TWI909997B - Tracking method for tongue and its rehabilitation system for tongue muscle - Google Patents

Abstract

The present invention provides a tracking method for tongue and a rehabilitation system for tongue muscle, wherein the tracking method for tongue comprises steps: placing a tongue suction device on a tongue, wherein the tongue suction device is configured with a tracking point; acquiring a first tongue muscle exercise image of the tongue; converting a color space of the first tongue muscle exercise image; filtering a specified color value in the color space to generate a filtering result; performing image analysis based on the filtering result to generate plural regional image shapes; determining aforementioned plural regional image shapes to generate a target object corresponding to the tracking point; drawing a contour information corresponding to the target object; identifying a coordinate values of the contour information; and calculating the coordinate values to generate a first position information corresponding to the tongue. By integrating with the tongue muscle rehabilitation system, the efficiency in rehabilitation is significantly enhanced.

Description

Tongue tracking methods and tongue muscle rehabilitation system

A tracking method and its rehabilitation system, particularly relating to a tongue tracking method and its tongue muscle rehabilitation system.

According to statistics from the Ministry of the Interior, in 2024, the number of people aged 65 and above in Taiwan reached more than 4 million, accounting for 17.8% of the total population, with an aging index of 147.9. From the perspective of international aging indicators, 7%, 14%, and 20% of the population aged 65 and above are used as indicators of different degrees of aging. Among them, 7% represents an aging society, 14% represents an aged society, and 20% represents a super-aged society. Therefore, according to the National Development Council's prediction, Taiwan will enter a super-aged society in 2025. This year, 40 countries and regions have already entered super-aged societies, and it is estimated that more countries will enter super-aged societies in the coming years.

With the arrival of a super-aged society, the demand for medical care is increasing. Domestic and international research indicates that age has a significant impact on tongue muscle strength, and diseases can also lead to tongue muscle dysfunction, such as sarcopenia, Parkinson's disease, stroke, or dementia. The risk of developing these diseases often increases with age. According to literature, tongue muscle rehabilitation methods can be broadly categorized into... There are two types of exercises: one is head exercises, and the other is mouth-circumference exercises. The head exercises, such as head lifting and head extension swallowing exercises, are mostly assisted by tongue muscle strength recording equipment. The mouth-circumference exercises, such as mouth opening and chin retraction resistance exercises, can be assisted by tongue positioners and bouncy balls, respectively. In this way, the motor ability of various tongue muscle groups can be improved.

Therefore, in existing technologies, devices for tongue muscle rehabilitation include IOPI Pro Device, oral function training kits, tongue exercisers, Electric Tongue Exerciser Kits, or traction devices. However, most of these tongue muscle rehabilitation devices lack interactivity. Devices for tongue rehabilitation games mainly use tongue depressors combined with surface electromyography (sEMG) to measure the user's tongue muscles and swallowing status and evaluate their swallowing effect. Another method is to use image sensing combined with a head-mounted display to obtain the tongue's movement status. However, these rehabilitation devices require a variety of auxiliary devices, which are not only expensive but also space-limited.

To address this problem, this invention provides a tongue tracking method and its tongue muscle rehabilitation system. Based on existing tongue suction devices, the tongue tracking method uses image recognition to obtain the tongue position and combines it with rehabilitation tasks to significantly improve the interactivity of tongue muscle rehabilitation. Moreover, it is not limited by space or equipment. Even if the user moves their head during rehabilitation, the tongue position can still be accurately obtained and the direction of tongue movement can be identified, effectively improving the rehabilitation effect.

The main purpose of this invention is to provide a tongue tracking method that performs color space conversion on the image of the first tongue muscle movement and performs image analysis through the selection of specified color values, effectively acquiring and tracking the position of the tongue suction device and greatly improving tracking accuracy.

Another objective of this invention is to provide a tongue muscle rehabilitation system that identifies the position of a tongue suction device using a tongue tracking method and enables a virtual object to move synchronously with the position of the tongue suction device through a control module, thereby completing the rehabilitation task, ensuring the correct execution of tongue muscle movements, and effectively improving rehabilitation results.

To achieve the above objectives, one embodiment of the present invention discloses a tongue tracking method, comprising the following steps: installing a tongue suction device on a tongue, the suction device having a tracking point; acquiring a first tongue muscle movement image of the tongue; converting a color space of the first tongue muscle movement image; filtering a specified color value in the color space to generate a filtering result; performing image analysis on the filtering result to generate a plurality of regional image shapes; determining the regional image shapes to generate a target object corresponding to the tracking point; drawing corresponding contour information based on the target object; identifying a coordinate value of the contour information; and calculating the coordinate value to generate a first position information corresponding to the tongue.

