JP4442651B2 - Image processing apparatus and program - Google Patents

Description

  The present invention relates to an image processing technique, and more particularly to an image processing technique for a scanned image.

  There is a technique for compressing a scanned image obtained by scanning a document (paper document or the like) with a scanner device at a high compression rate, and generating and storing a file having a relatively small image size. Examples of such a file include a PDF (Portable Document Format) file such as a highly compressed PDF.

  In such a file, a character area is distinguished from other areas, and processing according to the characteristics of each area is performed, so that character readability and file capacity reduction are realized. Specifically, with respect to the character area, in order to maintain the legibility of the character, data generated by performing binarization processing with high resolution is stored. On the other hand, for areas other than the character area, data that has been compressed at a high compression rate after being reduced in resolution without being binarized is stored in order to reduce the file capacity while maintaining gradation. The As for the character area, a predetermined range (word or line) is set as one unit, and one representative color is determined as the character color for each unit. According to this, the file size can be further reduced.

  In such a file, various problems occur when the characters become smaller. For example, when the character becomes small, it becomes difficult to extract the edge of the character. As a technique for solving such a problem, there is a technique described in Patent Document 1, for example.

JP 2002-10075 A

  By the way, regarding the above file, there is another problem, specifically, a problem regarding the color of characters.

  Specifically, when the character is thin, the color of the pixel near the edge of the character is likely to be mixed with the background color depending on scanning conditions and the like. For example, when the character is red and the background is green, the red color of the character in the scanned file (also referred to as a scan file) is not the original “red” but a color closer to the green (or yellow) side. . In other words, the character color of the scan file is influenced by the background color and becomes a color different from the original color on the original (paper document or the like). As a result, there is a problem that the legibility of the character is lowered.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a technique capable of improving the legibility of a character by eliminating the influence of a background color in a scanned image.

In order to solve the above problems, the invention of claim 1 is an image processing apparatus, which performs binarization processing on a scanned image and extracts a character portion corresponding to a character from a background. A character part extraction unit and a determination unit that determines a character color that is a representative color of the character part, and obtains a first color that is a candidate color of the character part based on a color component value of the character part; Determination to determine, as the character color, a second color obtained by correcting the first color so as to enlarge a difference between the first color and a background color that is the background color in the scanned image And a detecting means for detecting the thickness of the character, wherein the determining means changes the degree of enlargement of the difference at the time of correction according to the thickness of the character detected by the detecting means. as be one which in the character becomes narrower, the Characterized in that it is intended to increase the degree of expansion

According to a second aspect of the present invention, in the image processing apparatus according to the first aspect of the invention, the detecting means calculates a ratio of non-boundary pixels to the total number of pixels of the character portion as the thickness of the character. .

According to a third aspect of the invention, in the image processing apparatus according to the first or second aspect of the invention, the determining means determines an average color of at least some of the pixels included in the character portion as the first color. It is characterized by doing.

According to a fourth aspect of the invention, in the image processing apparatus according to the first or second aspect of the invention, the determining means determines the color of a representative pixel included in the character portion as the first color. And

The invention according to claim 5 is: a) a procedure for performing binarization processing on a scanned image and extracting a character part, which is a part corresponding to a character, from a background; b) a thickness of the character And c) obtaining a first color that is a candidate for the color of the character portion based on the color component value of the character portion, and using the first color and the background color in the scanned image A program for executing a procedure for determining a second color obtained by correcting the first color so as to enlarge a difference from a certain background color as a character color that is a representative color of the character portion. a is one of the steps c) is, c-1) the procedure b) in accordance with the thickness of said detected characters, a procedure for changing the degree of expansion of the difference in the correction time, the character The difference is increased so that the degree of enlargement of the difference increases as the thickness becomes thinner. The method includes a procedure for changing the degree of enlargement .
According to a sixth aspect of the invention, in the program according to the fifth aspect of the invention, the step b) includes a step of calculating a ratio of non-boundary pixels with respect to the total number of pixels of the character portion as the thickness of the character. And
According to a seventh aspect of the invention, in the program according to the fifth or sixth aspect of the invention, the procedure c) determines an average color of at least some of the pixels included in the character portion as the first color. It includes a procedure.
According to an eighth aspect of the present invention, in the program according to the fifth or sixth aspect , the step c) includes a step of determining a color of a representative pixel included in the character portion as the first color. It is characterized by.

