KR101195917B1 - Acquisition method of Tongue Diagnosis Region - Google Patents

Abstract

A method for extracting a tongue diagnosis region is disclosed. The method of extracting a tongue diagnosis region may include a first step of providing a tongue image implemented with RGB color coordinates externally, and LAB color coordinates including color coordinates of the tongue image including a first color region, a second color region, and a third color region. A second step of converting to and performing a histogram equalization process on the LAB color coordinates, a third step of extracting a tongue outline point and a tongue outline using the leveled LAB color coordinates, and the tongue outline point and the tongue Sampling a contour line and restoring to a smoothed tongue boundary surface, wherein the first color region is a contrast region, the second color region is a red and green complementary region, and the third region. Tongue diagnostic region extraction method characterized in that the complementary region of yellow and blue.

Description

Acquisition method of Tongue Diagnosis Region

The present invention relates to a method for diagnosing oriental medicine, and more particularly, to a method for extracting a tongue diagnosis region.

In oriental medicine, photography is called as ruin, questionnaire, and laps to diagnose using the senses. Dual diagnosis is a method of reporting and diagnosis, which is important in the diagnosis of oriental medicine because much of the information is taken visually.

Seoljin is one of these ruined methods to visually observe and diagnose the tongue.

There are two major categories of seoljin, the shape of the tongue and the color of the tongue is an important factor. The shape of the tongue obtains the tooth marks, the cracks on the outer surface of the tongue, the cracks and the volume of the tongue, and the color of the tongue obtains the color distribution of the tongues and tongues by obtaining the distribution of tongues and tongues in the tongue.

However, since the diagnosis results vary depending on the condition of the oriental medical doctor and the light source conditions around the psychological factors, standardization of the diagnosis is very urgent. This standardization requires a device that can quantitatively analyze the tongue, which consists of a light source, a camera, and software that analyzes images captured by the camera.

More importantly, it is necessary to configure the darkroom so that it is not exposed to other light around it. It is also necessary to ensure that the light source is not reflected from the tongue. A device has been developed to solve this problem, and it is necessary to analyze the acquired images. The most necessary thing to analyze the tongue is to correctly extract the diagnosis region of the tongue.

The problem to be solved by the present invention is to improve the method of extracting the tongue diagnosis area that changes according to the change in the lighting conditions to distinguish the appearance of the tongue and the surrounding environment to identify a new tongue diagnosis area that can accurately identify the presence of the scars or cracks, etc. It is to provide an extraction method.

In accordance with an aspect of the present invention, there is provided a method for extracting a tongue diagnosis region according to an embodiment of the present invention. The method includes: a first step of providing a tongue image implemented with RGB color coordinates; A second step of converting the LAB color coordinates including a color region and a third color region, and performing a histogram equalization process on the LAB color coordinates; A third step of extracting and a fourth step of sampling the tongue contour point and the tongue contour line and restoring the smoothed tongue boundary surface, wherein the first color region is a contrast region, and the second color region is Complementary red and green complementary regions, and the third region is characterized in that the complementary yellow and blue regions.

The extracting of the tongue outline point of the second step may include obtaining an x-axis center point of the tongue on the LAB color coordinates, and obtaining a point where the throat starts using the x-axis center point of the tongue on the LAB color coordinates. And obtaining a point at which the tongue starts on the LAB color coordinates, obtaining a boundary point above the tongue on the LAB color coordinates, and obtaining the tongue left and right points on the LAB color coordinates.

In the obtaining of the x-axis center point of the tongue, Equation 1 and Equation 2 are used in the first color area.

The point where the throat starts using the x-axis center point of the tongue on the LAB color coordinate is the point where the contrast value of the throat is the smallest, and the point where the interface between the lips and the throat is the largest. It features.

The starting point of the tongue on the LAB color coordinates is ΔA (xcent, y) = A (xcent, y + t) -A (xcent, yt) using Equation 3, wherein A The change in value is represented using the largest value.

The upper boundary point of the tongue on the LAB color coordinates moves from the lower point of the tongue to the lower point of the lip to find the upper point of the tongue, and the tongue value is equal to or smaller than the contrast value of the lower point of the lip and the lower point of the lip. Use as a disadvantage.

