JP4952507B2 - Image processing apparatus, image processing method, and image processing program - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program.

  Conventionally, various methods have been disclosed for extracting only pixels belonging to a specified ink color from an input image.

  First, as shown in FIG. 17, Non-Patent Document 1 discloses a phenomenon in which, in an image obtained by scanning a document, the color distribution in the RGB space by each pixel constituting the character can be regarded as linear. . This phenomenon will be further described. As shown in FIG. 17, when scanning a line figure pattern such as a character or a stamp, a gradation-like mixed color is generated in the outline portion depending on the pattern color and the background color. As a result, when the line figure pattern such as characters and stamps is uniform, the color distribution of each pixel is ideally one point, whereas the color mixture described above causes extreme dispersion in one direction. A large linear distribution. In addition, FIG. 17 is a figure for demonstrating a prior art.

  By utilizing the phenomenon disclosed in Non-Patent Document 1 described above, as a technique for extracting only pixels belonging to a specified ink color from an input image, a color distribution of a point on a certain color space belongs to a conventional technique. When determining whether or not the distance between the center of each color distribution (color center) and the point to be determined is determined, it is determined that the point to be determined belongs to the color distribution with the minimum distance. ing. For example, in this conventional method, the image processing apparatus calculates the distance from the center of the color distribution for each pixel. Then, the image processing apparatus determines that the calculated distance belongs to the color related to the closest color distribution, and extracts each pixel constituting the color distribution related to the designated ink color.

  In Patent Document 1, the color distribution of each pixel constituting the line figure pattern is approximated to be a line segment connecting the center of the background color distribution and the center of the color distribution of each pixel of the line figure, and the image processing apparatus A technique for extracting pixels included in a cylinder or ellipsoid surrounding the line segment as pixels included in the pattern color to be extracted is disclosed.

  In other words, in Patent Document 1, paying attention to the above-described characteristic distribution shape (a linear distribution with extremely large dispersion in one direction), the image processing apparatus has a color distribution of each pixel of the line figure pattern. An approximate curve is created by approximating that it is a straight line, and pixels included in a region near the created approximate straight line (in the example, a cylindrical region) are determined to be the same color pattern and extracted.

Worring and Todoran, Proc. of ICDAR (1999) JP 2001-291097 A (page 5-7, FIG. 1)

  By the way, as described below, the above-described conventional technique has a problem in that it cannot accurately extract a line figure pattern of a specified ink color.

  That is, since the color center closest to each pixel is not necessarily the color center to which each pixel belongs, in the above-described conventional method, the image processing apparatus uses pixels that belong to a color different from the color to which the pixel originally belongs. The line pattern pattern of the specified ink color cannot be extracted with high accuracy.

  Specifically, as shown in FIG. 17, when the distribution shape is a line that is separated from a sphere, a situation occurs in which the centers of adjacent different color distributions are closer to the center of the color distribution to which the user originally belongs. It becomes easy. Here, in the conventional method described above, when the image processing apparatus extracts a line figure pattern of a specified ink color in a form composed of a plurality of colors, the image processing apparatus uses a color distribution corresponding to the specified ink color. Also, a pixel that is close to a color distribution corresponding to another adjacent color is erroneously classified as the other color. As a result, the image processing apparatus drops pixels that are mistakenly classified into the other colors without being extracted as the designated ink color, causing a defect in the extraction result.

  For example, description will be made using the example shown in FIG. In FIG. 17, there are color distributions 1 and 2 of each pixel of a pattern having two different colors in the RGB space, and the color centers of the color distribution are C1 and C2, respectively. The pixel P1 is a pixel that originally belongs to the color distribution 1. In such a situation, in the conventional general technique, the image processing apparatus determines which distribution belongs to the proximity of the distance from the center. The image processing apparatus compares the distance (d1) from P1 to C1 with the distance (d2) from P1 to C2, and determines that the distance (d2) from P1 to C2 is smaller. , P1 is classified as a pixel included in the color distribution 2 corresponding to C2. That is, the image processing apparatus erroneously determines that the pixel P1 originally belonging to the color distribution 1 is a pixel belonging to the color distribution 2.

  Further, in the above-described Patent Document 1, a line figure pattern of a specified ink color cannot be extracted with high accuracy due to ink unevenness or a color mixture with a background color.

  For example, as shown in FIG. 18, in the case of a post-printed line figure pattern, a mixed color due to overpainting depending on the background color and the ink color due to the ink being not completely opaque or unevenness in the thickness of the ink. Happens. In the example shown in FIG. 18, the blue portion (such as (1) in FIG. 18) that is the color of the ink and the yellow portion (such as (2) in FIG. 18) that is the background color are mixed, resulting in yellow and blue. Are mixed colors ((3) in FIG. 18). Due to such color mixing, the color distribution generally deviates greatly from a linear shape. In addition, FIG. 18 is a figure for demonstrating a prior art.

  As a result, in the technique disclosed in Patent Document 1 described above, the image processing apparatus causes a defect in the extracted line figure pattern as shown in FIG. In other words, the technique disclosed in Patent Document 1 described above can produce a line figure pattern with almost achromatic black ink even in a print showing a color distribution as shown in FIG. 19 (1) or a post-printed pattern. Applicable. For this reason, the technique disclosed in Patent Document 1 described above is applied to a line figure pattern post-printed with a chromatic ink color in which the shape of the color distribution generally deviates from a straight line, as shown in FIG. In this case, the image processing apparatus causes an extraction omission to occur in a portion that protrudes from a straight line vicinity region (in the example, a cylindrical region), and cannot accurately extract a line figure pattern of a specified ink color.

  Accordingly, the present invention has been made to solve the above-described problems of the prior art, and an image processing apparatus, an image processing method, and an image processing apparatus that can extract a line figure pattern of a specified ink color with high accuracy. An object is to provide an image processing program.

  In order to solve the above-described problems and achieve the object, this image processing apparatus is designated as an extraction target from an image in which a pattern of ink of a predetermined color is post-printed on a background of a predetermined background color. An image processing apparatus that extracts pixels corresponding to a pattern of ink, and an index used for drawing a color distribution approximation curve of the ink on a color space for each ink that can be specified as the extraction target, and a color space The color distribution defined by associating the unique point coordinates indicating the unique coordinates of the ink not affected by the background color and the separation value used for separating the background color and the ink color on the color distribution approximation curve. A storage means for storing an approximate curve table; a background color coordinate determination means for determining a background color coordinate indicating a color space coordinate of the background color in the image; The index and the unique point coordinates corresponding to the link are read from the color distribution approximate curve table, and the curve drawn by the index based on the unique color coordinates and the background color coordinates determined by the background color coordinate determining means is Color distribution approximate curve determining means for determining as a color distribution approximate curve relating to ink, and the separation value corresponding to the ink designated as the extraction target is read from the color distribution approximate curve table, and by the color distribution approximate curve determining means Separation point determination means for determining a separation point between a designated ink pixel and a background pixel corresponding to the separation value on the determined color distribution approximation curve, and each pixel of the image developed in a color space From the above, the intersection of the color distribution approximate curve determined by the color distribution approximate curve determination means and the intersection is determined for each pixel. After that, a pixel in which the intersection point is in the direction of the unique point coordinates as viewed from the separation point determined by the separation point determination unit, and the distance in the color space with the intersection point is within a predetermined threshold value, And an extraction unit that extracts pixels corresponding to a pattern of ink designated as an extraction target.

  Further, in addition to the above-described requirements, the image processing apparatus performs post-printing with each of the two learning images having different background colors on which the pattern is post-printed only with the same ink that can be post-printed. First determining means for calculating each of the pattern color coordinates indicating the coordinates in the color space of each of the pixels constituting the pattern, and determining the average value or mode value of the calculated pattern color coordinates as the pattern color center coordinates The background color coordinate determination means determines the background color coordinates for each of the two images, and the pattern color center determined by the first determination means for each of the two images. Using the coordinates and the background color coordinates determined by the background color coordinate determining means, the color component having the unique point coordinates as an unknown number. A second determining unit that determines each approximate curve, and the eigenpoint coordinates that are unknown in the two color distribution approximate curves determined by the second determining unit are the same. Third determining means for calculating and determining point coordinates; and fourth determining means for substituting the eigenpoint coordinates determined by the third determining means for the unknown and calculating and determining the exponent. Substituting the eigenpoint coordinates determined by the third determining means into the unknown, and from each pixel of the image developed in the color space, to a color distribution approximation curve determined by substituting the unknown into the unknown Fifth determination means for determining an intersection intersected by dropping a perpendicular line for each pixel, and calculating and determining the separation value based on the frequency of pixels corresponding to each intersection on the color distribution approximation curve; The storage means is associated with the ink that can be post-printed, the eigenpoint coordinates determined by the third determination means, the index determined by the fourth determination means, It is a requirement to further comprise storing a color distribution approximate curve table defined in association with the separation value determined by the fifth determination means.

  This image processing apparatus can extract a line figure pattern of a specified ink color with high accuracy. Specifically, it is possible to extract a pattern based on the color of ink, not the observed color.

  Further, this image processing apparatus can determine information necessary for pattern extraction from two images.

  Exemplary embodiments of an image processing apparatus, an image processing method, and an image processing program according to the present invention will be described below in detail with reference to the accompanying drawings. In the following, main terms used in the present embodiment (explanation of terms), the outline and features of the image processing apparatus according to the first embodiment, the configuration of the image processing apparatus and the flow of processing will be described in order, and finally the present embodiment Various modifications to the examples will be described.

[Explanation of terms]
First, main terms used in Example 1 will be described. “Pattern” indicates a character, a line figure, or the like, for example, a character written by a person with a pen (ink) or a character formed by a stamp.

  “Post-printing” indicates that a pattern is described on some background. For example, on a printed standard form or paper (including paper with only a specific color such as blank paper), a person can write a character with a pen (ink), write a character with a stamp, This corresponds to transferring characters using carbon paper.

[Outline and Features of Image Processing Device]
Next, the outline and features of the image processing apparatus according to the first embodiment will be described. The image processing apparatus according to the first embodiment extracts pixels corresponding to a pattern with ink designated as an extraction target from an image obtained by post-printing a pattern with ink of a predetermined color on a background that is a predetermined background color. From the viewpoint of effect, it is possible to extract the line pattern pattern of the specified ink color with high accuracy. Specifically, it is not the observed color (actually it is discriminated visually) The main feature is that it is possible to extract a pattern based on the color of the ink (not the color that is produced).

  In short, as shown in FIG. 1, the image processing apparatus according to the first embodiment plots each pixel constituting the image (each point in FIG. 1 corresponds to each pixel) on the RGB space, An approximate curve (hereinafter referred to as a color distribution approximate curve) that approximates the distribution (hereinafter referred to as color distribution) shape of each point corresponding to each pixel is determined and corresponds to each point in the vicinity of the determined color distribution approximate curve. Each pixel is extracted as having the same color. For example, as illustrated in (1) of FIG. 1, the image processing apparatus according to the first embodiment extracts pixels included in a range of a certain distance from each point of the curve as the same color pattern, and extracts them. As shown in (2), the pixels included in the range of distance determined by each point of the curve are extracted as the same color pattern.

