KR20100077641A - Apparatus and method for image transfer using templates - Google Patents

Abstract

PURPOSE: A device and a method for image transfer using templates are provided to express the visual difference between objects comprising original video. CONSTITUTION: A regressive line determining unit(120) determines original regression line and standard regression line. A sum of distance from the original regression line to coordinate of color space corresponding to pixel color comprising an inputted original video is minimum. A sum of distance from reference regression line to coordinate on the color space corresponding to a color of pixels comprising a template is minimum. An image transformer transforms the original regression line and the original video by conversion matrix which make the line to be coincide with the standard regression line. The video transformer generates the transferred video.

Description

Apparatus and method for image transfer using templates

The present invention relates to an image conversion apparatus and method using a template, and more particularly, to an apparatus and method for converting the color distribution constituting the original image to the same as the color distribution constituting the template.

There is a method of using color contrast as a method for expressing the three-dimensional and perspective between the objects to be expressed in the two-dimensional image. This makes it easy to visualize the depth of the shape or partially overlapped objects by increasing the difference between the colors of the foreground and background, ie hue, saturation, and intensity, in a two-dimensional image that does not have depth information. To be recognized. Various methods of converting the color of the original image using the reference image have been studied to express the contrast effect of the color in the image.

One of them is a method of changing the distribution of the color of the original image to the distribution of the color of the reference image by moving the pixel data in the RGB color space using the conversion matrix. The transformation matrix is calculated using covariance and singular value decomposition (SVD). 1 is a diagram illustrating an example of converting an image in an RGB color space. Referring to FIG. 1, after the color distribution of the original image and the reference image represented in the RGB color space are represented by an ellipse, the converted result image may be obtained by matching the color distribution of the original image and the reference image. However, since the RGB color space is highly dependent on each axis of R (red), G (green), and B (blue), the resultant image can be easily obtained only when color conversions of similar colors are performed. Since the distribution data must be expressed as an ellipse, color conversion is difficult for data located outside the elliptic region.

The Munsell color wheel is a representative colorimetric system used to express the contrast effect of colors. Munsell color wheel is a method of displaying color by a mathematical method determined according to the Commission Internationale de I'Eclairage (CIE). Complementary colors facing each other in this color wheel achieve strong hue contrast when placed adjacent to each other, helping humans recognize objects.

On the other hand, the CIE L ^* a ^* b ^* color space is composed of an L ^* channel that represents only the brightness of the color in achromatic color, a ^* channel that represents red and green, and b ^* channel that represents blue and yellow. At this time, L ^* channel and a ^* b ^* plane are completely separated, so CIE L ^* a ^* b ^* color space is mainly used to change the color and saturation of the image without changing the brightness value. In addition, CIE L ^* a ^* b ^* color space is a color system that can express the color values of various color spaces. Therefore, when it is used as an intermediate medium in the color conversion process, all colors in the visible area can be preserved even during the color conversion process. There is an advantage.

CIE L ^* a ^* b ^* using the color space as an example of how to convert the colors in the original image, the way to automatically convert the color of the original image based on the number of templates generated by the CIE L ^* a ^* b ^* color space, There is this. FIG. 2 is a diagram illustrating various templates used for converting an original image in a CIE L ^* a ^* b ^* color space. The template as shown in FIG. 2 refers to a color distribution having an arbitrary straight or curved shape on a ^* b ^* plane. In this method, eight limited templates are used to convert colors of an image. However, if this method does not change the brightness and saturation, the saturation contrast cannot be expressed in the image, and color conversion is not easy between non-neighboring areas of the image. The disadvantage is that it should.

SUMMARY OF THE INVENTION The present invention has been made in an effort to provide an image conversion apparatus and method using a template capable of clearly expressing visual differences between objects constituting an original image while maintaining image quality of an original image having various colors.

Another technical problem to be solved by the present invention is a program for executing an image conversion method using a template that can clearly express visual differences between objects constituting the original image while maintaining the image quality of the original image having various colors. To provide a computer-readable recording medium for recording.

