KR20210109766A - Artificial tooth color reproduction device using Deep Learning - Google Patents

Abstract

Disclosed is a color reproduction device of an artificial tooth using deep learning, which is capable of shortening the time for manufacturing an artificial tooth. The color reproduction device of an artificial tooth using deep learning according to the present invention comprises: an image processing unit which receives an image of a natural tooth for color reproduction of an artificial tooth, and obtains a reference image through pre-processing of the received image; a color distribution information generation unit which extracts a part or the entirety of pixel information with respect to a reference image, generates a pixel information matrix for the pixel information, and classifies pixels of the pixel information matrix according to a similarity algorithm so as to group the reference image into at least two regions and generates color distribution information with respect to the reference image; and a pigment spraying unit which selects a pigment to be applied or penetrating to the artificial tooth according artificial tooth material information and the color distribution information, and sprays the selected pigment on a part or the entirety of a surface of the artificial tooth.

Description

Artificial tooth color reproduction device using Deep Learning

The present invention relates to a color reproduction apparatus for artificial teeth using deep learning, and more particularly, color information is obtained from natural teeth by a color information acquisition algorithm based on deep learning, and based on the obtained color information, it is applied to artificial teeth. It relates to a color reproduction device for artificial teeth using deep learning that reproduces colors by applying pigments.

In the process of manufacturing artificial teeth, a material that has recently been in the spotlight is zirconia. Zirconia has been widely used as a material for artificial teeth in recent years because its overall volume shrinkage rate is low even in high-temperature heat treatment.

However, since zirconia absorbs pigments in the process of coloring, the so-called coloring process, in order to reproduce the target color, it was necessary to compare the difference between the result of heat treatment and the reference photograph with the naked eye.

As such, in the conventional method, there was a difference in quality in the production result of artificial teeth according to the skill of the operator, and there was a problem in that mass production was practically difficult, such as taking a lot of work time due to repetitive coloring work.

Patent Document 1: Korean Patent No. 10-1767270 (Announcement Date: August 04, 2017) Patent Document 2: Korea Patent Publication No. 10-2008-0070070 (published on July 29, 2008) Patent Document 3: Korean Patent Registration No. 10-1656106 (Announcement Date: September 08, 2016)

The present invention, which was devised to solve the above problems, accurately acquires color information from natural teeth, and uses deep learning to reproduce the obtained color information in the same way in consideration of the characteristics of materials and pigments of artificial teeth. An object of the present invention is to provide an apparatus for color reproduction of artificial teeth used.

An apparatus for color reproduction of artificial teeth using deep learning according to an embodiment of the present invention is an image processing unit that receives an image of a natural tooth for color reproduction of an artificial tooth, and obtains a reference image through pre-processing of the received image ; Extracting some or all of pixel information for the reference image, generating a pixel information matrix for the pixel information, and classifying the pixels of the pixel information matrix according to a similarity algorithm to divide the reference image into at least two regions a color distribution information generation unit generating color distribution information for the reference image by grouping them into ; and a pigment spraying unit that selects a pigment to be applied or permeated to the artificial tooth according to the artificial tooth material information and the color distribution information, and sprays the selected pigment to some or all of the surface of the artificial tooth.

In this case, the image processing unit performs pre-processing of extracting a natural tooth part from the received image and converting it into the reference image.

Meanwhile, the color distribution information generating unit divides the reference image into a plurality of main blocks, calculates an average value for a plurality of pixels included in the plurality of main blocks, and allocates an average value to the plurality of main blocks to obtain the pixel. Create an information matrix.

In this case, the color distribution information generating unit divides at least one main block among the plurality of main blocks into a plurality of sub-blocks, calculates an average value of a plurality of pixels included in the plurality of sub-blocks, and The pixel information matrix is generated by allocating an average value to sub-blocks.

Meanwhile, the color distribution information generating unit compares the average values allocated to the plurality of main blocks and generates the color distribution information by the similarity algorithm that integrates the corresponding main blocks when the difference value is lower than a reference value.

On the other hand, the pigment spraying unit, a bad member for seating the artificial tooth; a position sensing member for detecting the position of the artificial tooth placed on the bed member; a discharging member discharging the pigment to the artificial tooth; and a control member configured to control the ejection member using the position information of the artificial tooth and the color distribution information.