Obtaining the tongue left and right points on the LAB color coordinates includes searching for a center point of the tongue, searching for a vertical boundary point of the tongue, and searching for a diagonal boundary point of the tongue.

The searching of the vertical boundary point of the tongue may be a step of finding a value at which the boundary of the A value is maximized by moving each of the left and right sides from the center point of the tongue.

The step of searching for the diagonal boundary is a step of finding a value at which the boundary of the A value is maximized by moving each of the left and right sides from the center point of the tongue, and the contrast value of the tongue surface is a predetermined value or more. Is considered to be not a tongue boundary.

The extracting of the tongue outline of the third step may include: a reference region division step of dividing the area of the tongue into at least four areas based on the tongue center point, a horizontal division extraction step of dividing the tongue horizontally; And a vertical boundary extraction step of vertically dividing the tongue.

The fourth step is characterized in that the smoothing using a Catmull-Rom Spline curve.

According to the present invention, the boundary surface is correctly detected in all three cases of an image photographed in a bright place, a dark image in which the illumination condition is changed, and a blurring image of the movement of the tongue. This is because in dark places or in motion, the boundary of the coordinates of A and L in the LAB coordinates is improved by the histogram equalization, so that the performance of tongue detection is not degraded.

This algorithm captures the structural features of the tongue and face to detect the tongue. The upper part is extracted using the interface between the throat and the tongue, and the lower part is extracted using the color difference.

The future direction of research is to distinguish the shape of the outer tongue and the environment to accurately identify the scars or cracks to determine the presence of disease. Such a system will evolve into a u-health system that can diagnose health.

1 is a flowchart illustrating a method of extracting a tongue diagnosis region according to an exemplary embodiment of the present invention.
FIG. 2 is a flowchart for explaining in detail the step of obtaining the tongue left and right points in the contour point search step shown in FIG. 1.
3 is a flowchart illustrating the outline extracting step illustrated in FIG. 1 in more detail.
4 is an exemplary view illustrating a plurality of regions divided from a center point to obtain a contour of a tongue according to an exemplary embodiment of the present invention.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described in detail with reference to the preferred embodiments of the present invention with reference to the accompanying drawings.

1 is a flowchart illustrating a method of extracting a tongue diagnosis region according to an exemplary embodiment of the present invention.

As shown in FIG. 1, the method for extracting a tongue diagnosis region 10 may include first steps S10 to fourth steps S50.

The first step S10 may be a step of providing a tongue image implemented with an external RGB color coordinate.

In a second step S20, color coordinates of the tongue image are converted into LAB color coordinates including a first color coordinate region L, a second color coordinate region A, and a third color coordinate region B. The histogram equalization processing may be performed on the LAB color coordinates.

The third step S40 may be a step of extracting the tongue outline points and the tongue outline lines using the leveled LAB color coordinates.

The fourth step S50 may be a step of restoring the tongue contour point and the tongue contour line by sampling and combining the tongue contour point and the tongue contour line.

The first color coordinate region L represents a light and dark region, the second color coordinate region A represents a red and green complementary color region, and the third color coordinate region B represents a yellow and blue complementary color region. .

More specifically, the first step S10 may be a step of providing a tongue coordinate image based on an RGB color (eg, red, green, and blue).

The second step S20 converts RGB color coordinates of the image acquired in the first step S10 into LAB color coordinates, and histogram equalization processing for leveling of each color distributed on the converted LAB color coordinates. processing may be performed to increase the clarity of the boundary by enlarging the color distribution of the tongue boundary surface in each color.

In this case, the first color coordinate region (eg, the contrast region L) and the second color coordinate region (eg, the region A of the red system) are mostly distributed in the tongue region represented by the LAB color coordinates. Changes in the values of the coordinate region L and the second color coordinate region A may be very important.

The third step S40 may be a step of extracting the tongue outline points and the tongue outline lines using the leveled LAB color coordinates.

At this time, extracting the tongue contour point of the third step (S40) (S30) is the step of searching the x-axis center point of the tongue on the LAB color coordinates (S31), using the x-axis center point of the tongue on the LAB color coordinates Obtaining a point where the throat begins, obtaining a point at which the tongue starts on the LAB color coordinates, obtaining a boundary point above the tongue on the LAB color coordinates, and obtaining the tongue left and right points on the LAB color coordinates Steps.