In order to achieve the above technical problem, an image conversion apparatus using a template according to the present invention generates a template represented by two-dimensional lines in the color space based on a reference color selected from a plurality of colors constituting the color space. A template generating unit; The original regression line having a minimum sum of distances from the coordinates in the color space corresponding to the colors of the pixels constituting the input original image and the distance from the coordinates in the color space corresponding to the colors of the pixels constituting the template. A regression line determining unit for determining a reference regression line having a minimum sum; And an image converter configured to generate a converted image by converting the original image by a transformation matrix to match the original regression line with the reference regression line.

In order to achieve the above technical problem, an image conversion method using a template according to the present invention includes a template represented by a two-dimensional line in the color space based on a reference color selected from a plurality of colors constituting the color space. Generating a template; The original regression line having a minimum sum of distances from the coordinates in the color space corresponding to the colors of the pixels constituting the input original image and the distance from the coordinates in the color space corresponding to the colors of the pixels constituting the template. A regression line determining step of determining a reference regression line having a minimum sum; And an image conversion step of generating the converted image by converting the original image by a transformation matrix for matching the original regression line with the reference regression line.

According to the image conversion apparatus and method using a template according to the present invention, by generating a template consisting of the reference color to convert the color of the original image, it is possible to convert the color and saturation while maintaining the image quality of the original image. In addition, the color of the original image can be easily converted by determining a regression line for the original image and the template in the two-dimensional plane and matching the original regression line with the reference regression line by the conversion matrix.

Hereinafter, exemplary embodiments of an image conversion apparatus and method using a template according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing the configuration of a preferred embodiment of an image conversion apparatus using a template according to the present invention.

Referring to FIG. 1, an image converting apparatus using a template according to the present invention includes a template generating unit 110, a regression line determining unit 120, an image converting unit 130, and a color mapping unit 140.

The template generator 110 generates a template represented by two-dimensional lines in the color space based on a reference color selected from a plurality of colors constituting the color space.

The template used for color conversion is made by considering the contrast effect of color, and is an arbitrary straight or curved color composed of several reference colors designated by a user on a specific color plane, for example, a ^* b ^* plane. Has a distribution. The color scheme between the reference colors is used to improve the visual recognition rate of the objects to be expressed in the design. For example, you can create a template that expresses complementary color contrast using colors such as red-green and yellow-blue facing each other on the Munsell color wheel described above, or use warm colors (red, orange, You can also create templates that use yellow, etc., cold colors (blue, indigo, etc.), and neutral colors (green, violet, etc.). In the present invention, a template is generated using colors that contrast with each other on a Munsell color wheel.

The process of creating a template consists of three steps. First, (1) a plurality of reference colors (hereinafter, expressed as C _i (i-1, ..., N)) to be selected are selected in consideration of color contrast. Next, (2) the selected N reference colors are arranged in the order of numbers on the Munsell color wheel, and then a gradation image is generated in the order listed. The gradation image expresses a plurality of reference colors that are disconnected from each other in a plurality of consecutive colors, and includes a plurality of colors appearing between two neighboring colors. Finally, (3) the colors used in the generated gradient image are distributed on the two-dimensional plane of the color space. If CIE L ^* a ^* b ^* color space is used, the colors are distributed in the a ^* b ^* plane. In addition, when the values of a ^* and b ^* channels are the same, and colors of different L ^* channels exist, the value of the largest L ^* channel is selected.

When the colors distributed in the two-dimensional plane are connected to each other in a straight or curved form, this corresponds to a template generated from the reference color. 2 shows an example of a reference color, a gradient image generated from the reference color, and a template shown in the a ^* b ^* plane. Since the a ^* b ^* plane of the chromatic color is complementary to the x-axis (red-green channel) and y-axis (yellow-blue channel), the shape of the template with the complementary contrast is mostly passed near the origin. In addition, since a ^* b ^* plane has a wide range of saturation values, it is possible to convert an image having a narrow range of saturation values to have a wide range of saturation values.