According to various embodiments of the present invention, since color information can be accurately obtained from natural teeth and the obtained color information can be reproduced in the same way in consideration of the characteristics of materials and pigments of artificial teeth, the manufacturing time of artificial teeth It is possible to shorten the process, lower the manufacturing cost, and exhibit the effect of mass production of high-quality artificial teeth.

1 is a block diagram exemplarily illustrating an apparatus for color reproduction of artificial teeth using deep learning according to an embodiment of the present invention;
FIG. 2 is a diagram exemplarily explaining an image pre-processing process for obtaining the reference image shown in FIG. 1;
FIG. 3 is a view exemplarily explaining a process of obtaining color distribution information by the color distribution information generating unit shown in FIG. 1;
4 is a view exemplarily explaining an example of applying a pigment according to the color distribution information shown in FIG. 3;
5 is a flowchart illustrating an example of a color reproduction method of an artificial tooth using deep learning according to another embodiment of the present invention.

Hereinafter, a preferred embodiment of the present invention will be exemplarily described with reference to the drawings. The following description describes preferred examples within the scope for understanding the technical spirit of the present invention, and is not limited to these examples. In addition, various design changes may be made within the scope of the technical spirit of the present invention.

1 is a block diagram illustrating an apparatus for color reproduction of artificial teeth using deep learning according to an embodiment of the present invention. Referring to FIG. 1 , the color reproduction apparatus 100 includes an image processing unit 110 , a color distribution information generation unit 130 , and a pigment ejection unit 150 .

The image processing unit 110 receives an image file of a natural tooth and generates a reference image file therefor. Specifically, the image processing unit 110 receives a work target image file including a gray card and natural teeth. The image processing unit 110 obtains a reference image through pre-processing on the received image.

The color distribution information generating unit 130 extracts some or all of pixel information for the reference image. The color distribution information generating unit 130 generates a pixel information matrix for the extracted pixel information. The color distribution information generating unit 130 classifies the pixels of the pixel information matrix according to a similarity algorithm and groups the reference image into at least two or more regions to generate color distribution information for the reference image.

The pigment injection unit 150 selects a pigment according to the artificial tooth material information and the color distribution information generated by the color distribution information generation unit 130 . The pigment injection unit 150 applies or penetrates the selected pigment to the artificial tooth. The pigment ejection unit 150 may be implemented by an inkjet method. For example, the pigment spraying unit 150 is a method of expressing an image by spraying a pigment on the surface of an uncolored artificial tooth. The pigment injection unit 150 can use four color pigments (black, red, blue, yellow), but in the case of natural teeth, the color is reproduced by adjusting the white balance, brightness, and saturation, so do not use four color pigments. Alternatively, the color may be colored by adjusting the brightness and saturation of the black pigment. The pigment injection unit 150 can reproduce the same color as that of natural teeth as the number of colors of the pigment increases. In addition, since the pigment ejection unit 150 can adopt the method applied to the existing ink-based printer as it is, the noise is low and the print quality is clear. The pigment spraying unit 150 may be implemented in various ways such as a piezo method, a thermal jet method, a bubble jet method, and the like.

FIG. 2 is a diagram exemplarily explaining an image pre-processing process for obtaining the reference image shown in FIG. 1 . Referring to FIG. 2 , the analysis image is divided into a natural tooth area and a gray card area serving as a standard for brightness/saturation. The natural tooth area contains the tooth image to be reproduced. Also, the gray card area may obtain information about the reference pixel value. The image processing unit 110 acquires a reproduction target tooth image region as a reference image through image preprocessing. In this case, the image processing unit 110 extracts the reference image together with information about the reference pixel value from the gray card area. The image processing unit 110 may also compare the pixel values of the reference image correspondingly based on the reference pixel values. A process of generating color distribution information using pixel values of a reference image will be separately described below.