According to the first to eighth aspects of the present invention, the first color that is a candidate for the color of the character portion is obtained based on the color component value of the character portion, and the background in the first color and the scanned image is obtained. Since the second color obtained by correcting the first color so as to enlarge the difference from the background color which is the color of the character is determined as the character color, the influence of the background color in the scanned image is eliminated and the character color The legibility can be improved. In particular, the degree of enlargement of the difference between the first color and the background color is changed according to the detected thickness of the character , and more specifically, the degree of enlargement of the difference increases as the character becomes thinner. since the, it is possible to achieve an appropriate degree of correction in accordance with the thickness of a character.

<1. First Embodiment>
<1-1. Configuration>
FIG. 1 is a schematic view showing an original digitizing system 100 (100A) according to the first embodiment. Here, the electronic document system 100 includes a multi-function peripheral (abbreviated as MFP) 1 (1A). The MFP 1 also functions as an image processing apparatus by executing predetermined image processing using the image processing unit 3.

  The document digitizing system 100 is a system that converts an image (also referred to as a document image) obtained by optically reading a document (such as a paper document) into electronic data. Specifically, when the MFP 1 generates a document image and executes image processing on the document image, the document is converted into electronic data.

  The MFP 1 is a device (also referred to as a multi function device) having a scanner function, a printer function, a copy function, a facsimile function, and the like. Specifically, the MFP 1 includes an image reading unit 2, an image processing unit 3, a print output unit 4, a communication unit 5, an input / output unit 6, a storage unit 8, and a controller 9. The above functions are realized by operating these units in a complex manner.

  The image reading unit 2 is a processing unit that optically reads a document placed at a predetermined position of the MFP 1 and generates an image (also referred to as a document image) of the document. The image reading unit 2 is also referred to as a scanner unit.

  The image processing unit 3 is a processing unit that performs various types of image processing on the scanned image generated by the image reading unit 2. The image processing unit 3 includes a character part extraction unit 31 and a character color determination unit 32. The character part extraction unit 31 performs a process of separating and extracting characters included in the scanned image from the background. Further, the character color determination unit 32 executes a process for determining the color (representative color) of the character part extracted by the character part extraction unit 31. These processes will be described in detail later.

  The print output unit 4 is an output unit that prints an image on various media such as paper based on image data related to the target image.

  The communication unit 5 is a processing unit capable of performing facsimile communication via a public line or the like. The communication unit 5 can perform network communication via the communication network NW. By using this network communication, the MFP 1 can exchange various data with a desired destination. The MFP 1 can also send and receive e-mails using this network communication.

  The input / output unit 6 includes an operation input unit 61 that receives input to the MFP 1 and a display unit 62 that displays and outputs various types of information.

  The storage unit 8 includes a storage device such as a hard disk drive (HDD). The storage unit 8 stores a document image or the like generated by the image reading unit 2 or the like.

  The controller 9 is a control device that controls the MFP 1 in an integrated manner, and includes a CPU and various semiconductor memories (such as a RAM and a ROM). Various functions of the MFP 1 are realized by operating various processing units under the control of the controller 9. For example, by scanning a desired image optically using the image reading unit 2 under the control of the controller 9, an image obtained by scanning a document (also referred to as a document image or a scanned image) is acquired, and a scanner function is realized. The

<1-2. Overview of image processing>
The MFP 1 has a function of performing predetermined image processing on a scanned image to generate a file in a predetermined format (for example, a PDF (Portable Document Format) file). Hereinafter, such a file generation process will be described. This file generation function is realized by the image processing unit 3 or the like.

  As shown in FIG. 2, first, a predetermined pre-process is performed on the scanned image acquired by the image reading unit 2 (step S11). This preprocessing can include, for example, image format conversion processing, resolution adjustment processing, and the like.