Further, although the reference color may be selected again to generate a new template, a template consisting of new reference colors may be generated by adjusting the position of a control point of the existing template. 3 shows an example of a new template generated by adjusting the position of the control point in the existing template. Referring to FIG. 3, a new template B is generated by changing the position of the control point from the existing template A shown in the a ^* b ^* plane. This is the same result as selecting a new reference color and creating a template.

The regression line determiner 120 includes an original regression line having a minimum sum of distances from coordinates in the color space corresponding to the colors of the pixels constituting the input original image and a color space corresponding to the colors of the pixels constituting the template. Determine the reference regression line with the smallest sum of distances from the coordinates.

In addition, the image conversion unit 130 generates the converted image by converting the original image by the conversion matrix to match the original regression line with the reference regression line.

In order to convert the color distribution in the original image into the same form as the template, a conversion matrix is required. Also, the transformation matrix is easily determined when the color distribution appears in the form of a straight line. Therefore, the regression line determiner 120 determines a regression line with respect to the original image and the template, respectively, in order to represent the distribution of colors constituting the original image and the template in a straight line. The regression line obtained by regressive analysis is a suitable straight line representing the distribution data of colors of an image, and means a straight line which minimizes the sum of distances from one point of each color data to a straight line. In addition, in order to determine the original regression line determined for the original image, the color distribution constituting the original image must be represented on the same two-dimensional plane as the template, and this is called color distribution data. When using the CIE L ^* a ^* b ^* color space, the color distribution is obtained by using the a ^* channel, which is a chromatic channel, on the x axis, and the b ^* channel, on the y axis, and having two-dimensional coordinates. In this case, when the values of a ^* and b ^* channels are the same and the colors of L ^* channels are different, the value of the largest L ^* channel is selected as in the case of creating a template.

The regression line determined from the original image and the template is determined by a parameter capable of minimizing an error represented by Equation 1 below. In Equation 1, α and β minimizing E become slopes and y-intercepts of the regression line.

Where E is the minimum squared error (MSE), (x _{i, j} , y _i, for the deviation between the original image or template consisting of m (height) x n (width) pixels and the original regression or reference regression line _{. j} ) is the coordinate value of the color appearing on the a ^* b ^* plane with respect to the pixels constituting the original image or template, α is the slope of the original regression line or the reference regression line, and β is the y-intercept of the original regression line or the reference regression line. to be. FIG. 4 illustrates the distribution data of colors constituting the original image on the a ^* b ^* plane and a regression line determined for the original image using Equation 1 above.

When both the original regression line for the original image and the reference regression line for the template are determined, the image conversion unit 130 matches the original regression line with the reference regression line for color conversion of the original image. In this case, since the original regression line and the reference regression line are located on a two-dimensional plane, a transformation matrix represented by a two-dimensional matrix is used to convert the slope of the original regression line into the slope of the reference regression line. This conversion matrix can be expressed as Equation 2 below.

Where M is the transformation matrix, S (S _x , S _y ) is the scaling matrix that matches the ratio of the distribution range of the colors of the original image and the template, and R (θ) is the rotation matrix that rotates the original regression line about the origin. , And T (-x _s , -y _s ) are shift matrices that allow the original regression line to pass through the origin.

The conversion of the original image using the transformation matrix of Equation 2 is performed by sequentially applying a moving matrix, a rotation matrix, and a scaling matrix to the distribution data of the colors of each pixel of the original image, that is, the a ^* b ^* plane. First, the moving matrix is a matrix for converting the original regression line represented by y = αx + β into a straight line passing through the origin, and is out of the range of a ^* b ^* plane when rotating or resizing the color distribution data. This is to prevent. Specifically, the moving matrix is expressed by the following Equation 3, and the straight line is moved using a smaller value among the x-intercept (-β / α) and y-intercept of the original regression line. Also, as the original regression line moves, the distribution data of the colors constituting the original image also move on the a ^* b ^* plane.