More specifically, the x-axis center point search step S31 of the tongue may be obtained by using Equations 1 and 2 shown below in the first color coordinate area L. FIG.

[Equation 1]

Here, L represents a value of the first color coordinate region, t represents an arbitrary integer, and d (x) represents an x-axis point at which the value of the first color coordinate region L is minimum according to the left and right values of the tongue. .

&Quot; (2) "

For example, x _cent represents a point on the x-axis having the smallest value among the points at which the value of the first color coordinate area L becomes the minimum.

Since most tongues are symmetrical, first, the position of the x-axis where the difference in the symmetry of the tongue is minimized is calculated by using Equation 2 in the first color coordinate region L as shown in Equation 1 below.

Next, the step of obtaining the point where the throat starts is a step of finding a point having the smallest value of the first color coordinate region L of the throat using the x-axis center point of the tongue on the LAB color coordinate. It is a step of obtaining a point ΔA representing a value at which the boundary surface of the lips and the throat y _below is the largest. For reference, the point where the throat starts can be found by using Equation 3 below.

&Quot; (3) "

[Delta] A denotes a value at which the second color coordinate region value becomes smallest according to an arbitrary t (e.g., t is an integer) on the y axis.

Next, obtaining the boundary point of the upper tongue on the LAB color coordinates to find the upper point of the tongue by moving from the lower point of the tongue to the lower point of the lips, the value of the first color coordinate region (L) of the lower point of the lips and the It may be a step of obtaining a value having a lower or lower contrast value of the point under the lip, and the value may be expressed as a tongue point.

In this case, the location point under the tongue may be randomly found at the position where the boundary between the value of the second color coordinate region A including the red component of the tongue and the facial skin is the largest.

FIG. 2 is a flowchart for explaining in detail the step of obtaining the tongue left and right points in the contour point search step shown in FIG. 1.

Next, as shown in FIG. 2, the step of obtaining the left and right tongue points on the LAB color coordinates may include searching for a center point of the tongue (S31), searching for a vertical boundary point of the tongue (S32), and the tongue. Retrieving a horizontal boundary point of (S33) and retrieving a diagonal boundary point of the tongue (S35).

The order of searching for the vertical boundary point of the tongue (S32) and the searching of the horizontal boundary point of the tongue (S33) may be changeable.

The searching of the center point of the tongue uses Equation 4 shown below.

&Quot; (4) "

For example, x _cent represents a point on the x-axis having the smallest value among the points at which the first color coordinate region L becomes the minimum, and y _cent represents a point _{above the} y-axis and a point below the y-axis. (y _below ) indicates a point with a median.

Searching for the vertical boundary point of the tongue (S32) is a point (x _left , y _cent ) that the boundary of the value of the first color coordinate region (L) is the maximum while moving each of the left and right from the center point of the tongue, It may be a step of finding (x _right , y _cent ).

In the searching of the diagonal boundary point (S35), reflection is generated on the surface of the tongue by illumination injected from the outside, where a saturation state of the value of the first color coordinate region L may occur, thereby increasing the boundary surface. have.

Accordingly, in order to suppress such a saturation state, each of the left and right sides may be moved from the center point of the tongue to find a value at which the boundary of the first color coordinate region L is maximized.

In this case, when the value of the first color coordinate region L on the surface of the tongue is greater than or equal to a predetermined value, it is preferable that the value of the tongue is not regarded as the tongue boundary value.

3 is a flowchart illustrating the contour extraction step illustrated in FIG. 1 in more detail, and FIG. 4 is an exemplary view illustrating a plurality of regions divided from a center point to obtain a contour of a tongue according to an exemplary embodiment of the present invention.

As shown in Figure 3, extracting the tongue contour (S35) is a reference region division step (S36) for dividing the region of the tongue into at least four or more regions based on the tongue center point, the tongue vertically Vertical division extraction step S37 for dividing and horizontal boundary extraction step S38 for dividing the tongue horizontally.

More specifically, the reference region dividing step (S36) may be a step of dividing the contour into four regions at 90 ° angles based on the center point of the tongue to obtain different contours, wherein the characteristics of the color extracted from each region are different from each other. Can be.

That is, referring to FIG. 4, the shape of the tongue is divided into four convex ranges, where Ul means Upper left, Ur means Upper right, and Ll Lower. left), and Lr means lower right.