  As described above, the image processing apparatus according to the first embodiment has one feature in that the pixel distribution is extracted by determining the color distribution approximation curve that approximates the shape of the color distribution. Therefore, in the following, first, a color distribution approximation curve used by the image processing apparatus according to the first embodiment will be described (an outline of a color distribution approximation curve, a method for calculating an ink color eigenpoint, a modification of a color distribution approximation curve), and then The outline and features of the first embodiment will be described.

[About color distribution approximation curve]
First, the outline of the color distribution approximation curve will be described with reference to FIG. FIG. 1 is a diagram for explaining the outline of the color distribution approximation curve in the first embodiment.

  The image processing apparatus according to the first embodiment assumes that the color distribution approximation curve of each pixel of the line figure pattern is a curve passing through the three points of the background color center, the pattern color center, and the ink color unique point, and approximates the color distribution. Determine the curve. That is, the color distribution approximation curve in the first embodiment includes the background color center shown in FIG. 1A, the pattern color center shown in FIG. 1B, and the ink color specific points shown in FIG. And a curve that passes through the three points.

  Here, the pattern color in the first embodiment is, for example, the color of a written character, and is not an ink-specific color but an actually observed color (a color including the influence of the background). Therefore, the color of the ink to be extracted (for example, the color of the color label “blue oil ballpoint pen” itself) and the pattern color are not generally equal, and the pattern color of the pattern to be extracted is the background color and the ink color. It is determined by.

  The pattern color center in the first embodiment is a pattern color (coordinate) determined by taking out a part of the pattern by some method and obtaining a pixel average (or mode value) of the extracted pattern color. . Further, the pattern color center in the first embodiment corresponds to (a) of FIG. 1 and is one point on the color distribution approximate curve line segment in the first embodiment.

  The background color in the first embodiment is a color indicated by a pixel in the background of the pattern, and specifically, a color indicated by “background” that is an image remaining after separating the “foreground (pattern)” from the image. .

  The background color center in the first embodiment is a color coordinate indicating a color space coordinate of the background color in the image, and specifically, a coordinate taken when the ink thickness is set to the limit of zero. For example, the background color center in the first embodiment is a background color in a portion where a predetermined color ink to be extracted is described, and a part of the background portion is extracted by some method, and the color average of those pixels (Or the mode) is determined.

  Specifically, in the case where characters are written in blue ink on white paper, the background color center in Example 1 is the portion where the characters are written in blue ink. This is a color and a state in which the influence of the blue ink is eliminated, and in reality, the background color is white. As shown in FIG. 1, the background color center in the first embodiment corresponds to (b) in FIG. 1, and is the end of the color distribution approximate curve line in the first embodiment. The background color center corresponds to “background color coordinates” recited in the claims.

  The ink specific color center in the first embodiment is a coordinate specific to ink that is not affected by the background color in the color space. For example, color coordinates taken when the thickness of the ink is set to an infinite limit are applicable. Specifically, the ink specific color center in the first embodiment is a color which is the color of the ink and is shown in a state where there is no influence of the background color. For example, when a character is written in blue ink on a transparent background, the color of the character written in blue ink corresponds to the ink specific color center in the first embodiment. As shown in FIG. 1, the ink specific color center in the first embodiment corresponds to (c) in FIG. 1, and is the end of the color distribution approximate curve line in the first embodiment. The ink unique color center corresponds to “unique point coordinates” recited in the claims.

  The ink color unique point in the first embodiment is generally a point having the lowest lightness of the color distribution of each pixel of the pattern or a point outside the distribution (due to the limit value) in normal actual data.

  Here, the relationship between the background color center, the ink color unique point, the pattern color, and the pattern color center in the color distribution approximate curve line in the first embodiment will be described again with reference to FIG. As shown in FIG. 1, each pixel plotted in the RGB space represents each pixel constituting the image, and the background color center in the first embodiment corresponds to (b) of FIG. The ink color center in FIG. 1 corresponds to (c) in FIG. 1, and the pattern color center in Example 1 corresponds to (a) in FIG. That is, the background color center and the ink unique color center in Example 1 are both ends of the approximate curve line segment, and the pattern color center in Example 1 is one point on the approximate curve line segment.

[Ink color eigenpoint calculation method]
Of the above three points (background color center, pattern color center, and ink color unique point), the background color center and the pattern color center are used when extracting pixels corresponding to the pattern of ink designated as the extraction target. Further, it can be calculated from one image in which a pattern to be extracted is described. On the other hand, as described above, the ink color unique point is calculated and determined as a point (coordinates) that does not depend on the background color, and is difficult to calculate from one image (from the image). The detected pattern is an ink color that has already been affected by the background color, and is not an ink color specific point that is indicated only by the ink color), and is thus calculated in advance.

  Therefore, a method for calculating the ink color specific point in the first embodiment will be described next with reference to FIGS. FIG. 2 is a diagram for explaining a method for calculating ink color specific points in the first embodiment. FIG. 3 is a diagram for explaining a method for calculating ink color specific points in the first embodiment. FIGS. 4A and 4B are diagrams for illustrating equations used to describe the color distribution approximation curve in the first embodiment.

  That is, for the purpose of calculating the ink color unique point in the first embodiment, which is an unknown coordinate, two color distribution approximate curves are created, and the coordinate that is the intersection is calculated. Specifically, two different color distribution approximation curves are determined from each of two images described with the same ink, with the ink color characteristic point unknown. Here, as described above, the pattern color center is one point on the line segment between the background color center and the ink color unique point in the first embodiment in the color distribution approximation curve in the first embodiment. It is expressed by an expression expressed by the center and the ink color unique point. For the background color center, an average value of the background color distribution is used as an approximate value. As a result, there are two equations (because they are created for two patterns) with one unknown value (the coordinates of the ink color unique point), and the coordinates of the ink color unique point in Example 1 are calculated. is there.

  Hereinafter, a specific example will be further described. As described above, the ink color unique point in the first embodiment is calculated and determined as a point (coordinate) that does not depend on the background color. Specifically, the ink color unique point in the first embodiment is calculated and determined from two images. Here, the two images have different background colors, and a line figure pattern is post-printed only with the same ink that can be post-printed. The ink color unique point in the first embodiment is calculated as an intersection of color distribution approximate curves determined for the two images, as will be described below.

  A method for calculating the ink color specific points will be described using the color distribution in the RGB space shown in FIG. Note that the description will be made on the assumption that the color coordinate system that handles each pixel constituting the image uses a linear color coordinate system such as linear RGB or XYZ.

  First, as shown in (2) and (3) of FIG. 2, images of line figure patterns having different background colors and post-printed only with the same color ink on the respective background colors (learning) Prepare two images. For example, in the example shown in (2) of FIG. 2, the background color is “CB0 (in the drawing, B0 is described as a subscript)” (for example, white), and the postprinted line figure pattern is “ “B”, and the color (pattern color) of “B” is “CA0 (A0 is described as a subscript in the drawing)”. Also, for example, in the example shown in (3) of FIG. 2, the background color is “CB1 (B1 is shown as a subscript in the drawing)”, and “A” is used as a post-printed line figure pattern. The color (pattern color) of “A” is “CA1 (A1 is described as a subscript in the drawing)”. That is, “B” post-printed in (2) of FIG. 2 and “A” post-printed in (3) of FIG. 2 are post-printed with the same ink.

  Here, when selecting two background colors, it is possible to reduce the error of the calculated intrinsic point of the ink color by selecting a light background color on one side and a dark background color on the other side. . For example, the error can be reduced by selecting white as one background color and using the pattern color center with the background color center calculated from an image having white as the background color.

  Based on such two images, the RGB space shown in (1) of FIG. 2 shows the color distribution of each pixel constituting the image on the RGB space for each of the two images. It is. In the RGB space shown in (1) of FIG. 2, a color distribution approximation curve has already been added for the convenience of explanation for the color distribution formed by each pixel of the background portion and the line pattern portion corresponding to each set. is doing. Here, in the following, the line figure pattern showing the color distribution by the image shown in (2) of FIG. 2 is referred to as pattern 1, and the line figure pattern showing the color distribution by the image shown in (3) of FIG. Will be described. Note that the two color distribution approximation curves are determined for patterns described using the same ink, and indicate the same ink color unique point. For this reason, it is assumed that the two color distribution approximation curves intersect at the ink color characteristic point.

  Color coordinates indicating the ink color specific points corresponding to the patterns 1 and 2 shown in (1) of FIG. 2, the background color center, and the pattern color center are represented by the coordinates shown in FIG.

  Here, the image processing apparatus according to the first embodiment assumes that the color distributions of the pattern 1 and the pattern 2 are approximated by a space curve of the same formula that intersects at one point, and from the actual observation result, Assume that the equations of the color distribution approximation curve are (1) and (2) in FIG. In the equations shown in FIGS. 4A and 4B, “t” and “s” represent parameters of the color distribution approximation curve. “Λ” and “μ” are curve constants that differ depending on the ink color, and take positive values. The curve constant corresponds to an “index” described in the claims.

  Next, in the image processing apparatus according to the first embodiment, the line segment of the color distribution approximate curve in the first embodiment relating to the pixels corresponding to the patterns 1 and 2 is the intersection (rc, gc, bc) (ink It is assumed that it exists between the color eigenpoints). Then, as shown in (3) and (4) of FIG. 4A, “t” and “s” that are parameters of the color distribution approximate curve in the first embodiment are selected.

  Here, the color distribution approximation curve in Example 1 corresponding to pattern 1 passes through the color center of pattern 1, the color distribution approximation curve corresponding to pattern 2 passes through the center of color of pattern 2, and distribution approximation When the parameters “t” and “s” of the curve are set to be “t = s = tm (tm is an arbitrary value that can be taken by“ t ”and“ s ”)”, in the first embodiment, In the color distribution approximation curve, the relational expression shown in (5) of FIG.

  Then, the relational expression (5) in FIG. 4A is substituted into (1) and (2) in FIG. 4A to solve the ink color eigenpoint T = (rc, gc, bc). , (6), (7) and (8) in FIG. 4-1 are obtained.

  Here, the background color distribution in the first embodiment will be described. Unlike the color distribution corresponding to the pattern color, the background color distribution does not take a distribution form with extremely large variance in one direction, but a distribution form concentrated at one point (background color center) and its vicinity. This is because, for example, in the color distribution corresponding to the pattern color, the color coordinate of each pixel varies depending on the thickness of the ink, whereas the background color is mostly uniform (for example, This is because the background color corresponds to the color of post-printed paper and the like, and the color coordinates of each pixel do not disperse and tend to concentrate on one point (and its surroundings).

