KR102128268B1 - Method and system for walking ability prediction using foot characteristics information - Google Patents

Abstract

A walking capability prediction system is provided. The walking capability prediction system includes a foot characteristic information acquisition unit that provides a virtual walking environment to a subject and acquires foot characteristic information of the subject according to the virtual walking environment; A motion detector for sensing the subject's body motion and extracting the subject's gait pattern information from the subject's body motion information; A database unit that stores the gait pattern information of the subject for each foot characteristic of the subject; A standard walking pattern information generating unit that analyzes a relationship between the subject's foot characteristic information and the subject's walking pattern information, and performs machine learning to generate standard walking pattern information corresponding to the subject's foot characteristics; And a walking ability prediction unit that predicts the walking ability of the new subject by substituting foot characteristic information acquired from the new subject into the standard walking pattern information.

Description

Method and system for walking ability prediction using foot characteristics information

It relates to a walking ability prediction method and system, and specifically to a walking ability prediction method and system using foot property information.

Gait is the most basic means of human mobility, including personal health information, pathological symptoms and intentions for movement. Through quantitative evaluation of gait, it is possible to diagnose and monitor musculoskeletal and neurological diseases, determine the effect on systematic treatment, and be used for surgical surgical decisions and postoperative review. These walking abilities can be analyzed from a kinematic, kinematic, and biosignal point of view, but to quantify walking abilities, expensive infrared cameras, experienced experimenters, expensive ground reaction forces, pressure sensors, and electromyography Personnel and equipment such as (Electromyography) are required.

Conventionally, the method for evaluating walking ability using the above-described equipment has not only a high cost due to the input of a large number of devices and manpower, but also has difficulty in taking accurate result data because patients with inconvenient behavior cannot perform all of the above-described experiments. There was.

Accordingly, there is a need for a system and a method capable of predicting walking ability and walking characteristics through simple tests.

United States Patent Publication US7467060B2 (December 16, 2008)

The present invention has been devised to solve the above-mentioned problems, and specifically, relates to a method and system for predicting walking ability using foot characteristic information.

A walking capability prediction system according to an embodiment of the present specification includes a foot characteristic information acquisition unit that acquires a foot depth image of a subject located on a scaffold and acquires foot characteristic information from the acquired foot depth image; A motion detector for sensing the subject's body motion and extracting the subject's gait pattern information from the subject's body motion information; A database unit for storing the walking pattern information of the target person according to the foot characteristic information of the target person; A standard walking pattern information generating unit that analyzes a relationship between the subject's foot characteristic information and the subject's walking pattern information, and performs machine learning to generate standard walking pattern information corresponding to the subject's foot characteristics; And a walking ability prediction unit that predicts the walking ability of the new subject by substituting foot characteristic information acquired from the new subject into the standard walking pattern information.

In one embodiment, the foot characteristic information may include a sole shape, a sole width, a position of a foot arch, a width of a foot arch, and an angle of a foot arch curve.

In one embodiment, the foot characteristic information acquisition unit may extract an MLA curve and an LLA curve from the foot depth image, and acquire the foot characteristic information based on at least one of the MLA curve and the LLA curve.

In one embodiment, the subject's gait pattern information includes time parameters and spatial parameters, and the time parameters include stride time, step time, stance time, swing time, single rim support time, double rim support time, cadence, and the like. Including, the spatial parameter may include a stride length, step length, walking speed, and the like.

In one embodiment, the standard walking pattern information generating unit may perform machine learning by using the foot characteristic information as an input value and using the subject's walking pattern information as an output value.

In one embodiment, the standard walking pattern information generation unit may generate the standard walking pattern information through deep learning.

The walking ability prediction method acquires a foot depth image of a subject located on a scaffold, obtains foot characteristic information from the acquired foot depth image, detects the subject's body motion, and walks the subject from the subject's body motion information By extracting pattern information, the subject's walking pattern information for each subject's foot characteristic information is stored, and the relationship between the subject's walking pattern information and the subject's foot characteristic information is analyzed, and machine learning is performed to machine the subject's foot Constructing standard walking pattern information corresponding to the characteristics; Obtaining foot characteristic information from a new subject; And predicting the walking ability of the new subject by substituting the foot gait information of the new subject into the standard walking pattern information.

