KR101145672B1 - A smile analysis system for smile self-training - Google Patents

Abstract

PURPOSE: A smile analysis system for smile training is provided to clearly analyze various locations of an individual tooth line by accurately detecting the discrete tooth line. CONSTITUTION: An individual tooth line is detected from an image which is inputted from a camera based on smile training database(200). Smile learning database includes shape model training database and profile model training database including location variation information landmarks. The smile tooth line is detected by using the tooth line.

Description

A smile analysis system for smile self-training}

The present invention, after detecting the lip line and the tooth region using the micro-learning database in the image input to the camera in real time, the individual tooth region is detected, the individual tooth region, after the second derivative in the transverse direction of the tooth region Microscopic analysis for micro training, characterized in that the detection using the vertical boundary detected using the average brightness of the column and the horizontal boundary detected using the average brightness of the row after the second derivative of the tooth region in the longitudinal direction. It is about the system.

Natural and attractive smiles not only play an important role in making the first impression, but also enhance the overall expression, such as preventing the tip of the mouth from sagging through tension and control of the facial muscles, providing a face that always looks fresh and vital.

However, in order to have a good smile, there is a way to visit the smile clinic and receive treatment, but it takes time and money.

Therefore, there is an urgent need for a system that not only recognizes one's smile in real time, but also analyzes, evaluates, and even trains to smile brightly.

Hereinafter, a description will be given of a conventionally developed facial expression recognition system or method.

First, the Republic of Korea Patent Publication No. 10-0874088 relates to a face recognition system using image variation sensing, characterized in that to determine the image variation in the photographed image and apply an application corresponding thereto.

This is different from the problem consciousness of the present invention in that it is positive in that it wants to cope with various image variations, but it is not specialized in recognizing and training smiles and covers the whole expression including smile expression.

Secondly, Korean Patent Laid-Open Publication No. 10-2010-0024261 is characterized by calculating a smile level after extracting a feature element that determines the smile level.

In addition, as a feature, the eye level or the mouth area is extracted, and the smile level is numerically evaluated by numerically evaluating the eye opening degree, the mouth opening degree, the curve following the eye edge line, and the curve following the mouth edge line. It is characterized by calculating.

This is positive in that it is specialized in the expression of laughter, but it does not deal with the various positional relationship of the lips and teeth as an important factor in bright smiles, and also lacks a method for training the smiles. It's different.

Third, the present applicant has proposed a real-time smile training system including application number 10-2011-0005930, image acquisition unit, microfeature detection unit, microlearning database and microanalysis and training unit.

However, the present applicant intends to propose a new function of precisely detecting an individual tooth region and a function of analyzing a smile using the detected individual tooth region.

The purpose of the present invention is to recognize, recognize, and analyze their own smiles in real time, and furthermore, to train bright smiles, while clearly recognizing and recognizing various positional relationships between lips and teeth, which are important factors of bright smiles. To provide a system that can be analyzed.

SUMMARY OF THE INVENTION An object of the present invention is to provide a system capable of precisely detecting individual tooth regions and clearly recognizing and analyzing various positional relationships of lips and individual tooth regions.

 The present invention, after detecting the lip line and the tooth region using the micro-learning database in the image input to the camera in real time, the individual tooth region is detected, the individual tooth region, after the second derivative in the transverse direction of the tooth region Microscopic analysis for micro training, characterized in that the detection using the vertical boundary detected using the average brightness of the column and the horizontal boundary detected using the average brightness of the row after the second derivative of the tooth region in the longitudinal direction. By providing a system, the technical problem is solved.

The present inventors smile analysis system for the smile training, it is possible to precisely detect the individual tooth region can clearly recognize and analyze the various positional relationship of the lips and the individual tooth region has a significant effect.

It also has the significant effect of not only analyzing your smile in real time, but also providing you with a bright smile training.