In a preferred embodiment, in the step of determining the shapes of the regional images and generating a target object corresponding to the tracking point, parameter values of the shapes of the regional images are extracted respectively, and an area value and perimeter of each regional image shape are calculated. A corresponding fitted ellipse is generated based on the area value and the perimeter. An ellipse aspect ratio and a roundness value are calculated based on the fitted ellipse. The target object is determined based on the ellipse aspect ratio and roundness value.

In a preferred embodiment, in the step of identifying a coordinate value of the contour information, the contour information is the approximate ellipse corresponding to the target object and a circumscribed rectangle corresponding to the approximate ellipse, and the coordinate value of the approximate ellipse and the circumscribed rectangle is identified. The coordinate value includes the coordinate value of the center point of the approximate ellipse and a first coordinate value and a second coordinate value of the circumscribed rectangle.

In a preferred embodiment, a regression analysis is performed using the first tongue muscle movement image, the contour information, the coordinate value, and the first position information to generate an analysis result; a second tongue muscle movement image is input; and the analysis result is used to determine the second position information of the tracking point in the second tongue muscle image.

In a preferred embodiment, in the step of calculating the coordinate value to generate a first position information corresponding to the tongue, a range of movement of the tongue is set, and the first position information is generated by calculating the coordinate value and the range of movement.

To achieve another objective mentioned above, one embodiment of the present invention discloses a tongue muscle rehabilitation system, comprising: a tongue suction device disposed on a tongue, the tongue suction device having a tracking point; an image capture module for acquiring an image of the movement of a first tongue muscle; a recognition module signal-connected to the image capture module for recognizing the image of the first tongue muscle to generate first position information corresponding to the tongue; a control module signal-connected to the recognition module for controlling the movement of a virtual object based on the first position information; and a rehabilitation module signal-connected to the control module for providing a rehabilitation task, causing the virtual object to move according to the rehabilitation task.

In a preferred embodiment, the recognition module performs color space conversion on the first tongue muscle movement image, and performs image analysis after filtering a specified color value to generate a target object. The module then draws corresponding outline information based on the target object, identifies a coordinate value of the outline information, and performs calculations based on the coordinate value and a range of movement of the tongue to generate the first position information. The target object corresponds to the tracking point of the tongue suction device.

In a preferred embodiment, the recognition module generates a plurality of regional image shapes after filtering the specified color values, extracts a parameter value of each regional image shape, calculates an area value and a perimeter of each regional image shape, generates a corresponding fitted ellipse based on the area value and the perimeter, calculates an ellipse aspect ratio and a roundness value based on the fitted ellipse, and determines the target object based on the ellipse aspect ratio and roundness value.

In a preferred embodiment, the contour information is obtained by identifying the coordinate values of the fitting ellipse and the circumscribed rectangle corresponding to the target object. The coordinate values include the coordinate value of the center point of the fitting ellipse and the first and second coordinate values of the circumscribed rectangle.

In a preferred embodiment, the recognition module acquires the first tongue muscle movement image, the contour information, the coordinate value, and the first position information to perform a regression analysis and generate a corresponding analysis result. When a second tongue muscle movement image is acquired from the image extraction module, the recognition module uses the analysis result to determine the tracking point in the second tongue muscle image to generate a second position information corresponding to the tongue.

The beneficial effects of this invention lie in its ability to quickly track the position of the tongue by combining the existing tongue suction device with a tongue tracking method, thereby controlling the tongue to complete rehabilitation tasks. This rehabilitation method does not require additional auxiliary equipment, which greatly reduces the construction cost. At the same time, it is not limited by the space of use and greatly increases the rehabilitation effect, solving the problem that existing technologies require a variety of auxiliary equipment and are not conducive to home use.

The relevant patent features and technical contents of this invention will be clearly presented in the following detailed description of the preferred embodiments with reference to the accompanying drawings.