  Using such a property of the color distribution of the background color, the background color center is approximated to be an average value of the background color distribution, and is represented by an expression (9) in FIG.

  Then, the coordinates of the ink color unique point are calculated by (9), (6), (7), and (8) in FIG. 4-1, and (10), (11), and ( 12) is obtained.

  Note that since two different colors are selected as the two background colors, the equation (13) in FIG. 4-2 is naturally satisfied. In addition, since the coordinates of the ink color specific points are not the same as the coordinates indicating the pattern color center and the background color center (or such two images are not used), each coordinate is shown in FIG. -2 (17) is satisfied.

[Modification of approximate color distribution curve]
As described above, the color distribution approximation curve in the first embodiment passes through the above three points (background color center, pattern color center, and ink color unique point). Here, as described above, it has been described that the ink color unique point is calculated in advance. Here, a case will be described in which the color distribution approximation curve in the first embodiment is transformed into an expression expressed by the background color center and the curve constant obtained from the pattern color center and the ink color unique point.

  That is, a modification of the color distribution approximate curve in the first embodiment is calculated using FIGS. 4A and 4B. As described above, in the expression of the color distribution approximation curve assumed in (1) and (2) of FIG. 4-1, the logarithm of each term is taken as “t = s = tm”, and the distribution approximation curve Solving with respect to the curve constants “λ” and “μ”, (14) and (15) in FIG. 4-2 are obtained.

  As a result, (by calculating these curve constants in advance), by using the calculated coordinates of the ink color specific points and the curve constants “λ” and “μ” described above, the “background color” A color distribution approximation curve in Example 1 in which “center” is an unknown value is obtained.

  That is, first, “λ” and “μ” are calculated using the “background color center” and the “pattern color center” of one of the two learning images. Here, the color distribution regarding the pattern of the same “color label” on an arbitrary background color is assumed to have the same parameter value (value t) at the center of the color distribution. This is especially true experimentally.

  Further, as described above, the color distribution approximate curve is a curve that passes through “background color center”, “pattern color center”, and “ink color unique point”. In addition, when calculating “curve constant” and “ink color characteristic point”, “background color center” and “pattern color center” are acquired to determine a color distribution approximation curve.

  On the other hand, when actually extracting a pattern, the “pattern color center” is not extracted, but the “background color center” and the “color label (more precisely, the“ curve associated with the color label ” Using the constants (“λ” and “μ”) ”and“ ink color characteristic point ”), a color distribution approximation curve is determined. This is because when a pattern is actually extracted, generally, a plurality of types of ink color patterns are post-printed on an image for extraction (for example, on a form), and the pattern extraction process is completed. Only after the restoration is performed, a pattern to be extracted from a plurality of types of ink color patterns is known, and at the time of extraction, the pattern color center is not (cannot) be acquired.

  In other words, when actually extracting a pattern, the color distribution approximate curve does not use the pattern center, but obtains the “background color center” and “color label” to obtain the “curve constant” obtained in advance. And the color distribution approximate curve is determined using “ink color characteristic point”. In addition, the color distribution approximation curve determined in this way passes through the “pattern color center” of the extracted pattern as a result, and the “background color center”, “pattern color center”, and “ink color unique point” It becomes a curve that passes.

  The “curve constant”, “ink color unique point”, and “separation value” described above are calculated for each “color label” and prepared in advance as a table (or prepared as a table). .

  Here, as the color distribution approximation curve in the first embodiment, the expressions shown in (1) and (2) of FIG. 4-1 may be used, but the present invention is not limited to this. For each RGB component, an equation of a color distribution approximate curve indicating coordinates may be derived and used. For example, when the background color center (rB, gB, bB) is acquired from an image obtained by scanning a form to be extracted, from (1) and (2) in FIG. The distribution approximation curve shown in (18) of No. 2 may be calculated and used.

  Further, although the color distribution approximation curve in Example 1 has been described as having “λ” and “μ” as the curve constants, the present invention is not limited to this, and (( 19) and (20) (corresponding to (1) and (2) in FIG. 4-1, respectively), “λ”, “μ”, and “γ” are used as the curve constants of the distribution approximate curve. It may be used. Note that “λ”, “μ”, and “γ”, which are curve constants of the color distribution approximation curve in the first embodiment, are also calculated by the ratio between “λ”, “μ”, and “γ”, respectively. In the color distribution approximate curve in the first embodiment, any one of “λ”, “μ”, and “γ” is assumed to be “1”, and the remaining two curve constants are used. It is what.

[About overview and features]
So far, the color distribution approximation curve used by the image processing apparatus according to the first embodiment has been described. Next, the outline and features of the image processing apparatus according to the first embodiment will be described with reference to FIG. FIG. 5 is a diagram for explaining the outline and features of the image processing apparatus according to the first embodiment.

  As shown in FIG. 5, the image processing apparatus according to the first embodiment includes a curve constant, an ink color unique point, and a background color on a color distribution approximation curve for each ink (color label) that can be post-printed on an image. A color distribution approximate curve table in which separation values used to separate ink colors are defined in association with each other is stored. The separation value is a value used to determine a separation point that divides a color distribution approximate curve line segment (a line segment sandwiched between the ink color unique point and the background color center) into two line segments.

  For example, in the example shown in the color distribution approximate curve table of FIG. 5, the image processing apparatus according to the first embodiment associates the color distribution “approximate curve table” with the color label “blue oil ballpoint pen” and λ “3.1. ”, Μ“ 2.9 ”, R“ 19 ”, G“ 18 ”, B“ 44 ”, and the separation value“ 0.81 ”.

  With such a configuration, the image processing apparatus according to the first embodiment, as shown in FIG. 5 (1), for example, “pattern (black) with black ink (in the drawing, for convenience of explanation, "Pattern (black)") "with a blue oil ballpoint pen" Pattern (blue) (In the drawing, for convenience of explanation, simply "Pattern (blue)") When the designation of the ink color (blue) is accepted as shown in (2) of FIG. 5, the image on the color space of the background color in the image is received as shown in (3) of FIG. The background color coordinates indicating the coordinates are determined. For example, in the example shown in (3) of FIG. 5, a user using the image processing apparatus according to the first embodiment surrounds a part of the extraction image (background color part) using a mouse (input unit) or the like. Then, the image processing apparatus according to the first embodiment calculates the average color value of the enclosed portion and determines the background color center.

  Then, as shown in (4) of FIG. 5, the image processing apparatus according to the first embodiment reads the curve constant and the ink color specific point corresponding to the ink designated as the extraction target from the color distribution approximate curve table, and reads A curve drawn by a curve constant with the ink color unique point and the determined background color center as a base point is determined as a color distribution approximation curve related to the ink.

  For example, the image processing apparatus according to the first embodiment will be described with reference to the example shown in the color distribution approximate curve table of FIG. 5. The unique point coordinates R “19” corresponding to the color label “blue oil ballpoint pen” and G Curve constants “18” and B “44” are read from the color distribution approximate curve table and read from the color distribution approximate curve table using the background color coordinates determined in (3) of FIG. 5 and the specific point coordinates as base points. A color distribution approximation curve is determined by λ “3.1” and μ “2.9”.

  Then, as shown in FIG. 5 (5), the image processing apparatus according to the first embodiment reads the separation value corresponding to the ink designated as the extraction target from the color distribution approximate curve table, and the color distribution approximate curve determination means. The separation point between the designated ink pixel and the background pixel corresponding to the separation value is determined on the color distribution approximation curve determined by the above.

  For example, the image processing apparatus according to the first embodiment reads out the separation value “0.81” corresponding to the color label “blue oil ballpoint pen” when described using the example shown in the color distribution approximate curve table of FIG. As shown in FIG. 5A, the image processing apparatus according to the first embodiment sets the ink color unique point on the color distribution approximate curve in the first embodiment to “t = 1”, and As shown in b), when the center of the background color on the color distribution approximate curve in Example 1 is set to “t = 0”, a point where “t = separation value (0.81)” is determined.

  Then, as shown in FIG. 5 (6), the image processing apparatus according to the first embodiment converts the color distribution approximation curve determined by the color distribution approximation curve determination unit from each pixel of the image developed in the color space. After determining the intersecting point for each pixel by dropping the perpendicular line, the intersecting point is in the direction of the eigenpoint coordinates when viewed from the separating point determined by the separating point determining means, and the distance in the color space from the intersecting point is A pixel that falls within a predetermined threshold is extracted as a pixel corresponding to the ink pattern designated as the extraction target.

  For example, the image processing apparatus according to the first embodiment drops a perpendicular line from the pixel Pi ′ to the color distribution approximation curve according to the first embodiment, as illustrated in (c) of (6) of FIG. The image processing apparatus according to the first embodiment calculates and acquires a value “t = ti” indicating the intersection, and calculates and acquires a distance “d (Pi)” between the intersection and the pixel. . The image processing apparatus according to the first embodiment calculates and acquires “t” and “d” for each of all the pixels.

  Thereafter, for example, as illustrated in (d) of (6) in FIG. 5, the image processing apparatus according to the first embodiment uses “t = ti” acquired for each pixel as the separation value “t = separation value (0). .81) "is closer to the ink color intrinsic point, and it is determined whether" d "acquired for each pixel is smaller than a predetermined threshold value, and a pixel satisfying the condition is extracted. In other words, for example, when the predetermined threshold is “10a”, the image processing apparatus according to the first embodiment has “1 <t <0.81” and “d <10a”. To extract.

  Then, as illustrated in (7) of FIG. 5, the image processing apparatus according to the first embodiment restores and outputs only each extracted pixel on the image. For example, the image processing apparatus according to the first embodiment restores and outputs only the pattern (blue).

  For this reason, the image processing apparatus according to the first embodiment can extract a pattern based on the color of ink, not the color that is actually discriminated visually, as in the main feature described above. .

[Configuration of image processing apparatus]
Next, the configuration of the image processing apparatus according to the first embodiment illustrated in FIG. 5 will be described with reference to FIGS. FIG. 6 is a block diagram showing the configuration of the image processing apparatus. FIG. 7 is a diagram for illustrating an example of information stored in the setting information storage unit 31. FIG. 8 is a diagram for explaining the separation value calculation unit 43. FIG. 9 is a diagram for explaining the distance / parameter calculation unit 54. FIG. 10 is a diagram for explaining the extraction determination unit 55. FIG. 11 is a diagram for explaining a method of setting a predetermined threshold value. FIG. 12 is a diagram for explaining the restoration unit.

  As shown in FIG. 6, the image processing apparatus according to the first embodiment includes an input unit 10, an output unit 20, a storage unit 30, a learning control unit 40, and a pattern extraction control unit 50.

