the straight line detection unit 410 can extract a straight line corresponding to the intersection point by extracting the intersection point of the curve. The straight line detection unit 410 is not limited to the case where a plurality of curved lines cross each other at one point and is not limited to the case of extracting a plurality of intersecting points. ) Is extracted as an intersection point corresponding to one straight line.

  Here, the edge pixels include not only those constituting the edge of the straight line image corresponding to the ruled lines but also those constituting the edge of the image such as characters. Therefore, the intersections include intersections formed by intersections of curves corresponding to several edge pixels (incidentally, edge pixels that are merely located on one straight line) that have nothing to do with the straight line image. Therefore, the straight line detection unit 410 needs to exclude such an intersection point.

  Specifically, the straight line detection unit 410 sets the above-described predetermined range in each of the areas where the intersections are densely targeted. Then, the straight line detection unit 410 counts the number of curves forming the intersection included in the predetermined range of each region (or the number of edge pixels respectively corresponding to those curves). Thus, the straight line detection unit 410 calculates the total number of curves (the number of curves) for each of the dense regions. Thereafter, the straight line detection unit 410 determines whether the calculated number of curves is larger than a threshold. Then, when the determination result that “the number of curves is larger than the threshold value” is obtained, the straight line detection unit 410 detects edge pixels respectively corresponding to those curves as one pixel group B (j) (symbol j is Represents a number that distinguishes pixel groups).

  Thereafter, for each of the extracted pixel groups B (j), the straight line detection unit 410 expresses the following equation (5) representing one straight line L (j) based on the coordinates of a plurality of edge pixels included in the pixel group. Calculate). Here, {−a (j) / b (j)} and {−c (j) / calculated based on coefficients a (j), b (j), and c (j) of equation (5) b (j)} represents the slope and y-intercept of the straight line L (j) in the xy coordinate system, respectively. For example, an approximation method such as the least square method is used to calculate the slope and the y intercept of the straight line L (j).

<Linear interval calculation unit>
Based on the straight line L (j) detected by the straight line detection unit 410, the straight line distance calculation unit 411 sets the distance d (k) between two adjacent straight lines L (k) and L (k + 1) to Calculate according to 6) (step S33 in FIG. 12). This equation (6) is, as shown in FIG. 14, the length of the perpendicular to the straight line L (k + 1) from any coordinate (x0 (k), y0 (k)) located on the straight line L (k) Is expressed as an interval d (k).

    Preferably, among the straight lines L (j) detected by the straight line detection unit 410, the straight line distance calculation unit 411 extracts one having a high probability of corresponding to the ruled line, and the distance d (k Calculate). This improves the comparison accuracy in the third comparison unit 412 described later. As an example, the straight line interval calculation unit 411 extracts, of the straight lines L (j), one having a straight line length equal to or greater than a predetermined length. As the predetermined length, for example, a length that is a predetermined ratio (for example, 70%) of the size of the character original (horizontal width in horizontal writing, vertical width in vertical writing) is employed.

<Third comparison unit>
The third comparison unit 412 determines whether the straight line L (j) detected by the straight line detection unit 410 corresponds to a ruled line, based on the interval d (k) calculated by the straight line interval calculation unit 411. (Step S34 in FIG. 12).

  Specifically, the third comparison unit 412 calculates an average value of the intervals d (k), and the fluctuation width of the intervals d (k) from the average value is within a predetermined error range (for example, ± 0.03 mm) If the resolution is 600 dpi, it is determined whether or not the number of pixels is about 1).

  Then, when the determination result that “the swing width is within the predetermined error range” is obtained, the third comparison unit 412 adds an interval d (k) (for example, an average value) to the existing document. It is determined whether or not the intervals between the ruled lines coincide with each other within a predetermined error range (for example, ± 0.03 mm; in the case of a resolution of 600 dpi, about one pixel in terms of the number of pixels). The interval between the ruled lines is stored, for example, in the storage unit 6, and the third comparison unit 412 reads the interval between the ruled lines from the storage unit 6 as necessary. The following exist as existing ruled lines. The numerical values shown in parentheses below are the interval between the ruled lines and the number of pixels corresponding to this (when the resolution is 600 dpi).

  As existing ruled lines, there are those attached to notes and stationery. In Japan, A-line (7 mm (168 pixels)), B-line (6 mm (144 pixels)), C-line (5 mm (120 pixels)), U-line (9 mm (216 pixels)), etc. exist as notebook ruled lines Do. In the West, college rules (7.15 mm (172 pixels)), spell lights (8.20 mm (197 pixels)), wide rules (8.77 mm (211 pixels)), legal rules (8.92 mm (215 pixels) ), Greg rule (8.80 mm (211 pixels)), and the like.