Where T (-x _s , -y _s ) is the moving matrix, α is the slope of the original regression line, and β is the y-intercept of the original regression line.

Next, the rotation matrix R (θ) rotates the original regression line, that is, the color distribution data of the original image in the clockwise or counterclockwise direction. The direction of rotation can be specified by the user, and the original regression line is rotated by the following equation (4) to match the slope of the original regression line with the slope of the reference regression line.

Where θ is the rotation angle of the original regression line, u is the slope of the reference regression line, and α is the slope of the original regression line.

Finally, the scaling matrix applied can be expressed as Equation 5 below, and is applied to match the ratio of the distribution range of the color of the original image and the template in the a ^* b ^* plane. The minimum rectangle that contains both the original image and the color distribution data of the template is called the bounding box. The scaling matrix parameter is determined by the ratio of the width and height of the bounding box for the original image to the bounding box for the template. Is determined. In addition, the scaling matrix changes the slope of the original regression line by matching the ratio of the width and height of the bounding box to the original image and the template.

Where S (S _x , S _y ) is the scaling matrix, B _t ^width is the ^width of the bounding box for the template, B _t ^height is the ^height of the bounding box for the template, and B _s ^width is the width of the bounding box for the original image , And B _s ^height are the height of the bounding box for the original image.

5 is a diagram illustrating an example of generating a converted image by sequentially applying a moving matrix, a rotating matrix, and a scaling matrix to an original image. Referring to FIG. 5, first, a moving matrix is applied to an original image so that the original regression line passes through the origin. The color distribution data of the original image also moves with the original regression image. In addition, if the original regression line has already passed the origin, it is not necessary to apply the shift matrix. Next, apply the rotation matrix to rotate the original regression line around the origin and apply the scaling matrix to match the slope of the original regression line with the slope of the reference regression line. At this time, if the inclination difference (S) between the reference regression line and the original regression line is larger than the preset error value (Error), the above process is repeated. The error value depends on the user's setting, and may preferably have a value of 0.1. When the difference between two regression lines is larger than the error value and the transform matrix is repeatedly applied, the moving matrix, the rotating matrix, and the scaling matrix are determined differently each time according to the converted original image at the time of applying the transform matrix.

In the transformed image generated by applying the transform matrix to the original image, the color distribution is almost the same as that of the template. However, in order to completely match the color distribution of the converted image with the template, it is necessary to replace data that does not correspond to the color distribution data of the template among the color distribution data of the converted image. To this end, the image conversion apparatus using the template according to the present invention further includes a color mapping unit 140.

The color mapping unit 140 replaces the replacement target pixel that does not correspond to the color of the pixels constituting the template among the pixels constituting the converted image with one of the pixels constituting the template.

The color distribution data, that is, the replacement process of the replacement pixel of the original image is performed using Euclidean norm as shown in Equation 6 below in the a ^* b ^* plane. That is, according to Equation 6, the distribution data of the color of the converted image is replaced with the distribution data of the color of the template having the minimum Euclidean norm value, so that the distribution of the colors of the two images can be completely matched.

Here, C (p) is distribution data of colors of the converted image to be replaced, and T (i) is distribution data of colors of the template.

FIG. 6 is a diagram illustrating an example of a converted image in which a replacement target pixel is replaced by the color mapping unit 140. Referring to FIG. 6, the color distribution of the transformed image that is identical to the regression line determined for the template is the same as that of the template that is represented by the curve of the color replacement.

7 is a flowchart illustrating a process of performing a preferred embodiment of the image conversion method using a template according to the present invention.

Referring to FIG. 7, the template generator 110 generates a template represented by two-dimensional lines in a color space based on a reference color selected from a plurality of colors constituting the color space (S710). In this case, the template generation process is performed by selecting a reference color and generating a gradient image based on the reference color. Next, the regression line determiner 120 corresponds to the original regression line having a minimum sum of distances from coordinates in the color space corresponding to the colors of the pixels constituting the input original image and the color corresponding to the colors of the pixels constituting the template. A reference regression line having a minimum sum of distances from spatial coordinates is determined (S720). The original regression line and the reference regression line are determined by calculating the slope and y-intercept to minimize the error of Equation 1.