FIG. 3 is a diagram exemplarily illustrating a process of obtaining color distribution information by the color distribution information generating unit shown in FIG. 1 . Referring to FIG. 3 , the color distribution information generating unit 130 obtains RGB values for all pixels of a reference image (see FIG. 3A ). The color distribution information generator 130 applies a matrix filter of a predetermined size to the reference image and groups the reference image into a plurality of rectangular regions including two or more pixels (refer to FIG. 3B ). The removal area and the remaining area are divided by calculating the average value of the pixels included in the plurality of rectangular areas, and dividing the boundary where the value change rapidly changes compared to the neighboring rectangular areas (refer to FIG. 3(c) ). The color distribution information generator 130 divides the remaining regions into at least two groups by comparing values in pixel units (refer to FIG. 3(d) ). The color distribution information generating unit 130 compares RGB values between adjacent pixels with respect to an algorithm for grouping into at least two groups for the remaining area, and if the difference value is less than or equal to a threshold value, it is determined as the same pixel and included in the same group, If the difference value is greater than or equal to the threshold, it is determined that the pixels are different from each other and divided into different groups.

4 is a view exemplarily explaining an example of applying a pigment according to the color distribution information shown in FIG. 3 . Referring to FIG. 4 , when divided into two regions in FIG. 3 , a dark pigment is applied to the first region, and a bright pigment is applied to the second region. In this case, the brightness values of the dark pigment and the light pigment are relatively determined, and based on the above-described color distribution information, a first basic hue value is determined in the first region and a second basic hue value is determined in the second region. In this way, it is possible to color the artificial teeth sufficiently close to the color tone characteristics of natural teeth only by coloring the color tones of the first area and the second area.

However, in order to reproduce the detailed color characteristics of natural teeth, secondary staining is required for highlighting tones. In FIG. 4 , the texture and detailed characteristics of natural teeth can be reproduced in more detail by reproducing the first emphasized tone region, the second emphasized tone region, and the third emphasized tone region through staining. As for the pigments applied to the first to third accentuated tone regions, a pixel showing a significant difference value from neighboring pixel values in color distribution information is determined as an accentuated tone region. For the pixels determined as an accent color region, a staining pigment is selected to express the corresponding region. Pigments for staining are different from pigments for coloring. Staining is a pigment applied to the surface after coloring the zirconia, heat-treating, and cooling the heat-treated product. The stain pigment uses a pigment that does not have a high absorption rate that penetrates into the interior of the zirconia.

5 is a flowchart illustrating an example of a color reproduction method of an artificial tooth using deep learning according to another embodiment of the present invention. Referring to Figure 5, the color reproduction method of artificial teeth is a step of receiving a natural tooth image file (available to support various image extensions such as JPEG, BMP, PNG, AI, etc.) (S510), the received natural tooth image file is pre-processed for reference Acquiring an image (S520), generating a pixel information matrix for a reference image (S530), applying a similarity algorithm to the pixel information matrix to group the reference image into a plurality of regions to generate color distribution information ( S540), selecting a coloring pigment and a stain pigment using the color distribution information and information on zirconia, which is the main material of the artificial tooth (S550), spraying the selected pigment on the surface of the artificial tooth according to the color distribution information (S560) ) is included.

In step S510, image file information of the natural tooth to be worked may be captured by using a gray card to obtain image file information or may be used by receiving an already acquired image file.

In step S520 , a reference image region including a natural tooth part serving as a reference for a work target is extracted from the image file. In this case, since the gray card area may be excluded from the reference image, a reference pixel value is extracted from the pixels of the gray card area and used as a reference value to generate color distribution information to be described below.

In step S530, a pixel matrix is generated for the entire reference image in order to obtain only a portion of the natural tooth region. Values are compared between adjacent pixels, and if the difference value is greater than the reference value, it is divided into different pixels, and if the difference value is less than the reference value, it is integrated into the same pixel. By processing these pixel matrix values by a similarity algorithm, only the natural tooth region can be extracted.

In step S540, color distribution values for natural tooth portions are compared and divided into at least two or more regions. When color distribution values are compared, a similarity algorithm is applied using pixel matrix information to divide into two groups. The similarity algorithm used here may be implemented by supervised learning or unsupervised learning by machine learning. As supervised learning algorithms, algorithms such as K-nearest neighbors, linear regression, logistic regression, support vector machines, decision trees and random forests, and neural networks can be applied. As unsupervised learning algorithms, K-means, hierarchical cluster analysis, and expected value maximization can be applied.