  In the image processing apparatus described below, as shown in FIG. 6, information necessary for extracting a specified specific color pattern (for example, ink color specific points, curve constants “λ”, “μ”, etc.) ) Will be described, and the same color pattern extraction control unit 50 that extracts a specified color pattern will be described. However, the present invention is not limited to this. For example, the image processing apparatus may store the necessary information in the storage unit 30 and include only the pattern extraction control unit 50. Moreover, you may utilize the learning control part 40 as another apparatus. In other words, the image processing apparatus according to the present invention only needs to be able to use necessary information (for example, ink color specific points, curve constants “λ”, “μ”, etc.) when extracting a specified specific color pattern. The necessary information may be stored as a table in the storage unit 30 and the present invention may be carried out. The learning control unit 40 may calculate the information and implement the present invention each time.

  Among these, the input unit 10 receives input of various types of information, and includes a keyboard, a mouse, a microphone, a scanner, and the like. For example, the input unit accepts input of two images input for the purpose of calculating ink color unique points, etc., accepts input of an image to be extracted, and specifies ink color and background color. Accept. The output unit 20 outputs various images and includes a monitor (or display or touch panel). For example, the output unit 20 outputs an input image or an image obtained by extracting only a designated color.

[Storage etc.]
The storage unit 30 stores data and programs necessary for processing by the pattern extraction control unit 50. In particular, the configuration information storage unit 31 is provided as closely related to the present invention. The setting information storage unit 31 corresponds to a “storage unit” described in the claims.

  The setting information storage unit 31 includes, for each ink that can be specified as an extraction target, a curve constant used for drawing the approximate color distribution curve of the ink in the color space, and an ink-specific value that is not affected by the background color in the color space. Used to separate the background color and the ink color on the color distribution approximate curve and the intrinsic point coordinates indicating the coordinates (for example, the coordinates taken when the thickness of the ink is infinite in the color space) A color distribution approximate curve table defined in association with the separation values to be stored is stored. The setting information storage unit 31 stores a color distribution approximate curve table in which ink color specific points, curve constants, and separation values determined by the learning control unit 40 described later are defined in association with each other. The information stored in the setting information storage unit 31 is used by a pattern extraction control unit described later.

  Specifically, as shown in FIG. 7, the setting information storage unit 31 associates the color constant of the color distribution approximation curve with a “color label” indicating information for uniquely identifying ink, as shown in FIG. “Curve constant”, “Ink color characteristic point”, and “Separation value” indicating the separation value are stored. More specifically, as shown in FIG. 7, the setting information storage unit 31 stores “λ” and “μ” as curve constants, and the ink color specific points in the RGB space as ink color specific points. “R”, “G”, and “B” indicating the color coordinates are stored. For example, in the example illustrated in FIG. 7, the setting information storage unit 31 associates the color label “black oil ballpoint pen” with λ “1.05”, μ “1.05”, and R “27”. , G “22”, B “27”, and separation value “0.8” are stored, and in association with the color label “blue oil ballpoint pen”, λ “3.1” and μ “2. 9 ", R" 19 ", G" 18 ", B" 44 ", and separation value" 0.81 "are stored.

[Learning control unit]
The learning control unit 40 has a program that prescribes a processing procedure for learning information necessary for extracting a specified specific color pattern and a memory for storing necessary data, and executes various processes using these programs. To do. For example, the learning control unit 40 has equations shown in (10) to (12) of FIG. 4-2 in the memory, and executes various processes. Particularly closely related to the present invention includes a representative color calculation unit 41, a color distribution approximate curve constant calculation unit 42, and a separation value calculation unit 43. Here, the representative color calculation unit 41 and a background designation unit 52 described later correspond to “background color coordinate determination means” described in the claims, and the color distribution approximate curve constant calculation unit 42 claims Corresponding to the “second determining means”, the “third determining means”, and the “fourth determining means” described in the above-mentioned range. Corresponds to “determining means”.

  In describing each part of the learning control unit 40, a paper having a uniform background color on which a line figure pattern is post-printed with ink to be extracted on the paper (for example, The report will be explained on the premise that learning color images (for two sheets) that are color images obtained by selecting two sheets with different background colors and scanning the two sheets are prepared. . Note that it is desirable that one of the two sheets used for preparing the learning color image has a white background.

  In addition, the learning color image here is an image used for the purpose of learning information necessary for extracting a specified specific color pattern, and when actually extracting a specific color. The following description will be made on the assumption that the image is different from the extraction target image which is an image used for. However, the present invention is not limited to the case where the learning color image and the extraction image are different from each other as described above, and the extraction image may be used as the learning color image (that is, extraction). Information necessary for extracting the specified specific color pattern may be learned using the image, and then the extraction process may be performed).

  The representative color calculation unit 41 determines the background color center for each of the two images. Further, the representative color calculation unit 41 determines the pattern color center. For example, a user who uses an image processing apparatus designates a part of a background area or a line figure pattern area by surrounding a background color part and a line figure pattern part with a mouse in each of two learning images. The representative color calculation unit 41 obtains the color of each pixel in each designated area and calculates the average value of the obtained colors, thereby calculating the pattern color center and the background color center in the two learning images. To do. The representative color calculation unit 41 stores the calculated pattern color center and background color center in the setting information storage unit 31.

  The representative color calculation unit 41 is not limited to the method of calculating the pattern color center and the background color center by calculating the average value of the acquired colors. For example, the representative color calculation unit 41 calculates the mode value. Thus, the pattern color center and the background color center may be calculated. For example, the representative color calculation unit 41 may calculate the pattern color center and the background color center independently by the image processing apparatus without the user specifying a part of the image. For example, the mode value may be calculated from the entire image. Specifically, the representative color calculation unit 41 creates a histogram using the frequency for each pixel constituting the image. For example, the representative color calculation unit 41 sets the mode value as the background color center, Alternatively, a value having a high frequency may be used as the pattern color center.

  The color distribution approximation curve constant calculation unit 42 uses a pattern color center and a background color center determined by the representative color calculation unit 41 for each of the two images, and a color distribution approximation curve having an unknown ink color eigenpoint. Respectively. Subsequently, the color distribution approximate curve constant calculation unit 42 calculates and determines the unknown ink color specific point coordinates, assuming that the unknown ink color specific points are the same in the two determined color distribution approximate curves. . Thereafter, the color distribution approximate curve constant calculation unit 42 substitutes the determined ink color specific point for the unknown and calculates and determines the curve constant.

  For example, the color distribution approximate curve constant calculation unit 42 substitutes the pattern color center and the background color center calculated by the representative color calculation unit 41 into (10) to (12) in FIG. Each color distribution approximate curve having an eigenpoint as an unknown is determined, and an ink color eigenpoint is determined. Thereafter, the color distribution approximate curve constant calculation unit 42 solves the curve constant of the color distribution approximate curve using the determined ink color eigenpoints as shown in (14) to (15) of FIG. The curve constant of the color distribution approximate curve is determined.

  A specific example of a method for calculating the ink color characteristic point is described in [Method for calculating ink color characteristic point] described above, and a specific method for calculating the curve constant of the color distribution approximate curve. The example is also described in [Modified example of color distribution approximation curve] described above, and is omitted here. Further, the color distribution approximate curve constant calculation unit 42 stores the calculated curve constant of the color distribution approximation curve and the ink color unique point in the setting information storage unit 31.

  The separation value calculation unit 43 is determined by substituting the eigenpoint coordinates determined by the color distribution approximate curve constant calculation unit 42 for the unknown and substituting the unknown for each pixel of the image developed in the color space. After determining an intersection point that intersects the color distribution approximation curve with a perpendicular line for each pixel, a separation value is calculated and determined based on the frequency of pixels corresponding to each intersection point on the color distribution approximation curve.

  For example, as shown in (1) of FIG. 8, the separation value calculation unit 43 includes a pattern color center calculated by the representative color calculation unit 41, a background color center, and a color distribution calculated by the color distribution approximate curve constant calculation unit 42. Using one of the two learning color images, a color distribution approximation curve corresponding to each pixel constituting the image is calculated using the curve constant of the approximate curve and the ink color characteristic point. When selecting one of the two learning color images, by selecting an image whose background color is white (or an image having a bright background color relative to other images), The accuracy can be improved.

  For example, as illustrated in (2) of FIG. 8, the separation value calculation unit 43 applies the color coordinates in the RGB space indicated by each pixel constituting the image to the calculated color distribution approximation curve. ) To obtain an intersection corresponding to each pixel. For example, in the example illustrated in (2) of FIG. 8, the separation value calculation unit 43 draws a perpendicular line from the pixel Pi to the color distribution approximation curve, and obtains an intersection Pi ′.

  For example, the separation value calculation unit 43 obtains a value indicating a corresponding intersection for each pixel. Note that the value indicating the intersection calculated here is a value for indicating the position on the line segment in the line segment of the color distribution approximation curve between the ink color unique point and the background color center. For example, when the ink color unique point is “1” and the background color center is “0”, the value indicating the intersection is a value between “0” and “1”. That is, as shown in (3) of FIG. 8, the separation value calculation unit 43 corresponds to Pi ′ when the ink color unique point is “t = 1” and the background color center is “t = 0”. As “t”, “t = ti” is calculated.

  Subsequently, for example, as illustrated in (4) of FIG. 8, the separation value calculation unit 43 calculates a separation value. For example, in the example shown in (4) of FIG. 8, the separation value calculation unit 43 sets “t” on the horizontal axis and the frequency (the number of pixels corresponding to the value on the horizontal axis “t”) on the vertical axis. Create a frequency histogram. Then, a separation value is calculated by binarization by discrimination. Specifically, in the frequency histogram shown in (4) of FIG. 8, “t = 1” indicates the ink color unique point, and “t = 0” indicates the background color center. For this reason, in the frequency histogram, as shown in (4a) of FIG. 8, it corresponds to a large color formed by each pixel corresponding to the background color and a pattern color as shown in (4b) of FIG. A gentle mountain will be formed. Here, the separation value calculation unit 43 calculates the boundary between the two peaks and sets it as the separation value “tb”. The separation value calculation unit 43 stores the calculated separation value “tb” in the setting information storage unit 31.

  The image processing apparatus according to the first embodiment learns (calculates) the curve constants and separation values of the color distribution approximation curve in advance in association with “color labels” for all the inks to be extracted, This is stored in the setting information storage unit 31.

[Pattern extraction control unit]
The pattern extraction control unit 50 includes a program that defines a processing procedure for extracting a specified specific color pattern and a memory for storing necessary data, and executes various processes using these programs. For example, the pattern extraction control unit 50 has expressions shown in (18) of FIG. 4-2 and (1) or (2) of FIG. 4-1 in the memory, and executes various processes. Particularly closely related to the present invention are the extraction target specifying unit 51, the background specifying unit 52, the color distribution approximate curve creating unit 53, the distance / parameter calculating unit 54, the extraction determining unit 55, and the restoring unit. 56.