Please refer to Figure 1, which is a flowchart of a method according to an embodiment of the present invention. As shown in the figure, a tongue tracking method of the present invention includes the following steps:

Step S1: Install a tongue suction device on a tongue, the tongue suction device having a tracking point;

Step S2: Acquire an image of the movement of one of the first tongue muscles;

Step S3: Change the color space of one of the first tongue muscle movement images;

Step S4: Filter the specified color values of one of the color spaces to produce a filter result;

Step S5: Perform image analysis based on the filtering results to generate multiple regional image shapes;

Step S6: Determine the shape of these regional images and generate a target object corresponding to the tracking point;

Step S7: Draw the corresponding outline information based on the target object;

Step S8: Identify one of the coordinate values of the contour information; and

Step S9: Calculate the coordinate value to generate one of the first position information corresponding to the tongue.

Please also refer to Figure 2, which is a schematic diagram of a system according to an embodiment of the present invention. As shown in the figure, a tongue muscle rehabilitation system S of the present invention includes: an image acquisition module 1, a recognition module 2, a control module 3, and a rehabilitation module 4. The recognition module 2 is connected to the image acquisition module 1 and the control module 3 by signal, and the rehabilitation module 4 is connected to the control module 3 by signal. The steps of the tongue tracking method are explained in detail below.

As shown in step S1, please also refer to Figure 3, which is a schematic diagram of one embodiment of the present invention. As shown in the figure, a tongue suction device H is provided on the tongue. The tongue suction device H includes, but is not limited to, an active rehabilitation tongue suction device H. Preferably, one end of the tongue suction device H is provided with a tracking point H1. The tracking point H1 can be a single color, and the type of color is not limited. In this embodiment, red is used as an example.

As shown in step S2, the image capture module 1 captures the first tongue muscle movement image of the user's tongue with the tongue suction device H. The image capture module 1 can be placed flat or suspended in front of the user.

As shown in step S3, the recognition module 2 converts the color space of the first tongue muscle movement image obtained in the previous step. In one embodiment, the color space of the first tongue muscle movement image can be RGB color space, and it is converted into HSV color space according to its RGB color space. The converted color space can more intuitively define the range of colors. RGB represents the three primary colors of light: red, green, and blue, respectively, while HSV represents hue, saturation, and value. Visually, the coordinate system of RGB color space is a three-dimensional rectangular coordinate system, while the coordinate system of HSV color space is a spherical coordinate system.

As shown in step S4, the identification module 2 can then filter specified color values based on the converted color space to generate a filtering result. In one embodiment, the specified color value corresponds to the color of the tracking point H1 set at one end of the tongue suction device H. Preferably, this embodiment uses red. Therefore, the specified color value can be defined for the hue, saturation, and brightness of red. Specifically, as shown in Table 1 below, it is the subject of this embodiment. The HSV color space adopted in the regulations has the following general values: red is divided into two parts in the hue range of 0 to 180, specifically the ranges of 0 to 10 and 156 to 180; saturation is defined as 43 to 255, excluding lighter areas; and lightness is defined as 46 to 255, excluding darker areas. However, these are not restrictions, and other specified color values may also be used.

Table 1. Color Range General Values of HSV Color Space black grey White red orange color yellow green blue blue Purple H _min 0 0 0 0 156 11 26 35 78 100 125 H _max 180 180 180 10 180 25 34 77 99 124 155 S _min 0 0 0 43 43 43 43 43 43 43 S _max 255 43 30 255 255 255 255 255 255 255 V _min 0 46 221 46 46 46 46 46 46 46 V _max 46 220 255 255 255 255 255 255 255 255 *H _min is the minimum hue value; H _max is the maximum hue value; S _min is the minimum saturation value; S _max is the maximum saturation value; V _min is the minimum lightness value; and V _max is the maximum lightness value.

As shown in step S5, the identification module 2 will perform image analysis based on the aforementioned color filtering results to generate multiple regional image shapes. In one embodiment, the outline pixel values of each region in the filtered image can be calculated, and regions below a preset pixel value can be filtered based on the values of each region. In this way, the specific regions to be retained (i.e., the green value portion in Figure 3), i.e., multiple regional image shapes, can be determined. In one embodiment, the preset pixel value of the specific regions to be filtered (i.e., the red value portion in Figure 3) can be 8 to 12, but is not limited to this.