1 is a main configuration of a smile analysis system for a smile training according to an embodiment of the present invention.
2 is an example of an image for constructing a shape model learning DB and a profile model learning DB in a microanalysis system according to an embodiment of the present invention.
3 is an example of an image for constructing a shape model learning DB and a profile model learning DB in a microanalysis system according to an embodiment of the present invention.
4 is an example in which a landmark is displayed to construct a shape model learning DB and a profile model learning DB in a microanalysis system according to an embodiment of the present invention.
5 is a view for explaining a process of building a profile model learning DB in a microanalysis system according to an embodiment of the present invention.
6 is an example of detecting facial features in the lip line and the tooth detector of the smile analysis system according to an embodiment of the present invention.
7 is an example of detecting a lip landmark in the lip line and the tooth detection unit of the smile analysis system according to an embodiment of the present invention.
8 is an example of detecting a tooth region in a lip line and a tooth detector of a smile analysis system according to an exemplary embodiment of the present invention.
FIG. 9 is an example of a result of horizontally differentiating the tooth region of FIG. 8 in the lip line and the tooth detector of the microanalysis system according to the exemplary embodiment of the present invention.
FIG. 10 is an example of average histogram processing of the results of FIG. 9 in the lip line and the tooth detector of the microanalysis system according to an exemplary embodiment of the present invention.
FIG. 11 is an example of a result of longitudinally differentiating the tooth region of FIG. 8 in the lip line and the tooth detector of the microanalysis system according to the exemplary embodiment of the present invention.
FIG. 12 is an example in which a lip line and a tooth detector of the smile analysis system according to an exemplary embodiment of the present invention process the mean value histogram.
13 is an example of individual tooth regions detected by the lip line and the tooth detector of the smile analysis system according to an exemplary embodiment of the present invention.
14 is an example of analyzing the position of the upper lip in the priority analysis module of the microanalysis system according to an embodiment of the present invention.
15 is an example of analyzing the curve of the upper lip in the upper lip curve analysis module of the microanalysis system according to an embodiment of the present invention.
16 and 17 illustrate an example of analyzing the relationship between maxillary tooth shape and the lip in the maxillary tooth and the lip relation analysis module of the microanalysis system according to the exemplary embodiment of the present invention.
18 is an example of analyzing the maxillary tooth contact relationship of the lower lip in the lower maxillary tooth contact analysis module of the microanalysis system according to an embodiment of the present invention.
19 is an example of analyzing the teeth exposed when smiling in the tooth exposure analysis module of the smile analysis system according to an embodiment of the present invention.
20 is an experimental data image for applying a micro analysis system according to an embodiment of the present invention.
FIG. 21 is a facial feature image detected by applying the microanalysis system according to an exemplary embodiment to the data of FIG. 20.
FIG. 22 is a lip line and an individual tooth region detected by applying the microanalysis system according to an embodiment of the present invention to the data of FIG. 20.
FIG. 23 is a diagram illustrating a result of applying a microanalysis system according to an embodiment of the present invention to the data of FIG. 20 and a hospital score.

The terms and words used in the present specification and claims should not be construed as limited to ordinary or dictionary terms and the inventor may properly define the concept of the term in order to best describe its invention It should be construed as meaning and concept consistent with the technical idea of the present invention.

Therefore, the embodiments described in the present specification and the configurations shown in the drawings are merely the most preferred embodiments of the present invention and are not intended to represent all of the technical ideas of the present invention. Therefore, various equivalents It should be understood that water and variations may be present.

Before describing the present invention with reference to the accompanying drawings, it should be noted that the present invention is not described or specifically described with respect to a known configuration that can be easily added by a person skilled in the art, Let the sound be revealed.

1 is a main configuration of a smile analysis system for a smile training according to an embodiment of the present invention.

The smile analysis system for smile training according to an embodiment of the present invention includes an image acquisition unit 100, a face shape model learning unit 200, a smile feature detection unit 300, and a smile analysis unit 400.

First, the image acquisition unit 100 is a means for receiving an image containing a face and transmits it to the microfeature detection unit 300, and includes an imaging means such as a web camera connected to a PC or a camera embedded in a mobile terminal. As a configuration, the facial image obtained in real time is transferred to the microfeature detection unit 300.

Preferably, the image acquisition unit 100 is also designed to perform a function of transferring the image transmitted to the mobile terminal or the image stored in the external hard disk via the data cable or other network means such as the Internet via a data cable.