  Next, an area determination process for distinguishing between a character area (also referred to as a first area) that is an area related to characters and another area (also referred to as a second area) is executed (step S13). Thereafter, by performing processing according to the characteristics of each area, it is possible to reduce the file capacity while maintaining the legibility of the characters. In addition, as a 2nd area | region, the area | region containing a photograph or a chart is illustrated.

  With respect to the image of the character area (first area), the process for the character area (step S15) and the lossless compression process (step S16) are performed. With respect to the character region, in order to maintain the legibility of the character, binarization processing or the like is performed with high resolution, and data DT is generated. The data DT is stored in a reversible compression format in a predetermined format (MMR (Modified Modified READ) or the like). According to the reversible compression, it is possible to avoid deterioration of the legibility of characters.

  On the other hand, regarding the image of the area other than the character area (second area), the data is subjected to the resolution reduction process without being subjected to the binarization process in order to reduce the file capacity while maintaining the gradation. A DG is generated (steps S17 and S18). The data DG is stored in a predetermined format (such as JPEG) after being irreversibly compressed. By irreversible compression, the file size of the generated file can be made very small.

  Then, a file in a predetermined format is generated by combining both data DT and DG (step S19).

<1-3. Character data generation processing>
Next, the image processing (step S15) for the character area will be described in more detail.

  FIG. 3 is a diagram showing an overview of the processing in step S15. Moreover, FIGS. 4-11 is a figure which shows the transition of the processing result regarding the red character of a white background. Specifically, FIGS. 4 to 7 are diagrams showing the transition of the processing result related to the character “I” having a size of 6 points, and FIGS. 8 to 11 are characters having a size of 10.5 points. It is a figure which shows the transition of the processing result regarding "I". 4 and 8 are diagrams showing the letter “I” on the document, and FIGS. 5 and 9 are diagrams showing the letter “I” in the scanned image. For example, when the MFP 1 scans an original to generate a scanned image as described above, the 6-point character “I” (see FIG. 4) on the original is obtained in the state shown in FIG. 5 in the scanned image. Similarly, the 10.5 point character “I” (see FIG. 8) on the document is obtained in the scan image as shown in FIG. 5 and 9, each pixel has a multi-tone value. 6 and 10 are diagrams showing a character “I” in an image obtained by binarizing the scanned image, specifically, an image before a correction process (step S24) described later, and FIG. 7 and FIG. 11 is a diagram showing the character “I” in the image after the correction process (step S24).

  Here, it is assumed that the MFP 1 acquires a color image as a scan image. Specifically, a color image (also referred to as an RGB image) composed of three primary color components of an R (red) component, a G (green) component, and a B (blue) component is acquired as a scan image. To do. Each color component image (that is, an R component image, a G component image, and a B component image) has gradation values of multiple levels (for example, 256 levels) corresponding to predetermined bits (for example, 8 bits). Therefore, the RGB image is configured as data having a bit number (for example, 24 bits) that is three times the predetermined bit number of each color component.

  In this character data generation process, in addition to RGB image data (also simply referred to as RGB data), “brightness data” is also generated based on RGB data. The brightness data is obtained by calculating the pixel value of each pixel as a value obtained by adding a value obtained by multiplying each color component value of RGB data by a corresponding coefficient. Further, the brightness data is acquired as a value having a gradation value corresponding to a predetermined number of bits (for example, 8 bits). The brightness data is also expressed as grayscale image data. In the following character extraction processing and the like, the calculation load can be reduced by basically using the brightness data.

  In step S21 (FIG. 3), binarization processing is performed on the brightness data of the character area. By this binarization processing, a portion (also referred to as “character portion”) corresponding to “character” in the character region of the scanned image is extracted separately from the background.

  For example, the 6-point character “I” on the scanned image in FIG. 5 is obtained as a pixel set indicating the character “I” as shown in FIG. 6 by the binarization processing in step S21. FIG. 12 is a diagram schematically representing this state. In FIG. 12, a small rectangle represents one pixel, and the state where the character “I” is configured as a set of 62 pixels is shown. In other words, the character “I” is constituted by the character portion CP which is an aggregate of all 62 pixels.