The image conversion unit 130 generates a converted image by converting the original image by a transformation matrix to match the original regression line with the reference regression line (S730). The transformation matrix consists of a movement matrix, a rotation matrix, and a scaling matrix. If the difference between the original regression line and the reference regression line is greater than a preset error value, the transformation matrix is repeatedly applied. Finally, the color mapping unit 140 replaces the replacement target pixel that does not correspond to the color of the pixels constituting the template among the pixels constituting the converted image with one of the pixels constituting the template (S740).

8 is a diagram illustrating an image converted by applying an image conversion apparatus and method using a template according to the present invention. 8, and color that make up the original image (A) are indicated by the distribution data of the color on the a ^* b ^* plane, a regression line is determined for the original image is represented in the a ^* b ^* plane (B). (C) shows the resultant image converted by the image converting apparatus using the template, and the distribution data of the colors showing the colors constituting the resultant image on the a ^* b ^* plane are represented by the same curve as the template.

9 is a diagram illustrating a result image generated by applying a template corresponding to each gradation image to the gradation image and the original image generated from the reference color. Referring to FIG. 9, the difference in color contrast that appears when different complementary color contrast templates are applied to the original image may be confirmed.

As mentioned above, a new template can be created by moving the control point position in the generated template. FIG. 10 is a diagram illustrating a result of applying a new template generated by moving a control point from an existing template to the original image of FIG. 9. The image conversion apparatus using the template according to the present invention can easily adjust the color and saturation of the image simply by modifying the shape of the template on the two-dimensional plane.

When white is added to the complementary contrast-based template, the range of applicable saturation is wider than that of the template without white, thereby improving the image quality. 11 is a diagram illustrating an original image that is variously converted according to whether or not white is included in a template. Referring to FIG. 11, when white is added to the complementary color contrast-based template, an image having improved brightness and saturation contrast can be obtained.

In the present invention, a gradation image is generated by randomly selecting a plurality of reference colors to generate a template. However, a template may be generated from an arbitrary image instead of a gradient image. 12 is a diagram illustrating a result of converting an original image using a template generated from an arbitrary reference image. Referring to FIG. 12, a resultant image D obtained by applying a template generated from the reference image B to the original image A is obtained. Here, the resultant image (C) transformed by the conventional method is obtained by using the mean and variance data of each color axis to convert between the representative color set of the original image and the representative color set of the reference image in L ^* a ^* b ^* color space. It is a video. Even when the same reference image is used from the (C) and (D) images, the converted image appears differently depending on the applied method. As described above, a technique of converting an image using a reference image may be usefully used to express the color contrast effect expressed in an arbitrary artwork on the image.

The color conversion according to the present invention is better shown in an image with a small number of colors used like cartoons. For example, an image created using a cartoon rendering algorithm needs to be re-rendered by resetting various parameters that determine color during color conversion. However, using the image conversion apparatus and method of the present invention can reduce this inconvenience. 13 is a diagram illustrating an image in which colors are converted by the image conversion apparatus according to the present invention.

According to the image converting apparatus and method according to the present invention, it is possible to naturally convert the hue and saturation of the original image using the template, and simply generate a new template by moving the control points of the template. The image conversion apparatus and method using such a template can be extended to the field of segmentation by simplifying a complex image using many colors.

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like, and may be implemented in the form of a carrier wave (for example, transmission via the Internet) . The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

Although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific preferred embodiments described above, and the present invention belongs to the present invention without departing from the gist of the present invention as claimed in the claims. Various modifications can be made by those skilled in the art, and such changes are within the scope of the claims.