The face shape model learning unit 200 is a configuration for constructing a shape model learning DB and a profile model learning DB necessary for detecting a facial feature, a lip line, and an individual tooth region in the microfeature detection unit 300. The active shape model (ASM, Active Shape Model) includes learning module, shape module learning DB and profile learning DB.

The active shape model learning module constructs a shape model learning DB and a profile model learning DB using a plurality of video images.

Here, the biggest reason for building a separate shape model learning DB and profile model learning DB is that the existing learning DB for face or facial expression recognition does not have enough images to smile and therefore the features of the present invention. This is because it is not suitable for detection and analysis.

2 and 3 are examples of images for constructing the shape model learning DB and the profile model learning DB in the microanalysis system according to an embodiment of the present invention, and FIG. 4 shows the shape model learning DB and the profile model learning DB. For example, landmarks are displayed.

Preferably, the active shape model learning module uses an image image photographed in three stages from an image image photographed in three directions of front, left and left inversion as shown in FIG. 2, and a smile smiling brightly in the expressionless manner as shown in FIG. 3. However, as shown in FIG. 4, 86 landmarks are displayed to build a shape model learning DB and a profile model learning DB.

Here, the 86 landmarks are represented by coordinate values representing the outline of the face, eyes, eyebrows, nose, mouth, and lips, and the number of landmarks in each part is 15 face outlines (numbers 0 to 14), 6 right eyebrows (15-20), 6 left eyebrows (21-26), 9 left eyes (27-35), 9 right eyes (36-44), 15 noses (45- 59), 25 mouths (60-84) and 1 ridge (85).

According to the design conditions, the shape model learning DB and the profile model learning DB may be constructed by changing the number and location of landmarks of each part.

The active shape model learning module constructs a shape model learning DB using 86 landmark coordinate values. In order to construct one model representing them, face shapes in multiple images have different sizes and directions. The normalized landmark coordinate values are subjected to linear geometric transformations such as zoom, shift, rotation, and the like so that the coordinates of the center of gravity have the same size and direction, and the coordinates of the center of gravity are the same.

에 대해 평균벡터

가 생성되는 것이 다음 식에 나와 있다.Here, the normalized vector of 86 landmarks

Mean vector for

Is generated by the following equation.

From here,

After generating the mean vector, a displacement vector representing how far from the mean vector the normalized vectors of each image is obtained, and then a covariance matrix is generated for the displacement vector.

에 대해 공분산행렬

를 생성하는 것이 다음 식에나와 있다.Where the displacement vector

Covariance matrix for

To generate the equation

From here,

After generating the covariance matrix, eigenvectors and eigenvalues are generated that satisfy the following equation.

는 고유벡터,

는 고유값이며,

가 되도록 정렬한다.
From here,

Is the eigenvector,

Is an eigenvalue,

Sort to be

The face shape of each image is generated using the eigenvector as shown in the following equation.

는 앞에서부터 k개의 고유벡터이고,

는 각 고유벡터의 가중치를 나타내는 벡터이다.
From here,

Is k eigenvectors from the beginning,

Is a vector representing the weight of each eigenvector.

는 고유벡터

방향으로 학습데이터가 분포하는 분산을 의미하게 된다.In other words, the face shape varies according to the weight of the eigenvector,

Is the eigenvector

It means the variance that the learning data is distributed in the direction.

범위가 되도록 한다.Therefore, in the case of normal distribution, most data are distributed within three times the standard deviation, so the weight is limited accordingly.

To be in range.

As such, the active shape model learning module generates the shape model learning DB by generating the face shape through the above process from the image of each image.

5 is a view for explaining a process of building a profile model learning DB in a microanalysis system according to an embodiment of the present invention.

The active shape model learning module generates a shape model learning DB and then generates a profile model learning DB around each landmark.

Here, the profile indicates what the image around the landmark looks like, and is generated using a one-dimensional vector of pixel change values sampled along a line perpendicular to the contour line of the landmark as shown in FIG. 5.

를 구하고, 전체 랜드마크에 대해 평균 프로파일 벡터

를 구한 후에, 공분산행렬

를 구하는 과정을 통해 프로파일모델 학습DB를 생성하게 된다.
Preferably, similarly to generating a shape model training DB, a profile vector for each landmark of the image image

To obtain the mean profile vector for the entire landmark.