In step S550, based on the color information of zirconia, appropriate zirconia information, pigment information, coloring method, etc. for implementing color distribution information is determined. The color distribution information is divided into a bright region, a middle region, and a dark region, and the type of zirconia is determined using the bright region, the pigment information is determined using the middle region, and the coloring method is determined using the dark region.

In step S560, the pigment is sprayed on the selected zirconia according to a coloring method determined using a coloring pigment and a stain pigment. For the coloring method, various methodologies may be selected according to precision. The coloring method (coloring, staining) is determined according to low precision, medium precision, and high precision, and the area to which the pigment is applied and the area to be overlapped with the area not to be applied are determined.

On the other hand, even though it has been described that all components constituting an embodiment of the present invention operate by being combined or combined into one, the present invention is not necessarily limited to this embodiment. That is, within the scope of the object of the present invention, all the components may operate by selectively combining one or more. In addition, all of the components may be implemented as one independent hardware, but some or all of the components are selectively combined to perform some or all functions of the combined components in one or a plurality of hardware program modules It may be implemented as a computer program having Codes and code segments constituting the computer program can be easily deduced by those skilled in the art of the present invention. Such a computer program is stored in a computer-readable non-transitory computer readable media, read and executed by the computer, thereby implementing an embodiment of the present invention.

Here, the non-transitory readable recording medium refers to a medium that stores data semi-permanently and can be read by a device, rather than a medium that stores data for a short moment, such as a register, cache, memory, etc. . Specifically, the above-described programs may be provided by being stored in a non-transitory readable recording medium such as a CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM, and the like.

In the above, preferred embodiments of the present invention have been illustrated and described, but the present invention is not limited to the specific embodiments described above, and it is common in the technical field to which the present invention pertains without departing from the gist of the present invention as claimed in the claims. Various modifications may be made by those having the knowledge of, of course, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

100: Apparatus for color reproduction of artificial teeth using deep learning
110: image processing unit
130: color division information generation unit
150: pigment injection unit

Claims (6)

In the color reproduction apparatus of artificial teeth using deep learning,
an image processing unit for receiving an image of a natural tooth for color reproduction of an artificial tooth, and obtaining a reference image through pre-processing of the received image;
Extracting some or all pixel information for the reference image, generating a pixel information matrix for the pixel information, and classifying the pixels of the pixel information matrix according to a similarity algorithm to divide the reference image into at least two regions a color distribution information generation unit generating color distribution information for the reference image by grouping into ; and
Artificial teeth using deep learning including; a pigment spraying unit that selects a pigment to be applied or penetrated to an artificial tooth according to the artificial tooth material information and the color distribution information, and sprays the selected pigment to some or all of the surface of the artificial tooth A device for color reproduction of teeth.
The method of claim 1,
The image processing unit,
Color reproduction apparatus for artificial teeth using deep learning, characterized in that pre-processing of extracting a portion of a natural tooth from the received image and converting it into the reference image is executed.
The method of claim 1,
The color distribution information generating unit,
dividing the reference image into a plurality of main blocks, calculating an average value for a plurality of pixels included in the plurality of main blocks, and allocating an average value to the plurality of main blocks to generate the pixel information matrix A color reproduction device for artificial teeth using deep learning.
4. The method of claim 3,
The color distribution information generating unit,
The pixel information by dividing at least one main block among the plurality of main blocks into a plurality of sub-blocks, calculating an average value for a plurality of pixels included in the plurality of sub-blocks, and assigning an average value to the plurality of sub-blocks A color reproduction device for artificial teeth using deep learning, characterized in that it generates a matrix.
5. The method of claim 3 or 4,
The color distribution information generating unit,
Artificial teeth using deep learning, characterized in that by comparing the average values assigned to the plurality of main blocks with each other and generating the color distribution information by the similarity algorithm that integrates the corresponding main blocks when the difference value is lower than the reference value color reproduction device.
The method of claim 1,
The pigment spraying unit,
a bed member for seating artificial teeth;
a position sensing member for detecting the position of the artificial tooth placed on the bed member;
a discharging member discharging the pigment to the artificial tooth; and
and a control member for controlling the ejection member using the position information of the artificial tooth and the color distribution information.

Images