  The background designating unit 52 corresponds to the “background color coordinate determining unit” described in the claims, and the color distribution approximate curve creating unit 53 is the “color distribution approximate curve determining unit” described in the claims. The distance / parameter calculation unit 54 and the extraction determination unit 55 correspond to “separation point determination means” and “extraction means” described in the claims. In this embodiment, it is assumed that the brightness of the foreground (pattern portion) is lower than that of the background by post-printing.

  The pattern extraction control unit 50 described below can extract a line figure pattern of a specified ink color with high accuracy by using the setting information stored in the setting information storage unit 31. Specifically, One of the features is that the pattern is extracted based on the ink color, not the observed color, and that the extraction omission is reduced as compared with the conventional method.

  The extraction target specifying unit 51 specifies a specific color to be extracted. Specifically, the extraction target designating unit 51 designates which “color label” is the specific color to be extracted, using the example shown in FIG. For example, when the user inputs the color label “black oil ballpoint pen” from the input unit, the extraction target designating unit 51 designates the black oil ballpoint pen. The extraction target designating unit 51 may estimate and use a specific color that is directly extracted from the extraction image. Note that the extraction target specifying unit 51 may output the color label and type to be extracted from the output unit 20 as a menu, and may accept selection from the user from a keyboard or the like.

  The background designating unit 52 determines the background color center indicating the color space coordinates of the background color in the image. For example, when the user surrounds a part of the extraction image (background color part) using a mouse (input unit) or the like, the background designating unit 52 calculates the color average value of the enclosed part and calculates the background. Specify the color center. The background designating unit 52 may calculate and estimate the mode value instead of calculating the average value. The background designating unit 52 is not limited to designating the background color center from the area surrounded by the user, and calculates the mode value from the entire pixels (or a predesignated part) constituting the extraction image. The background color center may be estimated and specified.

  The color distribution approximate curve creation unit 53 determines a color distribution approximate curve. Specifically, the curve constant and ink color specific point corresponding to the ink designated as the extraction target are read from the color distribution approximate curve table, and the background color center determined by the ink color specific point and the background designating unit 52 is the base point. The curve drawn by the curve constant is determined as a color distribution approximation curve related to the ink.

  For example, referring to the example shown in FIG. 7, the color distribution approximate curve creating unit 53 receives the λ “1” from the setting information storage unit 31 when the extraction target designating unit 51 designates the color label “black oil ballpoint pen”. .05 ”, μ“ 1.05 ”, R“ 27 ”, G“ 22 ”, and B“ 27 ”. In addition, the color distribution approximate curve creation unit 53 acquires color coordinates indicating the background color center designated by the background color center. Then, a color distribution approximation curve is created by substituting the acquired value into (18) in FIG.

  The distance / parameter calculation unit 54 determines, for each pixel, an intersection point where each pixel of the image developed in the color space intersects the color distribution approximate curve determined by the color distribution approximate curve creation unit 53 with a perpendicular line. Then, a value indicating the intersection on the color distribution approximate curve and a distance from the intersection to the pixel are calculated for each pixel. For example, as shown in FIG. 9, the distance / parameter calculation unit 54 draws a perpendicular line to the color distribution approximate curve created by the color distribution approximate curve creation unit 53 for each pixel constituting the extraction image. Then, the distance / parameter calculation unit 54 calculates a value “ti” indicating the intersection point for each pixel, as shown in (1) of FIG. 9, and as shown in (2) of FIG. The distance “d (Pi)” between the intersection and the pixel is calculated for each pixel.

  The extraction determination unit 55 reads the separation value corresponding to the ink designated as the extraction target from the color distribution approximation curve table, and corresponds to the separation value on the color distribution approximation curve determined by the color distribution approximation curve creation unit 53. The separation point between the designated ink pixel and the background pixel is determined.

  For example, the extraction determination unit 55 reads the separation value “0.81” corresponding to the color label “blue oil ball-point pen”, using the example shown in the color distribution approximate curve table of FIG. Then, as shown in FIG. 10, the image processing apparatus according to the first embodiment sets the ink color characteristic point on the color distribution approximate curve in the first embodiment to “t = 1”, and the color distribution approximate curve in the first embodiment. When the upper background color center is set to “t = 0”, a point “tb” that is “t = separation value (0.81)” is determined.

  In addition, the extraction determination unit 55 intersects each pixel of the image developed in the color space by the distance / parameter calculation unit 54 with a perpendicular line to the color distribution approximate curve determined by the color distribution approximate curve determination unit. Is determined for each pixel, and a pixel in which the intersection point is in the direction of the unique point coordinate as viewed from the separation point and the distance in the color space with the intersection point is within a predetermined threshold is designated as an extraction target. Extracted as pixels corresponding to the ink pattern.

  For example, the extraction determination unit 55 obtains “t = ti” obtained for each pixel closer to the ink color intrinsic point than the separation value “t = separation value (0.81)” and obtained for each pixel. It is determined whether “d” is smaller than a predetermined threshold, and pixels that satisfy the condition are extracted. In other words, for example, when the predetermined threshold is “10a”, the extraction determination unit 55 extracts pixels that are “1 <t = <0.81” and “d <10a”. .

  In other words, the extraction determining unit 55 determines each pixel corresponding to “t” larger than the separation value “tb” (1 ≧ t ≧ tb) (for example, “P2” corresponding to “t2”, “ The pixel corresponding to “t3” is “P3”), which is the extraction target pixel candidate, and is smaller than the separation value “tb” (tb> t ≧ 0). Each pixel corresponding to “t” (for example, “t1”) Pixel “P1”) to be removed as a background. In addition, the extraction determination unit 55 removes pixels that are not within the predetermined threshold “10a”. Then, the extraction determination unit 55 extracts each remaining pixel.

  Here, the extraction determination unit 55 sets a predetermined threshold as a method using a fixed threshold as shown in (1) of FIG. 11 or for each curve parameter as shown in (2) of FIG. Each pixel belonging to a specific color may be extracted by employing a method using a threshold value.

  Further, the extraction determination unit 55 determines whether or not each pixel is closer to the ink color fixed point than the separation value and whether or not the corresponding distance is within a predetermined threshold, and both requirements (intersection points are determined). Although a method of extracting pixels corresponding to a value satisfying a value that is closer to the ink color specific point and having a distance within a predetermined threshold has been described, the present invention is not limited to this. For example, extraction may be performed in multiple stages. For example, first, each pixel on the ink color specific point side as viewed from the separation value is extracted as each extraction target pixel candidate, and then, for each extraction target pixel candidate, a distance corresponding to each pixel is predetermined. It may be determined whether it is within the threshold.

  The restoration unit 56 outputs only each pixel extracted by the extraction determination unit 55. Specifically, an image is output in which only the pixels extracted by the extraction determination unit 55 are left out of the pixels constituting the extraction target image. In that case, black or white may be set as the background color according to the user's specification, and the line figure pattern color formed by each extracted pixel may be restored and output. Further, each extracted pixel may be converted into a color designated by the user, restored, and output.

  For example, as shown in FIG. 12, the background color is yellow (part (1) in FIG. 12), black letters (part (2) in FIG. 12), and letters written with a blue ballpoint pen (( 3) When an image having a part) is used as an extraction image and a blue ballpoint pen is extracted as a specific color, the restoration unit 56 restores and outputs only the part described with the blue ballpoint pen on the white image.

  This image processing apparatus is applied to an information processing apparatus such as a known personal computer, workstation, mobile phone, PHS terminal, mobile communication terminal, or PDA, and the pattern extraction control unit 50, learning control unit 40, and setting information described above. It can also be realized by installing each function of the storage unit 31. In this case, the learning control unit 40 may be omitted.

[Processing by image processing device]
Next, processing of the image processing apparatus will be described with reference to FIGS. 13 and 14. First, processing by the learning control unit will be described, and then processing by the pattern extraction control unit will be described. FIG. 13 is a flowchart showing the flow of processing by the learning control unit 40 in the first embodiment. FIG. 14 is a flowchart illustrating a flow of processing performed by the pattern extraction control unit 50 according to the first embodiment.

[Processing by learning control unit]
As shown in FIG. 13, when two learning color images are input (Yes in step S101), the representative color calculation unit 41 determines a background color center and a pattern color center (step S102). For example, the representative color calculation unit 41 obtains the color of each pixel in each designated area and calculates the average value of the obtained colors, thereby obtaining the pattern color center and the background color center in the two learning images. decide. Thereafter, the representative color calculation unit 41 stores the calculated pattern color center and background color center in the setting information storage unit 31.

  Then, the color distribution approximate curve constant calculation unit 42 uses the background color center and the pattern color center calculated by the representative color calculation unit 41 to determine the curve constant and ink color unique point of the color distribution approximation curve (step). S103). For example, the color distribution approximate curve constant calculation unit 42 substitutes the pattern color center and the background color center calculated by the representative color calculation unit 41 into (10) to (12) in FIG. Determine the point. Thereafter, the color distribution approximate curve constant calculation unit 42 solves the curve constants of the color distribution approximate curve using the calculated ink color eigenpoints as shown in (14) to (15) of FIG. The curve constant of the color distribution approximate curve is determined. Thereafter, the color distribution approximate curve constant calculation unit 42 stores the determined background color center and pattern color center in the setting information storage unit 31.

  Then, the separation value calculation unit 43 uses the pattern color center, the background color center, the curve constant of the distribution approximation curve, and the ink color eigenpoints to configure each pixel constituting one of the two learning color images. A color distribution approximation curve corresponding to is determined (step S104). Subsequently, the separation value calculation unit 43 draws a perpendicular line from each pixel (color coordinates in the RGB space indicated by each pixel) constituting the image with respect to the determined color distribution approximation curve, and corresponds to each pixel. A value indicating the intersection is obtained (step S105).

  Then, the separation value calculation unit 43 calculates a separation value (step S106). For example, in the example shown in (4) of FIG. 8, the separation value calculation unit 43 sets “t” on the horizontal axis and the frequency (the number of pixels corresponding to the value on the horizontal axis “t”) on the vertical axis. Create a frequency histogram. Then, a separation value is calculated by binarization by discrimination. More specifically, in the frequency histogram shown in (4) of FIG. 8, “t = 1” indicates the ink color unique point, and “t = 0” indicates the background color center. For this reason, in the frequency histogram, as shown in (4a) of FIG. 8, it corresponds to a large color formed by each pixel corresponding to the background color and a pattern color as shown in (4b) of FIG. A gentle mountain will be formed. Here, the separation value calculation unit 43 calculates the boundary between the two peaks and sets it as the separation value “tb”. Then, the separation value calculation unit 43 stores the calculated separation value in the setting information storage unit, and ends the process (END).

[Processing by pattern extraction control unit]
As shown in FIG. 14, when an image for extraction is input (Yes at Step S201), the extraction target specifying unit 51 specifies a specific color to be extracted (Step S202). For example, referring to the example shown in FIG. 7, which “color label” is specified as the specific color to be extracted.