After obtaining, covariance matrix

Through the process of obtaining the profile model training DB is created.

Hereinafter, the structure and function of the microfeature detection part 300 are demonstrated.

The microfeature detection unit 300 includes a face detection module, a lip line and a tooth detection unit 310, and the lip line and the tooth detection unit 310 includes a face feature detection module, a lip line detection module, and a tooth detection module.

When the image acquired in real time from the image acquisition unit 100 is transferred to the microfeature detection unit 300, the microfeature detection unit 300 detects a face using the face detection module, and then the lip line and the tooth detection unit 310. The microfeature is detected using the shape model training DB and the profile model training DB.

First, the face detection module performs a function of detecting a face in the image image transmitted from the image acquisition unit 100, for example, using a detection method using the Haar-like feature and the AdaBoost detection method or using only the AdaBoost method. It is designed to.

Preferably, the face detection module may be designed to adopt a different method for the purpose of the present invention in addition to the above-described method.

The facial feature detection module detects facial features such as face contour, eyes, eyebrows, nose, mouth, and lips using the face region, the shape model training DB, and the profile model training DB detected by the face detection module.

Specifically, first, the face shape formed in the shape model training DB is placed on the detected face area, and then the face shape corresponding to the detected face is changed while changing the shape, and the detected face shape and the detected face shape are detected. When the difference with the face area is below a certain level, the search stops.

Here, since the size or direction of the detected face area may be different from the size and direction of the face shape in the shape model training DB, it is designed to apply the enlargement / reduction and / or rotation transformation in advance.

After the face feature detection module searches for the face shape using the shape model training DB, the face feature detection module detects the face feature by using the profile model training DB to search for a profile model that is close to the change of each landmark.

를 구한 후에 프로파일모델과 가장 유사한 위치를 탐색한다.That is, with respect to the landmark of the detected face region, the profile vector at each position while moving the landmark along a line perpendicular to the contour line

After finding, we search for the location most similar to the profile model.

와 공분산행렬

를 사용하여 다음 식과 같은 Mahalnanobis 거리를 이용하여 구할 수 있다.For example, the mean profile vector of the profile model at which position the profile vector is similar to the profile model.

And covariance matrix

Using the Mahalnanobis distance, we can get

를 구하기 때문에, 일부 랜드마크는 불안정한 위치에 올 수 있고 결국 이미 탐색되었던 얼굴형태가 학습DB의 범위를 벗어날 수 있는 가능성이 있다.Here, profile vector while moving landmark

Since some of the landmarks may come in an unstable position, there is a possibility that the face shape that has already been searched may be out of the range of the learning DB.

와, 형태모델 학습DB에서 구할 수 있는

(

는 얼굴형태

의 근사화한 형태)의 차이가 최소화 되는

를 이용하여 상기 수학식 5를 반복 수행할 필요성이 있다.
Therefore, the face shape formed while moving the landmark

Wow, available from the shape model training DB

(

Face shape

Approximation of the difference)

It is necessary to repeat the above equation (5).

를 구하는 식이다.The following formula

To obtain.

는 형태모델을 영상 이미지의 형태공간으로 사상하는 변환을 의미한다.
From here

Means transforming the shape model into the shape space of the video image.

6 is an example of detecting facial features in the lip line and the tooth detector of the smile analysis system according to an embodiment of the present invention.

In summary, in the facial feature detection module, after searching for the face shape using the shape model training DB, the profile model learning DB is used to search for the change of the landmark for the smiling expression. Face features such as facial contours, eyes, eyebrows, noses, mouths, and lips are detected while limiting the range.

After the facial feature detection module detects the facial feature, the lip line detection module and the tooth detection module detect the lip line and the individual tooth region, respectively.

Herein, the lip line detection module and the tooth detection module may be configured to be included in the modules included in the lip and tooth analysis unit 410 included in the smile analyzer 400 or vice versa. Find out in advance.

7 is an example of detecting a lip landmark in the lip line and the tooth detection unit of the smile analysis system according to an embodiment of the present invention.