  Similarly, the 10.5 point character “I” on the scanned image of FIG. 9 is obtained as a pixel set indicating the character “I” as shown in FIG. 10 by the binarization processing in step S21. . FIG. 13 is a diagram schematically representing this state. In FIG. 13, a small rectangle represents one pixel, and the state where the character “I” is configured as a set of 196 pixels is shown. In other words, the character “I” is constituted by the character portion CP which is an aggregate of all 196 pixels.

  Here, the color of the character part CP is a predetermined range (one character, word, line, paragraph, block, etc.) including one or more characters as one unit, and one representative color for each unit is “ It is determined as “character color”. According to this, compared with the case where a different color is determined for each pixel constituting the character part CP, the file size can be further reduced. In the following, for simplification of description, the case where the character color is determined in units of one character will be exemplified.

  Therefore, the process of step S22 (FIG. 3) is executed as the first stage process. Specifically, the average color of all the pixels included in the character part CP is obtained as the first color V1. In other words, the first color V1 is obtained as a candidate for the representative color (“character color”) of the character part CP. For example, for all 62 pixels of the 6-point character “I” (see FIGS. 6 to 12), an average value is obtained for each color component, and a color constituted by the average value of each color component is obtained as the “first color”. be able to.

  However, the first color V1 may be different from the character color in the document. In particular, thin characters are often closer to the background color than the original color due to the influence of the background color. For example, as shown in FIG. 14, the “first color” related to a six-point character that is originally red (R = 255, G = 0, B = 0) is white (R = 255, G = 255, B = 255), it becomes a slightly dull red color (R = 246, G = 79, B = 81). FIG. 14 is a diagram showing a change in color before and after correction.

  Therefore, in this embodiment, in step S24 (FIG. 3), the first color V1 obtained by correcting the first color V1 so as to increase the difference between the first color V1 and the background color (background color) VB. The second color V2 is determined as the “character color”.

  Here, the degree of enlargement of the difference between the first color V1 and the background color VB at the time of correction is changed according to the thickness of the character. This is based on the knowledge that the degree of color mixing of characters varies depending on the thickness of the characters. Therefore, after obtaining the thickness of the character in step S23, the character color determining operation in step S24 is executed.

  First, the process of step S23 will be described.

  In step S23, the thickness of the character is detected.

  Specifically, first, in addition to obtaining the number of all pixels constituting the character part CP, the internal pixels (non-pixels) excluding the pixels (boundary pixels) on the boundary with the background (background) among all the pixels. The number of boundary pixels). In the case of FIG. 12, it is calculated that the total number of pixels is 62 pixels and the number of internal pixels is 16 pixels.

  Then, the ratio (internal ratio) α of internal pixels (non-boundary pixels) to all pixels is calculated. In the case of FIG. 12, the internal ratio α is calculated as 26% (= (16/62) × 100).

  In the case of FIG. 13, the total number of pixels is 196 pixels, the number of internal pixels is 102 pixels, and the internal ratio α is calculated as 52% (= (102/196) × 100).

  The ratio of the internal pixel (non-boundary pixel) to the internal ratio α, in other words, the total number of pixels of the character part CP increases as the character becomes thicker. Using this characteristic, in this embodiment, the internal ratio α is used as an index value representing “character thickness”.

  Next, the process of step S24 will be described.

  In step S24, the component values (V1r, V1g, V1b) of the first color V1 using the component values (VBr, VBg, VBb) of the background color VB and the like according to the following equation (1). Are corrected, and the component values (V2r, V2g, V2b) of the second color V2 are calculated. Note that the value β in the equation (1) is an appropriate constant, and is determined as 0.75 (= 75%), for example. The background color VB can be obtained by obtaining the average color of the background portion (background portion) determined as a portion other than the character portion CP by the binarization process (step S21).

  However, each calculated value (V2r, V2g, V2b) according to the equation (1) is corrected to the minimum value Vmin when it is smaller than the minimum value Vmin (for example, 0 (zero)), and the maximum value Vmax ( For example, when it is larger than 255), it is corrected to the maximum value Vmax.

  The correction principle based on Expression (1) will be described with reference to FIG. The symbol V1 is also used when collectively referring to the RGB color component values. The same applies to the symbols V2 and VB.