1 is a diagram illustrating an example of converting an image in an RGB color space;

2 is a diagram illustrating various templates used for converting an original image in a CIE L ^* a ^* b ^* color space;

3 is a block diagram showing the configuration of a preferred embodiment of an image conversion apparatus using a template according to the present invention;

4 is a diagram showing an example of a template shown in a reference color, a gradient image generated from the reference color, and a ^* b ^* plane;

5 is a diagram showing an example of a new template generated by adjusting the position of the control point in the existing template,

FIG. 6 is a diagram showing distribution data of colors constituting the original image shown in a ^* b ^* plane and a regression line determined for the original image; FIG.

7 illustrates an example of generating a converted image by sequentially applying a moving matrix, a rotating matrix, and a scaling matrix to an original regression line;

8 is a diagram illustrating an example of a converted image in which a replacement target pixel is replaced by the color mapping unit 140;

9 is a flowchart illustrating a process of performing a preferred embodiment of an image conversion method using a template according to the present invention;

10 is a diagram illustrating an image converted by applying an image conversion apparatus and method using a template according to the present invention;

FIG. 11 is a diagram illustrating a result image generated by applying a template corresponding to each gradient image to a gradient image and an original image generated from a reference color; FIG.

FIG. 12 is a diagram illustrating a result of applying a new template generated by moving a control point from an existing template to the original image of FIG. 9;

FIG. 13 is a view showing an original image which is variously converted according to whether a template includes white color;

14 is a diagram illustrating a result of converting an original image using a template generated from an arbitrary reference image, and

15 is a diagram illustrating an image in which colors are converted by the image conversion apparatus according to the present invention.

Claims (21)

A template generator for generating a template represented by two-dimensional lines in the color space based on a reference color selected from a plurality of colors constituting a color space; The original regression line having a minimum sum of distances from the coordinates in the color space corresponding to the colors of the pixels constituting the input original image and the distance from the coordinates in the color space corresponding to the colors of the pixels constituting the template. A regression line determining unit for determining a reference regression line having a minimum sum; And And an image converter configured to generate the converted image by converting the original image by a transformation matrix to match the original regression line with the reference regression line. The method of claim 1, The image using the template further comprises a color mapping unit for replacing the replacement pixel that does not correspond to the color of the pixels constituting the template among the pixels constituting the converted image to one of the pixels constituting the template. Inverter. The method according to claim 1 or 2, The reference color is an image conversion apparatus using a template, characterized in that selected in contrast colors in the Munsell color wheel. The method according to claim 1 or 2, And the image conversion unit applies the conversion matrix when the difference between the slope of the reference regression line and the original regression line is greater than a preset error value. The method according to claim 1 or 2, The original regression line and the reference regression line are determined by a parameter that minimizes the following equation A. Equation A

Where E is the minimum squared error (MSE) for the deviation between the original image and the original regression line or the template and the reference regression line consisting of m (height) x n (width) pixels, (x _{i, j} , y _{i, j} ) is the coordinate value of the color appearing on the a ^* b ^* plane with respect to the pixels constituting the original image or template, α is the slope of the original regression line or reference regression line, and β is the original regression line or reference Y-intercept of the regression line. The method according to claim 1 or 2, The image conversion unit using the template, characterized in that for generating the converted image by the conversion matrix represented by the following equation B: Equation B

Here, M is the transformation matrix, S (S _x , S _y ) is a scaling matrix that matches the ratio of the distribution range of colors of the original image and the template, and R (θ) rotates the original regression line about an origin. The rotation matrix, and T (-x _s , -y _s ) is a moving matrix for passing the original regression line through the origin. The method of claim 6, The moving matrix is an image conversion apparatus using a template, characterized in that represented by the following equation C: Equation C

Here, T (-x _s , -y _s ) is the moving matrix, α is the slope of the original regression line, and β is the y-intercept of the original regression line. The method of claim 6, The rotation matrix is an image conversion apparatus using a template, characterized in that for rotating the original regression line by the rotation angle shown in the following equation (D): [Equation D]

Is the rotation angle of the original regression line, u is the slope of the reference regression line, and α is the slope of the original regression line. The method of claim 6, The scaling matrix is a template characterized in that represented by the following equation E by a bounding box of a minimum rectangle each containing the distribution data of the color constituting the template and the original image on a two-dimensional plane in the color space Video conversion device using: [Equation E]