First, the lip line detection module extracts an area including the upper and lower order from the image image and simultaneously detects the lip line using the information detected by the facial feature detection module.

For example, in FIG. 7, both ends of the lips (RCh, LCh, landmarks 78 and 84), upper middle lower point (CLow, landmark 81), and lower middle upper point (CLab, landmark 75) are shown. The detection is shown.

8 is an example of detecting a tooth region in a lip line and a tooth detector of a smile analysis system according to an exemplary embodiment of the present invention.

The tooth detection module detects the tooth region image, detects the individual tooth region by performing histogram projection processing after performing the second derivative in the horizontal and vertical directions with respect to the tooth region image.

First, as illustrated in FIG. 8, a tooth region surrounded by landmarks 73 to 84 is detected.

Next, an operation of precisely detecting individual tooth regions is performed based on the detected tooth regions.

FIG. 9 is an example of a result of the second derivative of the tooth region of FIG. 8 in the horizontal direction, and it can be seen that the pixel value of the longitudinal boundary of each tooth is brighter than the pixel value inside the tooth surface.

Preferably, the second derivative is performed in the horizontal direction using a 3 × 3 size Sobel mask for vertical boundary detection, and of course, a different method may be employed depending on design conditions.

FIG. 10 is a result of average histogram processing of the result of FIG. 9 and is performed to find an accurate boundary.

In the image of FIG. 9, the average value of the vertical pixels for each column is obtained as shown in FIG. 10 (a), where a column having a larger value than the surrounding column is left and the remaining columns are removed. do.

Next, among the remaining heat, the heat corresponding to the centerline boundary line of the tooth is first determined, using number 81 of the landmarks as the center position of the lips.

That is, among the remaining heat, the average brightness value is relatively large and the heat closest to the landmark 81 is determined as the position of the central boundary line of the tooth.

After determining the centerline of the tooth, find the outer boundary of the middle two teeth.

In general, the size of the tooth is symmetrical, so the remaining columns have a distance that is a similar distance from the center of the tooth and the average brightness is selected to determine the vertical boundary of the middle two teeth.

Preferably, in the present invention, since the six teeth among the maxillary teeth are selected as the detection target, as shown in FIG. Remove.

Depending on the design conditions, the upper nine can be used when the eight teeth are to be detected.

After finding the longitudinal boundary of the middle two teeth, the second derivative is longitudinally performed in the tooth region of FIG. 8 to detect the horizontal boundary of the two teeth.

FIG. 11 shows that the pixel values of the horizontal boundary appear bright as a result of the vertical second derivative.

FIG. 12A is a result of the average value histogram processing of the result of FIG. 11 and is performed to find an accurate boundary.

Next, if only the row having the highest value is left in FIG. 12 (a) and the remaining rows of noise components are deleted, the horizontal boundary of the tooth can be found as shown in FIG. 12 (b).

Accordingly, the boundary between the two teeth in the center is detected by combining the longitudinal boundary of the two teeth and the horizontal boundary of the two teeth found in FIG. 12 (b).

In addition, this method can be repeated to detect six individual tooth regions among the maxillary teeth as shown in FIG. 13.

Hereinafter, the configuration and function of the microanalysis unit 400 will be described.

The smile analyzer 400 includes a lip and a tooth analyzer 410, and may further include a smile training module, a monitoring module, and a graphic processing module according to design conditions.

The lip and tooth analyzing unit 410 includes an upper rank analysis module, an upper lip curve analysis module, a maxillary and lower lip relationship analysis module, a lower maxillary tooth contact relationship analysis module, a tooth exposure analysis module, and a microsymmetry analysis module.

First, the lips and teeth analyzing unit 410 is a component for analyzing various positional relations between the lips and teeth, which are important factors of the bright smile, and analyzes the aesthetics of the smile.

The upper rank analysis module uses the lip line information and the tooth detection information, but analyzes and evaluates the positional relationship between the upper lip, the teeth and the upper gum when smiling, and FIG. 14 analyzes the position of the upper lip in the upper rank analysis module. An example is shown.