  Then, the background specifying unit 52 determines the background color center (step S203). For example, when the user surrounds a part of the extraction image (background color part) using a mouse (input unit) or the like, the background designating unit 52 calculates the color average value of the enclosed part and calculates the background. Determine the color center.

  Subsequently, the color distribution approximate curve creation unit 53 determines a color distribution approximate curve (step S204). For example, referring to the example shown in FIG. 7, the color distribution approximate curve creating unit 53 receives the λ “1” from the setting information storage unit 31 when the extraction target designating unit 51 designates the color label “black oil ballpoint pen”. .05 ”, μ“ 1.05 ”, R“ 27 ”, G“ 22 ”, and B“ 27 ”. In addition, the color distribution approximate curve creation unit 53 acquires color coordinates indicating the background color center designated by the background color center. Then, the color distribution approximate curve creating unit 53 creates a color distribution approximate curve by substituting the acquired value into (18) of FIG.

  After that, the distance / parameter calculation unit 54 calculates a value indicating the intersection and a distance (step S205). For example, as shown in FIG. 9, the distance / parameter calculation unit 54 draws a perpendicular line to the color distribution approximate curve created by the color distribution approximate curve creation unit 53 for each pixel constituting the extraction image. Then, the distance / parameter calculation unit 54 calculates a value “ti” indicating the intersection point for each pixel, as shown in (1) of FIG. 9, and as shown in (2) of FIG. The distance “d (Pi)” between the intersection and the pixel is calculated for each pixel.

  Then, the extraction determination unit 55 determines a separation point (step S206). For example, the extraction determination unit 55 reads the separation value “0.81” corresponding to the color label “blue oil ball-point pen”, using the example shown in the color distribution approximate curve table of FIG. Then, as shown in FIG. 10, the image processing apparatus according to the first embodiment sets the ink color characteristic point on the color distribution approximate curve in the first embodiment to “t = 1”, and the color distribution approximate curve in the first embodiment. When the upper background color center is set to “t = 0”, a point “tb” that is “t = separation value (0.81)” is determined.

  Then, the extraction determination unit 55 determines whether or not each pixel constitutes the background color (step S207). For example, the extraction determination unit 55 determines whether or not the value “t” indicating the intersection corresponding to the pixel is closer to the ink color specific point than the separation value.

  Then, the extraction determination unit 55 determines whether or not it belongs to the specific color that is the extraction target (step S208). For example, the extraction determination unit 55 determines, for each extraction target pixel candidate, whether or not the distance corresponding to each pixel is within a preset threshold value. Then, the extraction determining unit 55 extracts each pixel that can be said to satisfy the two requirements as a pixel belonging to the specific color to be extracted (step S209).

  The restoration unit 56 outputs only each pixel extracted by the extraction determination unit 55 (step S210). For example, as shown in FIG. 12, the restoration unit 56 has a yellow background (part (1) in FIG. 12), black characters (part (2) in FIG. 12), and characters written with a blue ballpoint pen. When a certain image (part (3) in FIG. 12) is used as an extraction image and a blue ballpoint pen is extracted as a specific color, the restoration unit 56 restores only the portion described with the blue ballpoint pen on the white image. And output. Then, the process ends (END).

[Effect of Example 1]
As described above, according to the first embodiment, for each ink that can be designated as an extraction target, a curve constant, an eigenpoint coordinate, and a separation value that are used to draw an ink color distribution approximate curve in the color space. The color distribution approximate curve table defined in association with is stored, the background color coordinates are determined, the curve constants and unique point coordinates corresponding to the ink specified as the extraction target are read from the color distribution approximate curve table, and the unique points The color distribution approximate curve drawn by the curve constant is determined based on the background color coordinates determined by the coordinate and background color coordinate determination means, and the separation value corresponding to the specified ink is read from the color distribution approximate curve table, and the color distribution The separation point corresponding to the separation value is determined on the approximate curve, and the intersection point that intersects each pixel of the image developed in the color space with a perpendicular to the color distribution approximate curve. After each element is determined, pixels that are in the direction of the eigenpoint coordinates when viewed from the separation point and whose distance in the color space from the intersection point is within a predetermined threshold correspond to the pattern of ink specified as the extraction target. Therefore, it is possible to extract the line figure pattern of the designated ink color with high accuracy. Specifically, it is possible to extract a pattern based on the color of ink, not the observed color.

  For example, in the conventional method, printing or ink is applied to a post-printed line pattern that is opaque, and when ink is applied to a post-printed line pattern that is not completely opaque, it is specified. Ink color line figure pattern cannot be extracted with high accuracy.

  In addition, for example, in the conventional method, when applied to a post-printed line figure pattern, the pattern color changes if the paper color changes even if the ink color is the same, so the pattern color to be extracted must be specified each time. I must. In other words, since the ink color and the pattern color are different, even if the patterns are described using the same ink color, it is necessary to determine the pattern color again because the background color is different.

  By applying the present invention compared to such a conventional method, a pattern based on the color of the ink can be used regardless of the nature of the ink (for example, the influence of the thickness, background color, transparency, etc.). Can be extracted.

  In other words, as described above, by calculating the curve constant of the color distribution approximation curve in the first embodiment (for example, the ink color characteristic point in the first embodiment, “λ”, and “μ”), it can be arbitrarily set. As long as the specified ink color on the background color is known, it is possible to easily calculate the color distribution approximation curve in the first embodiment for the color distribution corresponding to the ink color.

  For example, by applying the present invention, not only the mixed color of the pattern color and the background color of the line figure pattern contour portion that occurs at the time of scanning, but also the ink color caused by the ink being not completely transparent and uneven ink thickness In consideration of the color mixing phenomenon with the background color, it is possible to accurately extract only pixels belonging to the line figure pattern post-printed with the specified ink color from the input image.

  The same ink pattern on an arbitrary background color can be obtained simply by learning curve constants and the like necessary for determining a color distribution approximate curve from the pattern color distribution of line figures using the same ink as the two sets of background colors. An approximate curve of the distribution of can be immediately derived. That is, it is possible to extract a line figure pattern on an arbitrary background color by once specifying the ink color of the line figure to be extracted. Furthermore, in the present invention, it is also possible to estimate and restore the color on the white background of the line figure pattern post-printed on the colored background and causing color mixing.

  Thus, as shown in FIG. 15, when extracting the pattern shown in (1) of FIG. 15, the extraction result by the conventional method ((3) of FIG. 15) is compared with (2) of FIG. It is possible to extract with high accuracy as shown in FIG. FIG. 15 is a diagram for illustrating the effect of the image processing apparatus according to the first embodiment.

  Further, according to the first embodiment, for each of two images having different background colors, each of the pattern color coordinates indicating the coordinates on the color space of each pixel constituting the pattern described by the ink is calculated, and the calculated pattern The average or mode value of each color coordinate is determined as the pattern color center, the background color coordinate is determined for each of the two images, and the pattern color center and the background color center are used for each of the two images. Then, each color distribution approximation curve having an unknown ink color unique point is determined, and each of the determined ink color unique points in the two determined color distribution approximate curves is the same. The determined ink color eigenpoint is substituted for the unknown, and the curve constant is calculated and determined, and the determined ink color eigenpoint is substituted for the unknown and expanded into the color space. After determining for each pixel the intersection that intersects the color distribution approximation curve determined by substituting it into the unknown from each pixel of the image, the frequency of the pixel corresponding to each intersection on the color distribution approximation curve is determined. Since the separation value is calculated based on the information, information necessary for pattern extraction can be determined from the two images.

  Although the embodiments of the present invention have been described so far, the present invention may be implemented in various different forms other than the embodiments described above.

(1) System configuration, etc. Of the processes described in this embodiment, all or part of the processes described as being automatically performed can be manually performed, or can be manually performed. All or part of the described processing can be automatically performed by a known method. For example, the background color center may be specified automatically or manually. In addition, for information including processing procedures, control procedures, specific names, various data and parameters shown in the document and drawings (for example, FIG. 1 to FIG. 7, FIG. 13, FIG. 14, FIG. 16) It can be changed arbitrarily unless otherwise specified.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured. For example, in the example illustrated in FIG. 6, the representative color calculation unit 41 and the background designation unit 52 may be integrated, and the learning control unit 40 and the pattern extraction control unit 50 may be separated.

(2) Program In the first embodiment, the case where various processes are realized by hardware logic has been described. However, the present invention is not limited to this, and a program prepared in advance is executed by a computer. It may be realized by this. Therefore, in the following, an example of a computer that executes an image processing apparatus control program having the same function as that of the image processing apparatus described in the first embodiment will be described with reference to FIG. FIG. 16 is a diagram illustrating a program of the image processing apparatus according to the first embodiment.

  As shown in the drawing, the image processing apparatus 3000 according to the second embodiment is configured by connecting an operation unit 3001, a microphone 3002, a speaker 3003, a display 3004, a communication unit 3006, a CPU 3010, a ROM 3011, an HDD 3012, and a RAM 3013 via a bus 3009 or the like. Has been.

  The ROM 3011 includes a representative color calculation unit 41, a color distribution approximate curve constant calculation unit 42, a separation value calculation unit 43, an extraction target specification unit 51, a background specification unit 52, a color, and the like described in the first embodiment. A distribution approximate curve creation unit 53, a distance / parameter calculation unit 54, an extraction determination unit 55, a restoration unit 56, and a control program that exhibits similar functions, that is, as shown in FIG. 16, a representative color calculation program 3011a. Color distribution approximate curve constant calculation program 3011b, separation value calculation program 3011c, extraction target determination program 3011d, background designation program 3011e, color distribution approximate curve creation program 3011f, distance / parameter calculation program 3011g, extraction A determination program 3011h and a restoration program 3011i are stored in advance. Note that these programs 3011a to 3011i may be appropriately integrated or separated as in the case of each component of the image processing apparatus shown in FIG.

  Then, the CPU 3010 reads these programs 3011a to 3011i from the ROM 3011 and executes them, so that as shown in FIG. 16, for each program 3011a to 3011i, a representative color calculation process 3010a and a color distribution approximate curve constant calculation process. 3010b, separation value calculation process 3010c, extraction target determination process 3010d, background designation process 3010e, color distribution approximate curve creation process 3010f, distance / parameter calculation process 3010g, extraction determination process 3010h, and restoration process 3010i. To function. Each process 3010a to 3010i includes a representative color calculation unit 41, a color distribution approximate curve constant calculation unit 42, a separation value calculation unit 43, an extraction target designation unit 51, and a background designation unit 52 shown in FIG. Corresponds to a color distribution approximate curve creation unit 53, a distance / parameter calculation unit 54, an extraction determination unit 55, and a restoration unit 56, respectively.

  The HDD 3012 is provided with a setting information table 3012a. The setting information table 3012a corresponds to the setting information storage unit 31.