<Third determination unit>
The third determination unit 413 determines, based on the determination result of the third comparison unit 412, whether the character described in the text document is a handwritten character. Here, when the third comparison unit 412 determines that “the interval d (k) matches the interval between the ruled lines attached to the existing document”, the character document is attached with a ruled line. There is a high possibility of Then, there is a high possibility that the characters are handwritten on the document having the ruled lines. Therefore, when it is determined in the third comparison unit 412 that “the interval d (k) matches the interval of the ruled lines attached to the existing document”, the third determination unit 413 determines that “the character document It is determined that the character described in is a handwritten character "(step S35 in FIG. 12). On the other hand, when it is determined in the third comparison unit 412 that “the interval d (k) does not match the interval of the ruled lines attached to the existing document”, the third determination unit 413 “the text document It is determined that the character described in is not a handwritten character "(step S36 in FIG. 12). The third determination unit 413 calculates the number of straight lines L (j) detected by the straight line detection unit 410, and in consideration of the number, the characters described in the character original are handwritten characters. It may be determined whether or not

  According to the third determination process described above, in spite of the simple process, it is possible to determine with high accuracy whether the character described in the document is a handwritten character. The third determination process can be applied not only to Japanese but also to foreign languages.

  The third determination process described above may be realized by causing the image processing apparatus to execute a determination process program. Then, such a determination processing program may be recorded on a recording medium (hard disk, memory card, etc.) in a readable state.

[4] Other Embodiments [4-1] Fourth Embodiment A fourth determination process executed by the character determination unit 204b will be specifically described. The fourth determination process is a process of determining whether the character described in the text document is a handwritten character based on the total number of characters recognized from the RGB image data.

  When the fourth determination process is executed by the character determination unit 204b, according to the flow shown in FIG. 15, the character determination processing unit 402, the recognized character counting unit 404, the fourth comparison unit 414, and the like in the character determination unit 204b. Each of the fourth determination units 415 executes a process. The processes executed by the character recognition processing unit 402 and the recognized character counting unit 404 are as described in the first embodiment.

<Fourth Comparison Unit>
The fourth comparison unit 414 determines the number of characters Cr based on the number of recognized characters Cr output from the recognized character counting unit 404 and the total number of extracted characters (number of extracted characters Cb) obtained by the layout analysis unit 402 c of the character recognition processing unit 402. Calculate the ratio Cr / Cb. Thereafter, the fourth comparison unit 414 determines whether the calculated ratio Cr / Cb is smaller than a threshold Th4.

<Fourth determination unit>
The fourth determination unit 415 determines, based on the determination result of the fourth comparison unit 414, whether the character described in the text document is a handwritten character or a printed character. Specifically, when the fourth comparison unit 414 determines that “the ratio Cr / Cb is smaller than the threshold Th4”, the fifth determination unit 415 determines that the character described in the character original is a handwritten character. It is determined that On the other hand, when the fourth comparison unit 414 determines that "the ratio Cr / Cb is not smaller than the threshold Th4", the fourth determination unit 415 states that "the characters described in the character original are print characters". judge.

  The recognition character number Cr tends to be significantly smaller in the manuscript of handwritten characters than in the manuscript of print characters. The reason is that depending on the person writing the character, the form of the character varies, and therefore the accuracy in recognizing the character decreases. Therefore, the ratio Cr / Cb tends to be significantly smaller in the manuscript of handwritten characters as compared to the manuscript of print characters. Then, by setting the threshold value Th4 to an appropriate value based on such a tendency, the determination based on the magnitude relationship between the ratio Cr / Cb and the threshold value Th4 is realized.

  According to the fourth determination process described above, in spite of the simple process, it is possible to determine with high accuracy whether the character described in the document is a handwritten character. The fourth determination process can be applied not only to Japanese but also to foreign languages.

[4-2] Fifth Embodiment Each of the first to fourth determination processes described above may be individually executed by the character determination unit 204b, and some or all of them may be combined to perform character determination. It may be executed by the unit 204b.

  As an example, when all of the first to fourth determination processes are combined and executed by the character determination unit 204b, the character determination unit 204b is configured as shown in FIG. The processes executed by the units other than the comparison unit 416 and the comprehensive determination unit 417 are as described in the first to fourth embodiments. The comparison unit 416 includes all of the first comparison unit 405, the second comparison unit 408, the third comparison unit 412, and the fourth comparison unit 414 described above.