As shown in step S6, please refer to Figure 4, which is a schematic diagram of one embodiment of the present invention. As shown in the figure, the identification module 2 will determine which of the regional image shapes is the target object corresponding to the tongue suction device H, that is, find the tracking point H1 set at one end of the tongue suction device H. In one embodiment, in order to determine whether the regional image shapes are target objects, it is necessary to first extract the individual parameter values of each region, and calculate the area value (as shown in part (a) of Figure 4) and perimeter (as shown in part (b) of Figure 4) of each regional image shape to obtain the area of each regional image shape. The value and perimeter are used to generate corresponding approximate ellipses (as shown in parts (c) and (d) of Figure 4). Next, the ellipse ratio (as shown in part (e) of Figure 4) and roundness value (as shown in part (f) of Figure 4) are calculated based on these approximate ellipses. The target object is then determined based on the ellipse ratio and roundness value. The ellipse ratio ranges from 0.5 to 1.5, and the object with the highest roundness is determined to be the target object. Ideally, the roundness should be greater than 0.4, but this is not a limitation. The calculation formula is as follows:

As shown in step S7, please refer to Figure 5, which is a schematic diagram of one embodiment of the present invention. As shown in the figure, the identification module 2 draws the outline information of the target object obtained in the previous step. In one embodiment, the outline information drawn is the fitting ellipse corresponding to the target object (as shown by the green line in Figure 5) and the circumscribed rectangle corresponding to this fitting ellipse (as shown by the white box in Figure 5), but it is not limited to this.

As shown in step S8 and Figure 5, the identification module 2 identifies the outline information drawn in the previous step, thereby identifying the coordinate values of the fitted ellipse and the circumscribed rectangle. Preferably, the coordinate values include the coordinate values of the center point of the fitted ellipse. and the first coordinate value of the circumscribed rectangle and the second coordinate value The first coordinate value of the circumscribed rectangle can be the diagonal coordinate of the second coordinate value.

In one embodiment, the calculation method for coordinate values is explained as follows:

First, the coordinates of the center point of the ellipse are approximated. Defined as Furthermore, the coordinates of any point on the approximate ellipse are: Let the first and second coordinates of the circumscribed rectangle be _B1 ( _x1 , _y1 ) and _B2 ( _x2 , _y2 ) respectively, and combine them with the following formula:

(1)

(2)

(3)

(4)

Formulas (1) and (2) are coordinate algorithms for a fitted ellipse without tilt angle, while formulas (3) and (4) are coordinate algorithms for a fitted ellipse with tilt angle. The coordinate algorithm.

Since the circumscribed rectangle and the fitting ellipse are tangent to each other, the maximum and minimum values of the coordinates of the first coordinate _B1 and the second coordinate _B2 , calculated by formulas (3) and (4), are as follows:

As shown in step S9, after the identification module 2 performs calculations based on the acquired coordinate values, it can obtain the first position information corresponding to the tongue. In one embodiment, please refer to Figure 6A, which is a schematic diagram of an embodiment of the present invention. As shown in the figure, the range of movement of the tongue suction device H within the oral cavity can be further defined. When the distance between the image capture module 1 and the user is n cm, assuming that a moving boundary frame can completely cover the mouth, and considering that different users may have different mouth shapes, the range of the aforementioned moving boundary frame can be defined as the width LTB of the moving boundary frame, which is 1.5 times _{the width} of the mouth, and the height LTB is defined as _{the height} LTB of the moving boundary frame, which is 1.5 times the height of the mouth plus the thickness TSD _of the tongue suction device H. This serves as the threshold for the range of movement of the tongue suction device H within the oral cavity. For example, if the mouth width is 50 mm, then the width of the moving boundary frame... For example, if the mouth height is 20mm, then the width of the border frame can be moved. The coordinates of this moving bounding box include a third coordinate value _LTB1 ( _X1 , _Y1 ), a fourth coordinate value _LTB2 ( _X2 , _Y2 ), and a center point coordinate _LTBc ( _Xc , _Yc ), and the third coordinate value is the diagonal of the fourth coordinate value.

For example, as shown in Figure 6A, the image captured by image capture module 1 has a pixel value of 1280 pixels (length). 720 pixels (width) For example, the actual width of the move bounding box. and height Convert to pixel values and The calculation formula is as follows:

In this way, the coordinates of the center point of the moving bounding box are determined. The distance between the coordinates of the center point of the approximate ellipse and the distance between the center point ...