  The CPU 3010 reads the setting information table 3012a from the HDD 3012, stores it in the RAM 3013, and executes the image processing apparatus program using the setting information data 3013a stored in the RAM 3013.

  Further, each processing function performed in each device can be realized as hardware by wired logic.

(3) Others The image processing apparatus described in the present embodiment can be realized by executing a program prepared in advance on a computer such as a personal computer or a workstation. This program can be distributed via a network such as the Internet. The program can also be executed by being recorded on a computer-readable recording medium such as a hard disk, a flexible disk (FD), a CD-ROM, an MO, and a DVD and being read from the recording medium by the computer.

(Supplementary note 1) An image processing method for extracting pixels corresponding to a pattern with ink designated as an extraction target from an image obtained by post-printing a pattern with ink of a predetermined color on a background which is a predetermined background color. An image processing program to be executed by
A background color coordinate determination procedure for determining a background color coordinate indicating a color space coordinate of a background color in the image;
For each ink that can be designated as the extraction target, an index used to draw a color distribution approximation curve of the ink in a color space, and a unique point coordinate indicating a unique coordinate of the ink that is not affected by the background color in the color space The index corresponding to the ink designated as the extraction target from the color distribution approximation curve table in which the separation values used for separating the background color and the ink color on the color distribution approximation curve are defined in association with each other Color distribution approximation that reads out the unique point coordinates and determines the curve drawn by the index based on the unique point coordinates and the background color coordinates determined by the background color coordinate determination procedure as a color distribution approximate curve related to the ink Curve determination procedure;
The separation value corresponding to the ink designated as the extraction target is read from the color distribution approximation curve table, and designated on the color distribution approximation curve determined by the color distribution approximation curve determination procedure. A separation point determination procedure for determining a separation point between the ink pixel and the background pixel;
After each pixel of the image developed in the color space determines an intersection for each pixel by crossing the color distribution approximation curve determined by the color distribution approximation curve determination procedure for each pixel, the intersection is the separation point A pattern with ink designated as the extraction target is a pixel that is in the direction of the unique point coordinates as viewed from the separation point determined by the determination procedure and whose distance in the color space with the intersection is within a predetermined threshold. An extraction procedure for extracting pixels corresponding to
An image processing program for causing a computer to execute.

(Supplementary Note 2) For each of two learning images having different background colors on which a pattern is post-printed only by the same ink that can be post-printed, the color of each pixel constituting the pattern post-printed by the ink Calculating each pattern color coordinate indicating a spatial coordinate, and causing the computer to further execute a first determination procedure for determining an average value or mode value of each calculated pattern color coordinate as a pattern color center coordinate;
The background color coordinate determination procedure determines the background color coordinate for each of the two images,
For each of the two images, using the pattern color center coordinates determined by the first determination procedure and the background color coordinates determined by the background color coordinate determination procedure, the eigenpoint coordinates are unknown. A second determination procedure for determining the respective color distribution approximation curves,
Third determination procedure for calculating and determining the eigenpoint coordinates that are unknowns, assuming that the eigenpoint coordinates that are unknowns in the two color distribution approximation curves determined by the second determination procedure are the same. When,
A fourth determination procedure for substituting the eigenpoint coordinates determined by the third determination procedure for the unknown and calculating and determining the exponent;
Substituting the eigenpoint coordinates determined by the third determination procedure into the unknown and perpendicular to the color distribution approximation curve determined by substituting the unknown into the unknown from each pixel of the image developed in the color space A fifth determination procedure for calculating and determining the separation value based on the frequency of the pixel corresponding to each intersection on the color distribution approximation curve after determining the intersection for each pixel by lowering
Corresponding to the ink that can be post-printed, the unique point coordinates determined by the third determination procedure, the index determined by the fourth determination procedure, and the fifth determination procedure are determined. A storage procedure for storing in the color distribution approximate curve table defined in association with the separation value;
The image processing program according to appendix 1, wherein the computer is further executed.

(Supplementary Note 3) An image processing apparatus that extracts pixels corresponding to a pattern with ink designated as an extraction target from an image obtained by post-printing a pattern with ink of a predetermined color on a background that is a predetermined background color. And
For each ink that can be designated as the extraction target, an index used to draw a color distribution approximation curve of the ink in a color space, and a unique point coordinate indicating a unique coordinate of the ink that is not affected by the background color in the color space Storage means for storing a color distribution approximation curve table in which separation values used for separating the background color and the ink color on the color distribution approximation curve are defined in association with each other;
Background color coordinate determining means for determining a background color coordinate indicating a color space coordinate of a background color in the image;
The exponents and unique point coordinates corresponding to the ink designated as the extraction target are read from the color distribution approximate curve table, and the intrinsic color coordinates and the background color coordinates determined by the background color coordinate determination means are used as the base points according to the indexes. Color distribution approximation curve determining means for determining a drawn curve as a color distribution approximation curve relating to the ink;
The separation value corresponding to the ink designated as the extraction target is read from the color distribution approximation curve table, and is designated on the color distribution approximation curve determined by the color distribution approximation curve determination unit. Separation point determination means for determining a separation point between the ink pixel and the background pixel;
After determining, for each pixel, an intersection that intersects the color distribution approximation curve determined by the color distribution approximation curve determination means from each pixel of the image developed in the color space for each pixel, the intersection is the separation point A pixel pattern that is in the direction of the unique point coordinates as viewed from the separation point determined by the determination unit and that has a color space distance from the intersection point within a predetermined threshold value is designated as the extraction target. Extracting means for extracting as pixels corresponding to
An image processing apparatus comprising:

(Supplementary Note 4) For each of two learning images having different background colors on which a pattern is post-printed only by the same ink that can be post-printed, the color of each pixel constituting the pattern post-printed by the ink A first determining means for calculating each of the pattern color coordinates indicating the coordinates in space, and determining an average value or mode value of each of the calculated pattern color coordinates as the pattern color center coordinates;
The background color coordinate determining means determines the background color coordinate for each of the two images,
For each of the two images, using the pattern color center coordinates determined by the first determining means and the background color coordinates determined by the background color coordinate determining means, the eigenpoint coordinates are determined as unknowns. Second determining means for determining each of the color distribution approximation curves,
Third determination means for calculating and determining the unique point coordinates which are unknowns, assuming that the unique point coordinates which are unknown in the two color distribution approximation curves determined by the second determination means are the same. When,
Substituting the eigenpoint coordinates determined by the third determining means into the unknown, and calculating and determining the exponent;
Substituting the eigenpoint coordinates determined by the third determining means into the unknown and perpendicular to the color distribution approximation curve determined by substituting the unknown into the unknown from each pixel of the image developed in the color space A fifth determining means for calculating and determining the separation value based on the frequency of the pixel corresponding to each intersection on the color distribution approximate curve, after determining the intersection for each pixel by lowering
Further comprising
The storage means is associated with the ink that can be post-printed, the eigenpoint coordinates determined by the third determination means, the index determined by the fourth determination means, and the fifth The image processing apparatus according to appendix 3, further comprising: storing a color distribution approximate curve table defined in association with the separation value determined by the determination unit.

(Supplementary Note 5) Image processing in which a computer extracts pixels corresponding to a pattern with ink designated as an extraction target from an image obtained by post-printing a pattern with ink of a predetermined color on a background that is a predetermined background color A method,
The computer
A background color coordinate determination step for determining a background color coordinate indicating a color space coordinate of a background color in the image;
For each ink that can be designated as the extraction target, an index used to draw a color distribution approximation curve of the ink on the color space, and a unique point coordinate indicating the unique coordinates of the ink that are not affected by the background color on the color space And the color distribution approximation curve table in which the separation values used for separating the background color and the ink color on the color distribution approximation curve are defined in association with each other, the above-mentioned corresponding to the ink designated as the extraction target A color distribution that reads an exponent and a unique point coordinate and determines a curve drawn by the exponent based on the unique point coordinate and the background color coordinate determined in the background color coordinate determination step as a color distribution approximate curve related to the ink An approximate curve determination step;
The separation value corresponding to the ink designated as the extraction target is read from the color distribution approximation curve table, and designated on the color distribution approximation curve determined by the color distribution approximation curve determination step. A separation point determination step for determining a separation point between the ink pixel and the background pixel;
After determining, for each pixel, each intersection of the image developed in the color space by intersecting with the color distribution approximation curve determined by the color distribution approximation curve determination step for each pixel, the intersection is the separation point A pattern of ink specified as the extraction target is a pixel that is in the direction of the unique point coordinates as viewed from the separation point determined by the determination step and whose distance in the color space with the intersection is within a predetermined threshold. An extraction step of extracting as pixels corresponding to
An image processing method comprising:

(Supplementary Note 6) For each of the two learning images having different background colors on which the pattern is post-printed only by the same ink that can be post-printed, the color of each pixel constituting the pattern post-printed by the ink The method further includes a first determination step of calculating each of the pattern color coordinates indicating the spatial coordinates, and determining an average value or mode value of each of the calculated pattern color coordinates as the pattern color center coordinates,
The background color coordinate determination step determines the background color coordinate for each of the two images,
For each of the two images, using the pattern color center coordinates determined by the first determination step and the background color coordinates determined by the background color coordinate determination step, the eigenpoint coordinates are determined as unknowns. A second determination step for determining each of the color distribution approximation curves,
Third determination step of calculating and determining the unique point coordinates that are unknowns, assuming that the unique point coordinates that are unknowns in the two color distribution approximation curves determined by the second determination step are the same. When,
A fourth determination step of substituting the eigenpoint coordinates determined by the third determination step into the unknown and calculating and determining the exponent;
Substituting the eigenpoint coordinates determined in the third determination step into the unknowns and perpendicular to the color distribution approximation curve determined by substituting the unknowns from each pixel of the image developed in the color space A fifth determination step of calculating and determining the separation value based on the frequency of the pixel corresponding to each intersection on the color distribution approximation curve after determining the intersection intersected by lowering
Corresponding to the ink that can be post-printed, the unique point coordinates determined by the third determination step, the index determined by the fourth determination step, and the fifth determination step are determined. Storing in the color distribution approximate curve table defined in association with the separation value;
The image processing method according to appendix 5, further comprising:

  As described above, the image processing apparatus, the image processing method, and the image processing program according to the present invention are extracted as an object to be extracted from an image obtained by post-printing a pattern of ink of a predetermined color on a background that is a predetermined background color. This is useful for an image processing apparatus that extracts pixels corresponding to a specified ink pattern, and is particularly suitable for realizing the extraction of a pattern based on the ink color, not the observed color.