The upper lip is a good smile when the upper teeth are properly exposed. It is not desirable to expose the gums because the upper lip is high and the upper lip is too low to cover the upper teeth.

In the present invention, the presence of the upper gum is analyzed by the difference in pixel values between the upper boundary line and the upper lip line of the individual tooth region.

For example, if the lip line is 3 pixels or less between the upper boundary of the middle tooth and the tooth is greater than the width, 1.5 points, if the distance is less than 3 pixels, if the tooth height is not greater than the width 0.5 points, the distance is more than 3 pixels If it is big, the lip line is raised so that the upper gums are seen a lot and gives 1.0 points.

The upper lip curve module uses lip line information and tooth detection information, and analyzes and evaluates the curve of the upper lip. FIG. 15 shows an example of analyzing the curve of the upper lip in the upper lip curve analysis module.

The upper lip curvature is analyzed using the relationship between the two ends of the lips (RCh, LCh) and the lower middle point (CLow) of the upper lip.

For example, if the mouth tail LCh and RCh are higher than CLow, the upper lip is bent upwards toward the edge, so the aesthetics are good, so 1.5 points is higher, and if only one of the mouth tail LCh and RCh is higher than CLow, 1.0 point is lower than CLow of mouth tail LCh and RCh. If it is in the position, the upper lip is bent downward, so that the aesthetics are inferior.

The maxillary tooth and the lower lip relation analysis module is a module for analyzing and evaluating the relationship between the maxillary teeth and the lower lip using lip line information and tooth detection information. FIGS. 16 and 17 show the maxillary teeth in the maxillary tooth and the lower lip relation analysis module. And an example of analyzing the relationship of the latter.

In the present invention, the relationship between the maxillary teeth and the lower teeth is analyzed using the angle between the detected cutting edge of the maxillary teeth and the horizontal axis.

For example, if the detected angle between the lower cutting edge and the horizontal axis of the six maxillary teeth is 10 degrees or more and the lower cutting edge of the tooth is bent upward and aligned with the lower upper edge, the aesthetics are high. If smaller, the lower edge of the maxillary tooth is closer to a straight line, so the aesthetics are inferior.

The lower maxillary tooth contact analysis module uses lip line information and tooth detection information, but analyzes and evaluates the lower maxillary tooth contact relationship. FIG. 18 shows a lower maxillary tooth contact relationship in the lower maxillary tooth contact analysis module. An example of analyzing is shown.

In the present invention, the relationship between the upper maxillary tooth contact is analyzed using the distance between the lower boundary of the maxillary tooth and the lower landmark 75 (Clab).

For example, if the gap between the maxillary teeth and the lower CLab is more than 7 pixels wide, the aesthetics will be inferior. 0.5 point, less than 1 pixel between end point of maxillary mid incisor and CLab and greater than the height of mid incisor, or greater than 1 pixel between end point of mid incisor of maxilla and CLab If the tooth and the lower end meet properly, the esthetics are good, so give 1.0 points.

The tooth exposure analysis module uses lip line information and tooth detection information, and analyzes and evaluates tooth exposure. FIG. 19 shows an example of analyzing a tooth exposed when smiling in the tooth exposure analysis module.

In the present invention, it is judged that aesthetics are good in proportion to the number of maxillary teeth exposed at the time of micro smile, thereby giving a high score.

For example, six teeth among the teeth are detected, and a score is given based on how many areas of the entire left and right areas of the lips occupy the horizontal direction occupied by the detected teeth.

)이 70% 이하이면 2.0점, 70% ～ 80%이면 1.5점, 80% ～ 90%이면 1.0점, 90% ～ 100%이면 0점을 부여한다.
Percentage of the transverse sections of the six maxillary teeth in the middle

) Is less than 70%, 2.0 points, 70% to 80% 1.5 points, 80% to 90% 1.0 points, 90% to 100% 0 points.

The microsymmetry analysis module uses the lip line information and the tooth detection information, but performs a function of analyzing and evaluating the microsymmetry.

For example, in order to examine the degree of symmetry of the lip line, the relationship between the ends of the lips (RCh, LCh), the lower middle point (CLow) of the upper lip, and the upper middle point (CLab) of the lower lip are analyzed using the following equation.