First, to find the outline above the center of the tongue, the periphery of the tongue is ambiguous because the border of the tongue is near the throat, so find the minimum value in the y-axis direction in the transverse direction of the tongue, and then use the threshold value (T _lip). It is desirable to determine

Next, proceed from the central point of the tongue in the vertical direction, the first position of the first color coordinate area (L) is smaller than the threshold (T _lip ) as a boundary to determine the upper boundary of the center point of the tongue do.

It is desirable to find this process where the throat is present and to proceed this process at an angle of 45 ° in each area.

The reason for this is that a different method is required in areas other than the throat. Therefore, the position of the minimum value of the first color coordinate region L is found from 45 ° of each region. At this time, a local minimum within a predetermined range is obtained from the boundary surface found at the 45 ° position. The predetermined range may use Equation 5 shown below.

[Equation 5]

Where a is the previous x-axis value, and this is repeated up to 90 ° to find the x-axis value. width represents the boundary.

This process is applied together to the upper boundary of the left and right, and for reference, 20 described in Equation 5 may be any value.

Next, the vertical boundary extraction step (S37) may be a step of finding a boundary at which the difference of the A value is maximum within a predetermined range in the 45 ° vertical direction from the point below the tongue in the region below the tongue using Equation 6 have.

&Quot; (6) "

For reference, b denotes a previous y-coordinate, and height indicates a height from a lower point of the tongue to a boundary where the difference of the second color coordinate area value A becomes maximum.

That is, the vertical boundary extracting step S37 is a step of finding a difference between the tongue and the red color (eg, the value of the second color coordinate area A) between the tongue and the skin. Indicates the boundary where the A value of the direction) becomes the maximum.

Do this for each of the left and right borders. For reference, 20 described in Equation 6 may be any value.

Next, referring to FIG. 1, in the tongue image restoration step S50, after performing a sampling process using the tongue contour point and the tongue limbal line, a Catmull-Rom Spline curve is generated. It may be a step of smoothing process by using.

More specifically, referring to FIG. 3, the points extracted from four regions (first to fourth regions) may have rough curves due to noise generated in the tongue and its surroundings.

Therefore, points extracted for four areas (first to fourth areas) to remove the noise are extracted from each area according to Equations 7 and 8 described below (N _UI , N _Ur , Sampling is performed by dividing the data into 40 points according to the number of N _Ll and N _Lr ). The distribution to the 40 points may be for extracting sampling points of each region.

Subsequently, a Catmull-Rom Spline curve is obtained according to the number of points distributed, and a smoothing process is performed.

A detailed description of the Catmull-Rom Spline curve is well known and thus will be omitted.

[Equation 7]

[Equation 8]

Therefore, the step of restoring the tongue image (S50) restores the actual tongue shape by connecting each tongue contour point (S _Ur , S _Ul , S _Ll , S _Lr ) sampled based on 40 points. Can be.

For reference, each of the at least four regions (first to fourth regions) divided from the tongue center point may be divided into eight regions at 45 ° intervals again, and find an area where the error (e) is expected to be large. A feedback process can be performed to review the area again.

The reason for performing this feedback process is to find the scars or complex boundaries occurring in the region where the error (e) is expected to be large. The error (e) is obtained by comparing each of the regions divided into the eight regions. You can find it. Equation 9 described below may be an equation for finding an error e through a relative comparison of eight regions.

&Quot; (9) "

Where R is red, G is green, B is blue, (x, t) is a vector coordinate, and Bd (x) and S (x) are the boundaries before and after smoothing, respectively. Coordinate change before and after an interface is shown, and (alpha), (beta), (gamma) shows arbitrary values.

Therefore, according to the present invention, each boundary plane may be improved by histogram equalization of the coordinate value of the second color region and the coordinate value of the first color region relative to the LAB coordinates in a dark place or motion.

The tongue extraction method algorithm proposed in the present invention helps to detect the tongue by capturing the structural features of the tongue and face well, extracting the upper part of the center of the tongue using the interface between the throat and the tongue, and the lower part of the center of the tongue. It may be characterized by the extraction using the difference.

In addition, it is possible to determine the presence of the disease more easily by accurately identifying the scars or cracks by separating the shape and environment of the outer tongue.