FIG. 6 is a diagram for explaining an outline of a color distribution approximation curve in the first embodiment. FIG. 6 is a diagram for explaining a method for calculating an ink color specific point in the first embodiment. FIG. 6 is a diagram for explaining a method for calculating an ink color specific point in the first embodiment. FIG. 6 is a diagram for illustrating an expression used for explaining a color distribution approximation curve in the first embodiment. FIG. 6 is a diagram for illustrating an expression used for explaining a color distribution approximation curve in the first embodiment. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram for explaining an overview and features of an image processing apparatus according to a first embodiment. It is a block diagram which shows the structure of an image processing apparatus. It is a figure for showing an example of the information memorized by the setting information storage part. It is a figure for demonstrating a separated value calculation part. It is a figure for demonstrating a distance and parameter calculation part. It is a figure for demonstrating an extraction determination part. It is a figure for demonstrating the method of setting a predetermined threshold value. It is a figure for demonstrating a decompression | restoration part. 6 is a flowchart illustrating a flow of processing by a learning control unit according to the first embodiment. 6 is a flowchart illustrating a flow of processing by a pattern extraction control unit according to the first embodiment. FIG. 6 is a diagram illustrating an effect of the image processing apparatus according to the first embodiment. 1 is a diagram illustrating a program of an image processing apparatus in Embodiment 1. FIG. It is a figure for demonstrating a prior art. It is a figure for demonstrating a prior art. It is a figure for demonstrating a prior art.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Input part 20 Output part 30 Storage part 31 Setting information storage part 40 Learning control part 41 Representative color calculation part 42 Color distribution approximate curve index | exponent calculation part 43 Separation value calculation part 50 Pattern extraction control part 51 Extraction object designation | designated part 52 Background designation part 53 Color distribution approximate curve creation unit 54 Distance / parameter calculation unit 55 Extraction determination unit 56 Restoration unit

Claims (6)

An image that causes a computer to execute an image processing method for extracting pixels corresponding to a pattern with ink designated as an extraction target from an image obtained by post-printing a pattern with ink of a predetermined color on a background that is a predetermined background color A processing program,
A background color coordinate determination procedure for determining a background color coordinate indicating a color space coordinate of a background color in the image;
For each ink that can be designated as the extraction target, an index used to draw a color distribution approximation curve of the ink in a color space, and a unique point coordinate indicating a unique coordinate of the ink that is not affected by the background color in the color space The index corresponding to the ink designated as the extraction target from the color distribution approximation curve table in which the separation values used for separating the background color and the ink color on the color distribution approximation curve are defined in association with each other Color distribution approximation that reads out the unique point coordinates and determines the curve drawn by the index based on the unique point coordinates and the background color coordinates determined by the background color coordinate determination procedure as a color distribution approximate curve related to the ink Curve determination procedure;
The separation value corresponding to the ink designated as the extraction target is read from the color distribution approximation curve table, and designated on the color distribution approximation curve determined by the color distribution approximation curve determination procedure. A separation point determination procedure for determining a separation point between the ink pixel and the background pixel;
After each pixel of the image developed in the color space determines an intersection for each pixel by crossing the color distribution approximation curve determined by the color distribution approximation curve determination procedure for each pixel, the intersection is the separation point A pattern with ink designated as the extraction target is a pixel that is in the direction of the unique point coordinates as viewed from the separation point determined by the determination procedure and whose distance in the color space with the intersection is within a predetermined threshold. An extraction procedure for extracting pixels corresponding to
An image processing program for causing a computer to execute. For each of two learning images having different background colors on which a pattern is post-printed only by the same ink that can be post-printed, the coordinates on the color space of each pixel constituting the pattern post-printed by the ink are calculated. Calculating each pattern color coordinate to be displayed, and causing the computer to further execute a first determination procedure for determining an average value or mode value of each calculated pattern color coordinate as a pattern color center coordinate;
The background color coordinate determination procedure determines the background color coordinate for each of the two images,
For each of the two images, using the pattern color center coordinates determined by the first determination procedure and the background color coordinates determined by the background color coordinate determination procedure, the eigenpoint coordinates are unknown. A second determination procedure for determining the respective color distribution approximation curves,
Third determination procedure for calculating and determining the eigenpoint coordinates that are unknowns, assuming that the eigenpoint coordinates that are unknowns in the two color distribution approximation curves determined by the second determination procedure are the same. When,
A fourth determination procedure for substituting the eigenpoint coordinates determined by the third determination procedure for the unknown and calculating and determining the exponent;
Substituting the eigenpoint coordinates determined by the third determination procedure into the unknown and perpendicular to the color distribution approximation curve determined by substituting the unknown into the unknown from each pixel of the image developed in the color space A fifth determination procedure for calculating and determining the separation value based on the frequency of the pixel corresponding to each intersection on the color distribution approximation curve after determining the intersection for each pixel by lowering
Corresponding to the ink that can be post-printed, the unique point coordinates determined by the third determination procedure, the index determined by the fourth determination procedure, and the fifth determination procedure are determined. A storage procedure for storing in the color distribution approximate curve table defined in association with the separation value;
The image processing program according to claim 1, further causing the computer to execute. An image processing apparatus that extracts pixels corresponding to a pattern of ink designated as an extraction target from an image obtained by post-printing a pattern of ink of a predetermined color on a background that is a predetermined background color,
For each ink that can be designated as the extraction target, an index used to draw a color distribution approximation curve of the ink in a color space, and a unique point coordinate indicating a unique coordinate of the ink that is not affected by the background color in the color space Storage means for storing a color distribution approximation curve table in which separation values used for separating the background color and the ink color on the color distribution approximation curve are defined in association with each other;
Background color coordinate determining means for determining a background color coordinate indicating a color space coordinate of a background color in the image;
The exponents and unique point coordinates corresponding to the ink designated as the extraction target are read from the color distribution approximate curve table, and the intrinsic color coordinates and the background color coordinates determined by the background color coordinate determination means are used as the base points according to the indexes. Color distribution approximation curve determining means for determining a drawn curve as a color distribution approximation curve relating to the ink;
The separation value corresponding to the ink designated as the extraction target is read from the color distribution approximation curve table, and is designated on the color distribution approximation curve determined by the color distribution approximation curve determination unit. Separation point determination means for determining a separation point between the ink pixel and the background pixel;
After determining, for each pixel, an intersection that intersects the color distribution approximation curve determined by the color distribution approximation curve determination means from each pixel of the image developed in the color space for each pixel, the intersection is the separation point A pixel pattern that is in the direction of the unique point coordinates as viewed from the separation point determined by the determination unit and that has a color space distance from the intersection point within a predetermined threshold value is designated as the extraction target. Extracting means for extracting as pixels corresponding to
An image processing apparatus comprising: For each of two learning images having different background colors on which a pattern is post-printed only by the same ink that can be post-printed, the coordinates on the color space of each pixel constituting the pattern post-printed by the ink are calculated. Further comprising first determining means for calculating each of the pattern color coordinates shown, and determining an average value or mode value of the calculated pattern color coordinates as the pattern color center coordinates,
The background color coordinate determining means determines the background color coordinate for each of the two images,
For each of the two images, using the pattern color center coordinates determined by the first determining means and the background color coordinates determined by the background color coordinate determining means, the eigenpoint coordinates are determined as unknowns. Second determining means for determining each of the color distribution approximation curves,
Third determination means for calculating and determining the unique point coordinates which are unknowns, assuming that the unique point coordinates which are unknown in the two color distribution approximation curves determined by the second determination means are the same. When,
Substituting the eigenpoint coordinates determined by the third determining means into the unknown, and calculating and determining the exponent;
Substituting the eigenpoint coordinates determined by the third determining means into the unknown and perpendicular to the color distribution approximation curve determined by substituting the unknown into the unknown from each pixel of the image developed in the color space A fifth determining means for calculating and determining the separation value based on the frequency of the pixel corresponding to each intersection on the color distribution approximate curve, after determining the intersection for each pixel by lowering
Further comprising
The storage means is associated with the ink that can be post-printed, the eigenpoint coordinates determined by the third determination means, the index determined by the fourth determination means, and the fifth The image processing apparatus according to claim 3, further comprising a color distribution approximate curve table that is defined in association with the separation value determined by the determination unit. An image processing method in which a computer extracts pixels corresponding to a pattern of ink designated as an extraction target from an image obtained by post-printing a pattern of ink of a predetermined color on a background that is a predetermined background color. ,
The computer
A background color coordinate determination step for determining a background color coordinate indicating a color space coordinate of a background color in the image;
For each ink that can be designated as the extraction target, an index used to draw a color distribution approximation curve of the ink on the color space, and a unique point coordinate indicating the unique coordinates of the ink that are not affected by the background color on the color space And the color distribution approximation curve table in which the separation values used for separating the background color and the ink color on the color distribution approximation curve are defined in association with each other, the above-mentioned corresponding to the ink designated as the extraction target A color distribution that reads an exponent and a unique point coordinate and determines a curve drawn by the exponent based on the unique point coordinate and the background color coordinate determined in the background color coordinate determination step as a color distribution approximate curve related to the ink An approximate curve determination step;
The separation value corresponding to the ink designated as the extraction target is read from the color distribution approximation curve table, and designated on the color distribution approximation curve determined by the color distribution approximation curve determination step. A separation point determination step for determining a separation point between the ink pixel and the background pixel;
After determining, for each pixel, each intersection of the image developed in the color space by intersecting with the color distribution approximation curve determined by the color distribution approximation curve determination step for each pixel, the intersection is the separation point A pattern of ink specified as the extraction target is a pixel that is in the direction of the unique point coordinates as viewed from the separation point determined by the determination step and whose distance in the color space with the intersection is within a predetermined threshold. An extraction step of extracting as pixels corresponding to
An image processing method comprising: For each of two learning images having different background colors on which a pattern is post-printed only by the same ink that can be post-printed, the coordinates on the color space of each pixel constituting the pattern post-printed by the ink are calculated. A first determination step of calculating each of the pattern color coordinates shown, and determining an average value or mode value of each of the calculated pattern color coordinates as the pattern color center coordinates;
The background color coordinate determination step determines the background color coordinate for each of the two images,
For each of the two images, using the pattern color center coordinates determined by the first determination step and the background color coordinates determined by the background color coordinate determination step, the eigenpoint coordinates are determined as unknowns. A second determination step for determining each of the color distribution approximation curves,
Third determination step of calculating and determining the unique point coordinates that are unknowns, assuming that the unique point coordinates that are unknowns in the two color distribution approximation curves determined by the second determination step are the same. When,
A fourth determination step of substituting the eigenpoint coordinates determined by the third determination step into the unknown and calculating and determining the exponent;
Substituting the eigenpoint coordinates determined in the third determination step into the unknowns and perpendicular to the color distribution approximation curve determined by substituting the unknowns from each pixel of the image developed in the color space A fifth determination step of calculating and determining the separation value based on the frequency of the pixel corresponding to each intersection on the color distribution approximation curve after determining the intersection intersected by lowering
Corresponding to the ink that can be post-printed, the unique point coordinates determined by the third determination step, the index determined by the fourth determination step, and the fifth determination step are determined. Storing in the color distribution approximate curve table defined in association with the separation value;
The image processing method according to claim 5, further comprising:

Images