Specifically, please refer to Figures 6B to 6C, which are schematic diagrams of one embodiment of the present invention. As shown in the figures, when the distance between the user and the image capturing module 1 is 60cm, a mouth width of 50mm can be converted to 100 pixels, a moving bounding box width of 75mm can be converted to 150 pixels, and a moving bounding box width of 60mm can be converted to 120 pixels. The coordinates of the center point of the moving bounding box are... and the coordinates of the center point of the approximate ellipse At that time, and The calculation results are as follows:

Right now

Right now

At this time, the horizontal movement distance of the tongue suction device H is... as follows:

Therefore, the positive or negative value can be used to determine the left or right direction of the tracking point H1 (i.e., the tongue). When, it is interpreted as the tracking point H1 (i.e., the tongue) moving to the left, or when When the time is right, it is interpreted as tracking point H1 (i.e., the tongue) to the right.

At this time, the vertical movement distance of the tongue suction device H is... as follows:

Therefore, the direction of the tracking point H1 (i.e., the tongue) can be determined by the positive or negative value. When, it is interpreted as tracking point H1 (i.e., the tongue) moving upwards, or when When the time is right, it is interpreted as tracking point H1 (i.e., the tongue) downwards.

In one embodiment, the pixel values of the image obtained by the image capturing module 1 can be used to set dynamic adjustment of coordinates (DAC) so that the position of the rectangular area can be dynamically adjusted with the image capturing module 1. For example, if the pixel value of the image capturing module 1 is 1280x720 (with an aspect ratio of 16:9), the dynamic adjustment coordinates are set to the height pixel size of the image capturing module 1 divided by 720, i.e., DAC = 720/720.

In one embodiment, the coordinate mapping formula can be used to determine the final size of the area mapped to the display, as shown in Figure 6B. Taking the X-axis as an example, let's assume the coordinates of the upper left corner of the final mapped area are... And the coordinates of the lower right corner of the final mapping range are When the display resolution is 1080p, the calculation formula is as follows:

As shown in Figure 6C, taking the Y-axis as an example, the calculation formula is as follows:

In one embodiment, the identification data obtained in the above steps can be subjected to regression analysis. First, the identification module 2 acquires the first tongue muscle movement image, contour information, coordinate values and first position information for regression analysis, thereby generating corresponding analysis results. Then, the identification module 2 can use this analysis result to determine the second position information corresponding to the tongue in the second tongue muscle movement image it has acquired, but this is not limited to this.

To illustrate the operation of this embodiment in more detail, specific examples are provided below:

Rehabilitation module 4 provides rehabilitation tasks, which can be divided into left-right, up-down, or circular tongue rehabilitation. The rehabilitation tasks can be presented in a game mode. The user can set the tongue suction device H on the tongue, and after the image capture module 1 obtains the tongue muscle movement image, the recognition module 2 identifies the tracking point H1 of the tongue suction device H to generate the corresponding position information of the tongue. Based on this, the direction of movement of the tracking point H1 (i.e., the tongue) is determined, such as left-right, up-down, and circular. At the same time, the movement distance is read. Based on the first position information, the control module 3 controls the movement of the virtual object. At this time, the tongue can be moved according to the needs of the rehabilitation task, so that the virtual object moves accordingly to complete the corresponding rehabilitation task.

In summary, the tongue tracking method and tongue muscle rehabilitation system of one embodiment of the present invention obtain the position of the tracking point (i.e., the tongue) on the tongue suction device through image recognition, and combined with rehabilitation tasks, greatly enhances the interactivity of tongue muscle rehabilitation. Moreover, it is not limited by space or equipment. Even if the user moves his head during the rehabilitation process, the tongue position can still be accurately obtained and the direction of tongue movement can be identified, which greatly improves the rehabilitation effect.

However, the above description is merely a preferred embodiment of the present invention and should not be used to limit the scope of the present invention. Any simple equivalent changes and modifications made in accordance with the scope of the patent application and the description of the invention shall still fall within the scope of the present invention.

1: Image Acquisition Module 2: Recognition Module 3: Control Module 4: Rehabilitation Module H: Tongue Suction Device H1: Tracking Point S: Tongue Muscle Rehabilitation System S1: Step S2: Step S3: Step S4: Step S5: Step S6: Step S7: Step S8: Step S9: Step

Figure 1: A flowchart illustrating a method according to an embodiment of the present invention; Figure 2: A system diagram according to an embodiment of the present invention; Figure 3: An implementation diagram according to an embodiment of the present invention; Figure 4: An implementation diagram according to an embodiment of the present invention; Figure 5: An implementation diagram according to an embodiment of the present invention; Figure 6A: An implementation diagram according to an embodiment of the present invention; Figure 6B: An implementation diagram according to an embodiment of the present invention; and Figure 6C: An implementation diagram according to an embodiment of the present invention.