  The comprehensive determination unit 417 determines whether the character described in the text document is a handwritten character by comprehensively considering the plurality of determination results in the comparison unit 416. Specifically, when the first to fourth determination processings show that the determination result that the characters described in the text document are handwritten characters in at least one item, the comprehensive determination unit 417 It outputs a comprehensive judgment result that "the characters described in the character original are handwritten characters". When the determination result that “the characters described in the text is a handwritten character” is obtained for two or three items or all the items, the comprehensive determination unit 417 “described in the text A comprehensive judgment result may be outputted that "the character is a handwritten character".

  According to this determination process, in spite of the simple process, it is determined with higher accuracy whether the character described in the document is a handwritten character. Note that this determination process is applicable not only to Japanese but also to foreign languages.

[4-3] Sixth Embodiment The document determination process and the character determination process executed by the automatic document determination unit 204 (the document determination unit 204a and the character determination unit 204b) of the image processing apparatus described above are the operations of the image forming apparatus. The present invention is applicable not only to the case where the copy mode is selected as the mode, but also to the case where another operation mode (image transmission mode, facsimile transmission / reception mode, etc.) requiring reading of an original is selected.

  Then, even in another operation mode, after the document is read by the image input device 1, the image processing device 2 executes image processing based on the RGB analog signal generated by the image input device 1. In the following, with reference to FIG. 17, the process executed by the image processing apparatus 2 when the image transmission mode is selected will be described.

<< Image processing device (processing when selecting image transmission mode) >>
In the image processing apparatus 2, an A / D conversion unit 201, a shading correction unit 202, an input processing unit 203, an automatic document determination unit 204, an expansion processing unit 205, an area separation processing unit 206, a color correction unit 207, a spatial filter unit 209, Each of the output tone correction unit 210 and the formatting processing unit 212 executes processing. The processing performed by each unit other than the color correction unit 207, the spatial filter unit 209, the output tone correction unit 210, and the formatting processing unit 212 is the same as the processing at the time of selecting the copy mode, and thus the description is omitted. .

<Color correction unit>
The color correction unit 207 is a display device in which the color space is generally spread from the RGB space for RGB image data (RGB image data after expansion processing when expansion processing is performed by the expansion processing unit 205) Convert to a color space (for example, sRGB color space) adapted to the display characteristics of The image data after conversion is input to the spatial filter unit 209 as it is without being subjected to the processing in the black generation / under color removal unit 208. Hereinafter, the image data generated by the conversion of the color space is referred to as "R'G'B 'image data".

<Space filter section>
The spatial filter unit 209 subjects the R′G′B ′ image data generated by the color correction unit 207 to processing such as enhancement processing and smoothing processing. At this time, determination results (original determination data and area separation data) in the original document automatic determination unit 204 and the area separation processing unit 206 are used as needed.

<Output tone correction unit>
The output gradation correction unit 210 performs output γ correction processing on the R′G′B ′ image data so that the image can have appropriate brightness. The processed R′G′B ′ image data is input to the formatting processing unit 212 as it is without being subjected to the processing by the halftone generation unit 211.

<Formatting Processing Unit>
The formatting processing unit 212 converts the file format of the R′G′B ′ image data processed by the output tone correction unit 210 into PDF (Portable Document Format).

<< Transceiver device >>
The transmitting and receiving device 5 transmits and receives image data via the network. Specifically, the transmitting and receiving device 5 has a function of receiving image data from an external connection device such as a personal computer and a function of transmitting image data to the external connection device. The image data received by the transmission / reception device 5 is stored, for example, in the storage unit 6.

  As an example of transmission, the transmitting / receiving device 5 sends the PDF file generated by the formatting processing unit 212 to the external connection device through the network or communication line in a state where the PDF file is attached to the e-mail by the mail processing unit (not shown). Send.

[4-4] Seventh Embodiment The document determination process and the character determination process performed by the automatic document determination unit 204 (the document determination unit 204a and the character determination unit 204b) of the image processing apparatus described above are not limited to the multifunction device. The present invention is also applicable to a color image reading apparatus such as a scanner.