  For example, when the internal ratio α is equal to β (for example, 75%), k = 1. At this time, the component value V2 of the second color is equal to the component value V1 of the first color according to the expression (1). Referring to FIG. 15 in particular, when the difference (VB−V1) between the background color component value VB and the first color component value V1 is subtracted from the background color component value VB, as in Equation (1), It can be seen that the value is V1.

  On the other hand, assuming that the internal ratio α is smaller than β (for example, 75%), k> 1. For example, when α = 26% and β = 75%, k = 1.49 (149%). In this case, as shown in FIG. 15, a value obtained by expanding the difference between the background color component value VB and the first color component value V1 from the background color component value VB to k times (1.49 times) ( A value obtained by subtracting (VB−V1) × k) is calculated as a value V2. In other words, the component values (V2r, V2g, V2b) of the second color are the corresponding component values (V1r, V1g, V1b) of the first color and the corresponding component values (VBr, VBg, V1b) of the background color. VBb) is determined so as to enlarge the difference (specifically, the absolute value of the difference value).

  For example, when the character is thin (in the case of a 6-point character), as shown in FIG. 14, the color V1 (R = 246, G = 79, B = 81) is the color V2 (R = 242, G = 0, B = 0).

  Specifically, the R component V2r is calculated as V2r = 255− (255−246) × 1.49 = 241.6. The G component V2g is calculated as V2g = 255− (255−79) × 1.49 = −7.2, and the B component V2b is calculated as V2g = 255− (255−81) × 1.49 = −. Calculated as 4.3. After that, the numbers after the decimal point are rounded to an integer, and the negative numbers are corrected to zero to calculate the above values (V2r = 242, V2g = 0, V2b = 0).

  According to this, even when the original color V0 is calculated as the color V1 approaching the background color due to the influence of the background color in step S22, the color V1 can be corrected to the color V2 separated from the background color. it can. In this case, in particular, the G component and B component (value 0) of the color V2 after correction are compared with the G component and B component (value) of the background color compared to the color V1 (value 79 or value 81) before correction, respectively. 255). Therefore, the corrected color V2 is a color that is separated from the background color as a whole. By expressing a character with such a color (that is, a color in which the difference from the background color is widened), the color of the character is clear and the character is easy to see. That is, the legibility of characters can be improved.

  The value k is changed according to the index value α representing the thickness of the character. Specifically, the value k decreases as the character becomes thicker (the index value α increases). Conversely, the value k increases as the character becomes thinner (the index value α decreases).

  Here, as shown in FIG. 14, the color V1 (R = 233, G = 20, B = 16) before correction when the character is thick (in the case of a 10.5 point character) is when the character is thin (6 Compared to the original color V1 (R = 246, G = 79, B = 81) before correction in the case of a point character, it is relatively close to the original color (R = 255, G = 0, B = 0).

  Therefore, in this embodiment, when the character is thick, the degree of correction is made relatively small by using the equation (1) in which the value k becomes relatively small. According to this, excessive correction can be prevented. On the contrary, when the character is thin, the value k is set to a relatively large value by the equation (1), and the degree of correction is made relatively large.

  As described above, the degree of correction can be optimized by changing the degree of enlargement of the difference between the first color and the background color at the time of correction according to the thickness of the character.

<1-4. When the background color is green>
FIGS. 16 to 23 are diagrams showing changes in processing results when the background color is not white but green (although the character color is red as described above), and FIGS. It corresponds. 16 to 19 are diagrams showing the transition of the processing result related to the character “I” having a size of 6 points, and FIGS. 20 to 23 are related to the character “I” having a size of 10.5 points. It is a figure which shows the transition of a process result. 16 and 20 are diagrams showing the letter “I” on the document, and FIGS. 17 and 21 are diagrams showing the letter “I” in the scanned image. 18 and 22 are diagrams showing the character “I” in the image obtained by binarizing the scan image, specifically, the image before the correction process (step S24). FIG. 19 and FIG. FIG. 10 is a diagram showing the letter “I” in the image after the correction process (step S24).