Although the present invention has been described with reference to the embodiments illustrated in the drawings, this is merely exemplary, and it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

S10: Tongue image input S21: Color change
S22: Histogram leveling S30: Contour point extraction
S35: Contour Extraction S50: Tongue Image Restoration by Smoothing Technique

Claims (13)

A first step of providing a tongue image implemented in the RGB color coordinates from the outside;
A second step of converting color coordinates of the tongue image into LAB color coordinates including a first color region, a second color region, and a third color region, and performing a histogram equalization process on the LAB color coordinates;
Extracting tongue outline points and tongue outlines using the leveled LAB color coordinates; And
And a fourth step of sampling the tongue contour point and the tongue contour line and restoring the smoothed tongue boundary surface.
The first color region is a contrast region, the second color region is a red and green complementary color region, the third color region is a yellow and blue complementary color region extraction method of the tongue.
The method of claim 1,
Extracting the tongue contour point of the third step,
Obtaining an x-axis center point of the tongue on the LAB color coordinates;
Obtaining a starting point of the throat using the x-axis center point of the tongue on the LAB color coordinates;
Finding a point at which the tongue on the LAB color coordinates starts;
Obtaining a boundary point above the tongue on the LAB color coordinates; And
And obtaining the left and right dots of the tongue on the LAB color coordinates.
The method of claim 2,
Obtaining the x-axis center point of the tongue,
Equation 1 and Equation 2 are used in the first color coordinate region, and Equation 1 is

Is,
Equation 2 is

Where L represents a first color coordinate region value, t represents an arbitrary integer, and d (x) is the x axis of which the first color coordinate region L value becomes the minimum according to the left and right values of the tongue. And x _cent represents a point on the x-axis having the smallest value among the points where the first color coordinate area (L) value is the minimum.
The method of claim 2,
The starting point of the throat using the x-axis center point of the tongue on the LAB color coordinates,
And a point at which the contrast value of the throat is the smallest value and a point representing the value at which the boundary between the lips and the throat is the largest.
The method according to claim 3,
The starting point of the tongue on the LAB color coordinates,
ΔA (x _cent , y) is maximized using Equation 3, _ΔA (x _cent , y) = A (x _cent , y + t) -A (x _cent , yt), A is a tongue diagnosis region extraction method representing a change value of the second color coordinate region.
The method of claim 2,
The boundary point above the tongue on the LAB color coordinate is
A method for extracting a tongue diagnosis region by moving from a point below the tongue to a point below the lip to find a point on the tongue, and using a value less than or equal to the contrast value of the point below the lip and the contrast value of the point below the lip as the tongue point.
The method of claim 2,
Obtaining the left and right points of the tongue on the LAB color coordinates,
Retrieving a center point of the tongue;
Retrieving a vertical boundary point of the tongue;
Searching for a horizontal boundary point of the tongue; And
Tongue diagnostic region extraction method comprising the step of searching the diagonal boundary of the tongue.
The method of claim 2,
Obtaining a boundary point above the tongue,
And finding the difference between the boundary between the value of the second color coordinate region including the red component of the tongue and the facial skin.
The method of claim 7, wherein
Searching for the vertical boundary point of the tongue,
And a step of finding a value at which a boundary of the second color coordinate area value is maximized by moving left and right sides from the center point of the tongue, respectively.
The method of claim 7, wherein
Searching for the diagonal boundary point,
And moving each of the left and right sides from the center point of the tongue to find a value at which the boundary of the second color coordinate region value is maximum.
If the first color coordinate area value of the surface of the tongue is greater than a predetermined value it is considered that the tongue boundary value is not the tongue diagnosis region extraction method.
The method of claim 1,
Extracting the tongue outline of the third step,
A reference region dividing step of dividing the region of the tongue into at least four regions based on the tongue center point;
A vertical boundary extraction step of vertically dividing the tongue center point; And
And a horizontal division extraction step of horizontally dividing the tongue center point horizontally.
The method of claim 11,
The horizontal division extraction step and the vertical boundary extraction step are fed back to each other, characterized in that the feedback.
The method of claim 1,
The fourth step is a method for extracting a tongue diagnosis region, characterized in that for performing a smoothing process using a Catmull-Rom Spline curve.

Images