In one embodiment, the foot characteristic information may include a sole shape, a sole width, a position of a foot arch, a width of a foot arch, and an angle of a foot arch curve.

In one embodiment, the foot characteristic information acquisition unit may extract an MLA curve and an LLA curve from the foot depth image, and acquire the foot characteristic information based on at least one of the MLA curve and the LLA curve.

In one embodiment, the subject's gait pattern information includes time parameters and spatial parameters, and the time parameters include stride time, step time, stance time, swing time, single rim support time, double rim support time, cadence, and the like. Including, the spatial parameter may include a stride length, step length, walking speed, and the like.

In one embodiment, the step of constructing the standard gait pattern information may perform machine learning using the foot characteristic information as an input value and the gait pattern information of the subject as an output value.

In one embodiment, the step of constructing the standard walking pattern information may generate the standard walking pattern information through deep learning.

By using the standard walking pattern information corresponding to the constructed foot characteristic information, it is possible to predict the actual walking ability of a new subject with only the acquired foot characteristic information.

Accordingly, input of equipment and manpower required for walking prediction may be reduced.

In addition, as a simple test is performed, even a patient with a lot of difficulty in conducting a foot property information acquisition test may obtain actual walking prediction data.

1 is a block diagram of a walking ability prediction system according to an embodiment of the present invention.
2 is an exemplary view of a foot characteristic information acquisition unit.
3 is an exemplary view showing a foot depth image and foot characteristics obtained therefrom.
4 is an exemplary diagram illustrating spatiotemporal parameters sensed by a motion detector.
5 is a flowchart of a method for predicting walking ability using foot characteristic information according to an embodiment of the present invention.
6 is a graph showing experimental results of a walking prediction method according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. DETAILED DESCRIPTION The following detailed description, together with the accompanying drawings, is intended to describe exemplary embodiments of the present invention, and is not intended to represent the only embodiments in which the present invention may be practiced. The following detailed description includes specific details to provide a thorough understanding of the present invention. However, one skilled in the art will appreciate that the present invention may be practiced without these specific details. Certain terms used in the following description are provided to help understanding of the present invention, and the use of these specific terms may be changed to other forms without departing from the technical spirit of the present invention.

1 is a block diagram of a walking ability prediction system according to an embodiment of the present invention. 2 is an exemplary view of a foot characteristic information acquisition unit. 3 is an exemplary view showing a foot depth image and foot characteristics obtained therefrom. 4 is an exemplary diagram illustrating spatiotemporal parameters sensed by a motion detector.

1 to 4, the walking ability prediction system 10 includes a foot characteristic information acquisition unit 100, a motion detection unit 110, a database unit 120, a standard walking pattern information generation unit 130, and It includes a walking ability prediction unit 140.

The walking ability prediction system according to the embodiments may have an aspect that is entirely hardware or partially hardware and partially software. For example, the walking capability prediction system of the present specification and each unit included therein may collectively refer to a device for transmitting and receiving data of a specific format and content in an electronic communication method and software related thereto. In this specification, terms such as “part”, “module”, “server”, “system”, “device”, or “terminal” refer to a combination of hardware and software driven by the hardware. Is intended. For example, the hardware here may be a data processing device comprising a CPU or other processor. Also, software driven by hardware may refer to a running process, an object, an executable, a thread of execution, a program, or the like.

Also, each part of the walking ability prediction system is not intended to refer to a separate component that is physically separated. In FIG. 1, the foot characteristic information acquisition unit 100, the motion detection unit 110, the database unit 120, the standard walking pattern information generation unit 130, and the walking capability prediction unit 140 are separated blocks that are separated from each other. Although it is illustrated as, it is merely functionally classifying the devices constituting the walking capability prediction system by the operation executed by the device. Therefore, depending on the embodiment, the foot characteristic information acquisition unit 100, the motion detection unit 110, the database unit 120, the standard walking pattern information generation unit 130 and the walking capability prediction unit 140 may be partially or All of them may be integrated in the same single device, one or more may be implemented as separate devices that are physically separated from other parts, or components that are communicatively connected to each other under a distributed computing environment.