  FIG. 24 is a diagram showing the color before and after correction when the background color is not white but green.

  As shown in FIG. 24, for example, for a 6-point character, the character color V1 before correction (R = 195, G = 23, B = 1) is the character color V2 after correction (R = 220, G = 0, B). = 0). That is, the character color V1 before correction is more dull than the original red color (R = 255, G = 0, B = 0) due to the influence of the background green color (R = 145, G = 201, B = 52). Even when the color (R = 195, G = 23, B = 1), the corrected character color V2 is a color (R = 220, G = 0, B = 0). Therefore, the observer can visually recognize characters with relatively clear colors. Therefore, the legibility of characters can be improved.

  For the 10.5 point character, as shown in FIG. 16, the uncorrected color V1 (R = 211, G = 16, B = 1) is the corrected color V2 (R = 226, G = 0, B = 0). According to such correction, the color V1 of the character before correction is the original red color (R = 255, G = 0, B = 0) due to the influence of the green background (R = 145, G = 201, B = 52). ), The corrected character color V2 is corrected to a color with a larger difference from the background.

  In FIG. 17 to FIG. 19 and FIG. 21 to FIG. 23, the background portion (or the ground portion) adjacent to the character portion CP is treated as the above-described second region in the same manner as the photograph or chart, and the character portion CP. And is shown in a synthesized state. Therefore, the background portion is not binarized like the character portion CP, and is shown as having multi-level gradation values. Further, the green background color (R = 145, G = 201, B = 52) in the above calculation is acquired as the average color of such a background portion.

<1-5. When the background color is blue>
FIG. 25 is a diagram illustrating the color before and after correction when the background color is blue instead of green (although the character color is red as described above). Furthermore, FIGS. 26 to 33 are diagrams showing the transition of the processing results when the background color is blue instead of white, and correspond to FIGS. 4 to 11 respectively. 26 to 29 are diagrams showing the transition of the processing result related to the letter “I” having a size of 6 points, and FIGS. 30 to 33 are related to the letter “I” having a size of 10.5 points. It is a figure which shows the transition of a process result. 26 and 30 are diagrams showing the letter “I” on the document, and FIGS. 27 and 31 are diagrams showing the letter “I” in the scanned image. FIGS. 28 and 32 are diagrams showing the character “I” in an image obtained by binarizing the scanned image, specifically, the image before the correction process (step S24). FIGS. FIG. 10 is a diagram showing the letter “I” in the image after the correction process (step S24).

  As shown in FIG. 25, for a six-point character, for example, the color V1 before correction (R = 192, G = 80, B = 74) is the color V2 after correction (R = 255, G = 77, B = 31). ). According to such correction, the character color V1 before correction is changed to the original red color (R = 255, G = 0, B = 0) due to the influence of the blue color of the background (R = 26, G = 86, B = 161). ), The corrected character color V2 is a color (R) that has a larger difference from the background (R = 192, G = 80, B = 74). = 255, G = 77, B = 31). Therefore, the observer can visually recognize characters with relatively clear colors. Therefore, the legibility of characters can be improved.

  For the 10.5 point character, as shown in FIG. 25, the uncorrected color V1 (R = 225, G = 63, B = 59) is the corrected color V2 (R = 255, G = 58, B = 36). According to such correction, the character color V1 before correction is changed to the original red color (R = 255, G = 0, B = 0) due to the influence of the blue color of the background (R = 26, G = 86, B = 161). ), The corrected character color V2 is a color (R) having a larger difference from the background (R = 225, G = 63, B = 59). = 255, G = 58, B = 36).

  27 to 29 and FIGS. 31 to 33, as in FIGS. 17 to 19 and FIGS. 21 to 23, the background portion is not binarized like the character portion CP. It is shown as having graded gradation values. Further, the background color blue (R = 26, G = 86, B = 161) in the above calculation is obtained as the average color of such a background portion.

<2. Second Embodiment>
The second embodiment is a modification of the first embodiment. Below, it demonstrates centering on difference with 1st Embodiment.

  In the first embodiment, the case where the MFP 1 functions as an image processing apparatus is illustrated, but in the second embodiment, the case where a computer (such as a personal computer) functions as an image processing apparatus is illustrated.