S1: Steps

S2: Steps

S3: Steps

S4: Steps

S5: Steps

S6: Steps

S7: Steps

S8: Steps

S9: Steps

Claims (10)

A tongue tracking method includes the following steps: installing a tongue suction device on a tongue, the suction device having a tracking point; acquiring a motion image of a first tongue muscle on the tongue; converting a color space of the first tongue muscle motion image; filtering a specified color value in the color space to generate a filtering result; performing image analysis on the filtering result to generate a plurality of regional image shapes; determining the regional image shapes to generate a target object corresponding to the tracking point; drawing corresponding contour information based on the target object; identifying a coordinate value of the contour information; and calculating the coordinate value to generate a first position information corresponding to the tongue. According to the tongue tracking method described in claim 1, in the step of determining the shape of the regional images and generating a target object corresponding to the tracking point, a parameter value of the shape of the regional images is extracted respectively, and an area value and a perimeter of each regional image shape are calculated. A corresponding fitted ellipse is generated based on the area value and the perimeter. An ellipse axis ratio and a roundness value are calculated based on the fitted ellipse. The target object is determined based on the ellipse axis ratio and the roundness value. According to the tongue tracking method described in claim 2, in the step of identifying a coordinate value of the contour information, the contour information is the approximate ellipse corresponding to the target object and a circumscribed rectangle corresponding to the approximate ellipse, and the coordinate value of the approximate ellipse and the circumscribed rectangle is identified. The coordinate value includes the coordinate value of the center point of the approximate ellipse and a first coordinate value and a second coordinate value of the circumscribed rectangle. The tongue tracking method according to claim 1 includes the following steps: performing a regression analysis on the first tongue muscle motion image, the contour information, the coordinate value and the first position information to generate an analysis result; inputting a second tongue muscle motion image; and determining a second position information of the tracking point in the second tongue muscle image based on the analysis result. According to the tongue tracking method described in claim 1, in the step of calculating the coordinate value to generate a first position information corresponding to the tongue, a range of movement of the tongue is set, and the first position information is generated by calculating the coordinate value and the range of movement. A tongue muscle rehabilitation system includes: a tongue suction device disposed on a tongue, the tongue suction device having a tracking point; an image capture module for acquiring an image of the movement of a first tongue muscle on the tongue; a recognition module signal-connected to the image capture module for recognizing the image of the first tongue muscle movement to generate first position information corresponding to the tongue; a control module signal-connected to the recognition module for controlling the movement of a virtual object based on the first position information; and a rehabilitation module signal-connected to the control module for providing a rehabilitation task, causing the virtual object to move according to the rehabilitation task. According to the tongue muscle rehabilitation system described in claim 6, the recognition module performs color space conversion on the first tongue muscle movement image, and performs image analysis after filtering a specified color value to generate a target object. The target object is used to draw a corresponding outline information, and a coordinate value of the outline information is identified. The first position information is generated based on the coordinate value and a range of movement of the tongue. The target object corresponds to the tracking point of the tongue suction device. According to the tongue muscle rehabilitation system described in claim 7, the recognition module generates a plurality of regional image shapes after filtering the specified color values, extracts a parameter value of each regional image shape, calculates an area value and a perimeter of each regional image shape, generates a corresponding fitted ellipse based on the area value and the perimeter, calculates an ellipse axis and a roundness value based on the fitted ellipse, and determines the target object based on the ellipse axis and roundness value. According to the tongue muscle rehabilitation system described in claim 8, the contour information is based on the approximate ellipse corresponding to the target object and a circumscribed rectangle corresponding to the approximate ellipse, identifying the coordinate values of the approximate ellipse and the circumscribed rectangle, the coordinate values including the coordinate value of the center point of the approximate ellipse and a first coordinate value and a second coordinate value of the circumscribed rectangle. According to the tongue muscle rehabilitation system described in claim 7, the recognition module acquires the first tongue muscle movement image, the contour information, the coordinate value, and the first position information to perform a regression analysis and generate a corresponding analysis result. When a second tongue muscle movement image is acquired from the image extraction module, the recognition module uses the analysis result to determine the tracking point in the second tongue muscle image to generate a second position information corresponding to the tongue.