  Specifically, as shown in FIG. 18, the color image reading apparatus includes an image input device 1, an image processing device 2, a storage unit 6, and a control unit 7. In the image processing apparatus 2, each of the A / D conversion unit 201, the shading correction unit 202, the input processing unit 203, the automatic document determination unit 204, the expansion processing unit 205, the color correction unit 207, and the formatting processing unit 212 Execute the process The details of the process executed by each unit are as described in the first embodiment and the sixth embodiment.

  The PDF file generated by the formatting processing unit 212 is output from the image processing apparatus 2. Then, the output image data is transmitted to the external connection device through the network or the communication line.

  The above description of the embodiments should be considered in all respects as illustrative and not restrictive. The scope of the present invention is indicated not by the embodiments described above but by the claims. Further, the scope of the present invention is intended to include all modifications within the scope and meaning equivalent to the claims.

204 automatic document determination unit 204a document determination unit 204b character determination unit 205 expansion processing unit 401 edge extraction unit 402 character recognition processing unit 402a signal conversion unit 402b binarization processing unit 402c layout analysis unit 402d character recognition unit 403 edge counting unit 404 recognition Character counting unit 405 First comparing unit 406 First judging unit 407 Continuous character counting unit 408 Second comparing unit 409 Second judging unit 410 Straight line detecting unit 411 Straight interval calculating unit 412 Third comparing unit 413 Third judging unit 414 Fourth Comparison unit 415 Fourth determination unit 416 Comparison unit 417 General determination unit Ca Number of consecutive characters Cb Extraction number of characters Ce Edge pixel number Cr Recognition number of characters Th1, Th2, Th4 threshold

Claims (8)

A character recognition processing unit that recognizes characters included in image data of a document;
A recognition character counting unit that calculates the total number of the recognized characters by counting the number of characters recognized by the character recognition processing unit;
A determination unit that determines whether the character described in the document is a handwritten character based on the total number calculated by the recognized character counting unit;
An image processing apparatus comprising: An edge extraction unit which extracts edge pixels constituting an edge of an image included in the image data;
An edge counting unit that calculates the total number of edge pixels by counting the number of edge pixels extracted by the edge extracting unit;
A first comparison unit that determines whether a first ratio, which is a ratio of the total number of edge pixels to the total number calculated by the recognized character counting unit, is greater than a first threshold;
And further
The determination unit determines that the character described in the document is the handwritten character when the first comparison unit determines that the first ratio is larger than the first threshold. The image processing apparatus according to claim 1. The character recognition processing in the recognized character in part, by the size counts the number of Rubun shaped to continuously be determined and more than a predetermined one and the same continuous character count for calculating the total number of consecutive characters Department,
A second comparator for determining whether the second rate Ru Oh the ratio of the total number of characters said consecutive with respect to the total calculated by said recognized character counting section is smaller than the second threshold value,
And further
The determination unit determines that the character described in the document is the handwritten character when the second comparison unit determines that the second ratio is smaller than the second threshold. The image processing apparatus according to claim 1. An edge extraction unit that extracts edge pixels constituting an edge of an image included in image data of a document;
A straight line detection unit that detects a straight line corresponding to an edge of an image forming a straight line included in the image data, based on the edge pixel extracted by the edge extraction unit;
A linear interval calculation unit that calculates an interval of the straight lines detected by the straight line detection unit;
A determination unit that determines whether the character described in the document is a handwritten character based on the interval calculated by the linear interval calculation unit;
An image processing apparatus comprising:   The linear interval calculation unit is configured to extract one of the straight lines detected by the straight line detection unit that has a length equal to or greater than a predetermined length, and calculates the interval for the extracted straight line. Image processing device.   The expansion processing unit according to any one of claims 1 to 5, further comprising: an expansion processing unit that performs expansion processing on the image data when the determination unit determines that the character described in the document is the handwritten character. The image processing device according to one. In the image processing device,
Character recognition processing for recognizing characters contained in image data of a document;
Recognition character counting processing for calculating the total number of the recognized characters by counting the number of characters recognized by the character recognition processing;
A determination process of determining whether the character described in the document is a handwritten character based on the total number calculated by the recognized character counting process;
An image processing program that executes In the image processing device,
Edge extraction processing for extracting edge pixels constituting an edge of an image included in image data of a document;
Straight line detection processing for detecting a straight line corresponding to an edge of an image constituting a straight line included in the image data, based on the edge pixel extracted by the edge extraction processing;
A straight line interval calculation process for calculating the interval of the straight lines detected by the straight line detection process;
A determination process of determining whether the character described in the document is a handwritten character based on the interval calculated by the linear interval calculation process;
An image processing program that executes

Images