  FIG. 34 is a schematic diagram showing an original digitizing system 100 (100B) according to the second embodiment. Here, the document digitizing system 100 includes a plurality of scanner devices 80 and a computer 90 such as a personal computer. The scanner device 80 may be a scanner-dedicated device, or may be a multifunction device such as the MFP 1 described above.

  The computer 90 is connected to a plurality of scanner devices 80 via a network NW so as to be communicable. The computer 90 receives the scan image generated by each scanner device 80 via the network NW, and performs the above-described image processing on the scan image.

  Specifically, the computer 90 reads the program PG from a recording medium 91 (for example, a flexible disk, a CD-ROM, a DVD-ROM, etc.) on which a predetermined program PG is recorded, and uses the program PG using the CPU or the like. By executing these functions, the same function as that of the image processing unit 3 is realized. The program PG may be supplied by a recording medium, or may be supplied by downloading via the Internet.

  As a result, the computer 90 can realize image processing similar to that of the first embodiment, and can create a file in which the character color is clear.

  Further, the computer 90 can create the above-described file (for example, PDF file) based on the images transmitted from the plurality of scanner devices 80. In particular, the computer 90 sets the parameter β in the above equation (1) to a value unique to each scanner device and stores it in the storage unit. For example, β = 0.75 is set for a certain scanner device, and β = 0.85 is set for another scanner device. According to this, even if the characteristics of the scanner devices 80 are different from each other, it is possible to perform processing according to the characteristics.

<3. Modification>
Although the embodiments of the present invention have been described above, the present invention is not limited to the contents described above.

  For example, in each of the above-described embodiments, the case where the value k is changed according to the thickness (α) of the character is exemplified, but the present invention is not limited to this, and the value k may be a fixed value.

  Moreover, although the case where the average color of all the pixels contained in the character part CP (refer FIG. 12 etc.) is calculated | required as 1st color V1 is illustrated in the said embodiment, it is not limited to this.

  For example, an average color of a part of all the pixels included in the character part CP (for example, all or part of the non-boundary pixels) may be obtained as the first color V1.

  Further, the color of the representative pixel included in the character part CP may be determined as the first color V1. More specifically, the color of the representative pixel R1 at the center of the character portion CP in FIG. 12 may be determined as the first color V1.

  Further, in each of the above embodiments, the case where the character color is determined in units of one character has been exemplified. However, as described above, a certain range (word, line, paragraph, block, or the like) is set as one unit. The “character color” may be determined for each unit.

  For example, one character color may be determined for each word. Specifically, the “character color” of a plurality of characters included in a word may be obtained, and the average value may be determined as the “character color” of the word.

  Alternatively, one character color may be determined for each “line”. Specifically, the “character color” of a plurality of characters included in the “line” is obtained, and the average value is determined as the “character color” of the “line”.

  Alternatively, one character color may be determined for each character block composed of a plurality of lines (or a plurality of paragraphs). In this case, the character color is determined for each of a plurality of lines, and among the neighboring lines, the character colors that are close to each other are combined to form a character block, and the character color for the block is determined again. preferable. According to this, since a plurality of rows whose colors are similar to each other can be combined into one block to manage “color”, it is possible to efficiently reduce the file size.