Here, S (S _x , S _y ) is the scaling matrix, B _t ^width is the ^width of the bounding box for the template, B _t ^height is the ^height of the bounding box for the template, and B _s ^width is for the original image. The width of the bounding box, and B _s ^height is the ^height of the bounding box for the original image. The method according to claim 1 or 2, And the color mapping unit replaces the replaceable pixel with a pixel having a minimum Euclidean distance from the replaceable pixel in the color space among pixels constituting the template. A template generation step of generating a template represented by two-dimensional lines in the color space based on a reference color selected from a plurality of colors constituting the color space; The original regression line having a minimum sum of distances from the coordinates in the color space corresponding to the colors of the pixels constituting the input original image and the distance from the coordinates in the color space corresponding to the colors of the pixels constituting the template. A regression line determining step of determining a reference regression line having a minimum sum; And And an image conversion step of generating the converted image by converting the original image by a conversion matrix to match the original regression line with the reference regression line. The method of claim 11, The method may further include a color mapping step of replacing a target pixel that does not correspond to the colors of the pixels constituting the template among the pixels constituting the converted image with one of the pixels constituting the template. Image conversion method. The method of claim 11 or 12, The reference color is an image conversion method using a template, characterized in that selected in contrast colors in the Munsell color wheel. The method of claim 11 or 12, In the image conversion step, if the difference between the slope of the reference regression line and the original regression line is larger than a preset error value, the image conversion method using a template, characterized in that for applying the conversion matrix. The method of claim 11 or 12, The original regression line and the reference regression line are the image conversion method using a template, characterized in that determined by the parameter to minimize the following equation (A): Equation A

Where E is the minimum squared error (MSE) for the deviation between the original image and the original regression line or the template and the reference regression line consisting of m (height) x n (width) pixels, (x _{i, j} , y _{i, j} ) is the coordinate value of the color appearing on the a ^* b ^* plane with respect to the pixels constituting the original image or template, α is the slope of the original regression line or reference regression line, and β is the original regression line or reference Y-intercept of the regression line. The method of claim 11 or 12, In the image conversion step, the image conversion method using a template, characterized in that for generating the converted image by the conversion matrix represented by the following equation (B): Equation B

Here, M is the transformation matrix, S (S _x , S _y ) is a scaling matrix that matches the ratio of the distribution range of colors of the original image and the template, and R (θ) rotates the original regression line about an origin. The rotation matrix, and T (-x _s , -y _s ) is a moving matrix for passing the original regression line through the origin. The method of claim 16, The moving matrix is an image conversion method using a template, characterized by the following equation C: Equation C

Here, T (-x _s , -y _s ) is the moving matrix, α is the slope of the original regression line, and β is the y-intercept of the original regression line. The method of claim 16, The rotation matrix is an image conversion method using a template, characterized in that for rotating the original regression line by the rotation angle shown in the following equation (D): [Equation D]

Is the rotation angle of the original regression line, u is the slope of the reference regression line, and α is the slope of the original regression line. The method of claim 16, The scaling matrix is a template characterized in that represented by the following equation E by a bounding box of a minimum rectangle each containing the distribution data of the color constituting the template and the original image on a two-dimensional plane in the color space Image conversion method using: [Equation E]

Here, S (S _x , S _y ) is the scaling matrix, B _t ^width is the ^width of the bounding box for the template, B _t ^height is the ^height of the bounding box for the template, and B _s ^width is for the original image. The width of the bounding box, and B _s ^height is the ^height of the bounding box for the original image. The method of claim 11 or 12, In the color mapping step, an image conversion using a template, wherein the replacement target pixel is replaced with a pixel having a minimum Euclidean distance with the replacement target pixel in the color space among pixels constituting the template. Way. A computer-readable recording medium having recorded thereon a program for executing an image conversion method using a template according to claim 11 or 12.

Images