The sum of the distance between Rch and CLab (Q ₁ ) divided by the distance between Rch and CLow (P ₁ ) divided by the sum of the distance between LCh and CLow (P ₂ ) and the distance between LCh and CLab (Q ₂ ) (X) If it is between 0.9 and 1.1, it is considered as symmetrical and one point is given (microsymmetry = symmetry).

If the value is any other value, it is judged asymmetrically and 0.5 points are given (microsymmetry = asymmetry).

The accuracy was analyzed by applying the microanalysis system of the present invention to experimental image data not used in the learning database.

20 shows 20 experimental data images taken at a dental hospital, and FIG. 21 is a facial feature image detected by applying the analysis system according to the present invention to the data of FIG. 20.

22 is a lip line and an individual tooth region detected by applying the analysis system according to the present invention to the data of FIG. 20.

FIG. 23 is a diagram illustrating a microanalysis result of applying a microanalysis system according to an embodiment of the present invention to the data of FIG. 20 and a hospital score result.

Hospital rating results were scored by a professional at the dental clinic manually evaluating smiles on these images.

As a result of the comparison, the mean error between the manual evaluation by a specialist in the dental clinic and the evaluation rating of the microanalysis system was 0.77.

Since the perfect score was 9 points, the average error was 8.6%, indicating that the results of manual analysis at the dental clinic and microanalysis using the proposed microanalysis system did not show any significant difference.

According to the design conditions, after analyzing the various positional information of the lips and teeth for esthetic analysis of the smile in the lips and teeth analysis unit 410, the system of the present invention displays the evaluation score and comments of the smile on the screen In order to provide a process for the user to train the smile, it may include a smile training module, a monitoring module and a graphic processing module.

First, the graphic processing module is configured to display an evaluation score, a comment, a training process, etc. graphically on the screen by interworking with the lip and tooth analysis unit 410, a monitoring module, and a smile training module.

Depending on the design conditions, the graphics processing module can of course be configured to provide an interface that can utilize the system of the present invention as a whole.

The smile training module performs a function of providing a training program so that users can systematically train smiles.

The monitoring module checks whether the user maintains the same smile as the presented image for a predetermined time, transmits the result to the graphic processing module, and controls to move to the next step when the presented training is completed.

In addition, by extracting and storing the screen of the training process, the user monitors the stored image after the training to perform a function to check their smile training process.

Depending on the design conditions, the function of the micro training module may be included in the monitoring module or vice versa.

1 to 23 describe only the main matters of the present invention, and the present invention is limited to the configuration of FIGS. 1 to 23 as long as various designs are possible within the technical scope. Or not.

100: image acquisition unit
200: face shape model learning unit
300: micro feature detection unit
310: lip line and teeth detection unit
400: microanalysis
410: lips and teeth analysis unit

Claims (6)

After detecting the lip line and the tooth region using the micro-learning database in the image input to the camera in real time, individual tooth regions are detected.
The smile learning database,
A shape model training database including face shape information, and a profile model training database including position change information of a plurality of landmarks attached to an outline of a face main element,
Individual tooth areas,
Detecting using the vertical boundary detected after the second derivative of the tooth region in the horizontal direction using the average brightness of the column, and using the horizontal boundary detected using the average brightness of the row after the second derivative of the tooth region in the longitudinal direction Smile analysis system for smile training, characterized in that.
delete The method of claim 1,
Analyze the position of the upper lip using the detected individual tooth regions,
And analyzing the difference between the upper boundary of the individual tooth region and the pixel value of the lip line. The method of claim 1,
Analyze the relationship between the maxillary teeth and the lower teeth using the detected individual tooth regions,
And analyze using the angle between the lower cutting edge and the horizontal axis of the individual tooth region. The method of claim 1,
Analyze the relationship between the maxillary teeth and the lower teeth using the detected individual tooth regions,
And a pixel value difference between the boundary of the individual tooth region and the center point of the lower lip line. The method of claim 1,
Analyzing the teeth exposed during the smile using the detected individual tooth area,
And analyzing by using the ratio of the horizontal length of the individual tooth region to the total length of the left and right lips.

Images