1 is a schematic diagram illustrating a document digitization system according to a first embodiment. FIG. 2 is a diagram illustrating an overview of image processing in an MFP. It is a figure which shows the outline | summary of the process regarding a character. It is a figure which shows the character "I" (6 point character) on a manuscript. It is a figure which shows the character "I" (6 point character) in a scan image. It is a figure which shows the character "I" (6-point character) before a correction process. It is a figure which shows the character "I" (6 point character) after a correction process. It is a figure which shows the character "I" (10.5 point character) on a manuscript. It is a figure which shows the character "I" (10.5 point character) in a scan image. It is a figure which shows the character "I" (10.5 point character) before a correction process. It is a figure which shows the character "I" (10.5 point character) after a correction process. It is a figure which shows typically the character "I" (6-point character) before a correction process. It is a figure which shows typically character "I" (10.5 point character) before a correction process. It is a figure which shows the change of the color before and behind correction | amendment (when a background is white). It is a figure for demonstrating the correction principle. It is a figure which shows the 6 point character (green background) on a manuscript. It is a figure which shows the 6 point character (green background) in a scan image. It is a figure which shows the 6 point character (green background) before a correction process. It is a figure which shows 6 point characters (green background) after a correction process. It is a figure which shows the character 10.5 point character (green background) on a manuscript. It is a figure which shows the 10.5 point character (green background) in a scanning image. It is a figure which shows the 10.5 point character (green background) before a correction process. It is a figure which shows the 10.5 point character (green background) after a correction process. It is a figure which shows the change of the color before and behind correction | amendment (when a background is green). It is a figure which shows the change of the color before and behind correction | amendment (when a background is blue). It is a figure which shows the 6 point character (blue background) on a manuscript. It is a figure which shows the 6 point character (blue background) in a scan image. It is a figure which shows 6 point characters (blue background) before a correction process. It is a figure which shows 6 point characters (blue background) after a correction process. It is a figure which shows the character 10.5 point character (blue background) on a manuscript. It is a figure which shows the 10.5 point character (blue background) in a scanning image. It is a figure which shows the 10.5 point character (blue background) before a correction process. It is a figure which shows the 10.5 point character (blue background) after a correction process. It is the schematic which shows the original digitization system which concerns on 2nd Embodiment.

Explanation of symbols

1 MFP (image processing device)
90 Computer (image processing device)
91 Recording Medium 100, 100A, 100B Electronic Document System PG Program R1 Representative Pixel V1 First Color V2 Second Color VB Background Color α Internal Ratio

Claims (8)

An image processing apparatus,
Character part extraction means for performing binarization processing on the scanned image and separating and extracting a character part that is a part corresponding to the character from the background;
Determining means for determining a character color that is a representative color of the character portion, wherein a first color that is a candidate color of the character portion is obtained based on a color component value of the character portion; Determining means for determining, as the character color, a second color obtained by correcting the first color so as to enlarge a difference between a background color that is a background color of the scanned image and the background color;
Detecting means for detecting the thickness of the character;
With
Said determining means, depending on the thickness of the character detected by the detecting means, be one that changes the degree of expansion of the difference in the correction time, as the character becomes narrower, the expansion of the difference An image processing apparatus characterized by increasing the degree .   The image processing apparatus according to claim 1.
  The image processing apparatus according to claim 1, wherein the detection unit calculates a ratio of a non-boundary pixel to a total number of pixels of the character portion as a thickness of the character.   The image processing apparatus according to claim 1 or 2,
  The image processing apparatus according to claim 1, wherein the determining unit determines an average color of at least some of the pixels included in the character portion as the first color.   The image processing apparatus according to claim 1 or 2,
  The image processing apparatus according to claim 1, wherein the determining unit determines a color of a representative pixel included in the character portion as the first color.   On the computer,
  a) A procedure for performing binarization processing on the scanned image and separating and extracting a character portion that is a portion corresponding to the character from the background;
  b) a procedure for detecting the thickness of the character;
  c) obtaining a first color that is a candidate for the color of the character part based on the color component value of the character part, and calculating a difference between the first color and a background color that is the background color of the scanned image. Determining a second color obtained by correcting the first color so as to enlarge as a character color that is a representative color of the character portion;
A program for executing
  The procedure c)
    c-1) A procedure for changing the degree of enlargement of the difference at the time of correction according to the thickness of the character detected in step b), and the degree of enlargement of the difference as the character becomes thinner A program comprising a procedure for changing the degree of enlargement of the difference so as to increase.   The program according to claim 5, wherein
  The procedure b) includes a procedure for calculating a ratio of non-boundary pixels to the total number of pixels in the character portion as the thickness of the character.   In the program according to claim 5 or 6,
  The procedure c) includes a procedure for determining an average color of at least some pixels included in the character portion as the first color.   In the program according to claim 5 or 6,
  The procedure c) includes a procedure for determining a color of a representative pixel included in the character portion as the first color.