The foot characteristic information acquisition unit 100 includes at least a footrest 101, a camera 102, and a foot characteristic information generation unit 103. The foot characteristic information acquisition unit 100 may acquire a foot depth image of the foot 20 of the measurement target located on the footrest 101. As shown in FIG. 2, the camera 102 is located under the footrest 101 to acquire a foot depth image of the foot 20 of the measurement target located on the footrest 101. The camera 102 may be photographed toward the sole surface of the foot 100 of the measurement target located on the scaffold 101. The footrest 101 may be a transparent material or an opaque material. Also, the camera 102 may be a depth and color (RGBD) camera, or a depth camera, but is not limited thereto. In the following, an example of acquiring foot characteristic information based on a foot depth image obtained using a depth and color camera will be described.

The foot depth image photographed by the camera 102 is provided to the foot characteristic information generation unit 103, and the foot characteristic information generation unit 103 may perform a pre-processing operation of the acquired foot depth image. For example, the foot characteristic information generation unit 103 may filter noise of the foot depth image. After filtering, the foot characteristic information generation unit 103 may convert each pixel to a 3D point cloud. Thereafter, each pixel of the color image can be mapped to each pixel of the depth image to generate 3D point cloud color information.

3(a) shows a depth-colored foot depth image generated after the filtering operation described above.

The foot characteristic information generation unit 103 may acquire foot characteristic information from a foot depth image. Foot characteristic information is any information related to the sole, and may include arbitrary information about the shape, overall shape, and characteristics of each part of the sole. In one embodiment, the foot characteristic information may include a foot shape, a foot width, a position of the foot arch, a width of the foot arch, and an angle of the foot arch curve.

Specifically, the foot characteristic information generating unit 103 may extract a Medial Longitudinal Arch (MLA) line and a Lateral Longitudinal Arch (LLA) line from the foot depth image. Here, the MLA line is a line connecting the heel and the first metatarrsal joint. The first metatarsal joint may be the metatarsal joint of the thumb or index finger in the case of the right foot. The LLA line is a line connecting the heel and the second metatarsal joint. The second metatarsal joint may be the metatarsal joint of the middle finger or ring finger toe. Here, the heel, the first metatarsal bone, and the second metatarsal bone correspond to a part of the foot skeleton that touches the ground, and an arch connecting each point may be formed.

The foot characteristic information generation unit 103 may obtain the foot contact surface and the foot depth information from the foot depth image, and extract the MLA curve from the MLA line based on the information, and extract the LLA curve from the LLA line. .

3(b) is an exemplary view showing an MLA line. The foot characteristic information generation unit 103 may acquire foot characteristic information based on at least one of the MLA curve and the LLA curve. The foot characteristic information generating unit 103 acquires foot characteristic information (foot arch height, foot length, foot width, arch curve angle, etc.) which are parameters representing foot characteristics based on the MLA curve, and based on the LLA curve. Characteristic information can be obtained. Each foot characteristic information extracted in one embodiment may be combined according to a predetermined ratio. However, the present invention is not limited thereto, and the foot characteristic information generation unit 103 may extract foot characteristic information using an MLA curve or an LLA curve.

The foot characteristic information acquisition unit 100 may acquire foot characteristic information while the subject takes a static fixed posture. Here, the fixed posture may include at least one of a posture in which the measurement target is a knee bent at a predetermined angle (for example, 90 degrees), a posture with both knees extended, and a posture with one foot. The foot characteristic information acquisition unit 100 may acquire the above-described foot characteristic information in each fixed posture, and the amount of change in a parameter according to each posture may be calculated. However, the present invention is not limited thereto, and the foot characteristic information acquisition unit 100 may acquire the above-described foot characteristic information even in a position other than the fixed posture described above by the subject, and the foot characteristic while the subject takes a dynamic posture. Information can also be obtained. In some embodiments, the foot characteristic information acquisition unit 100 may further include a supporting member that assists a subject to easily take a posture, and a display member that guides acquisition of current foot characteristic information and displays a process of being acquired. have.

Foot characteristic information acquired by the foot characteristic information acquisition unit 100 may be provided to the database unit 120. Here, the foot characteristic information acquisition unit 100 may further collect the body information of the measurement target, such as the subject's age information, gender information, walking-related disease information, and leg body characteristics, in addition to the foot characteristic information, and collected body information May be included in the foot characteristic information and provided to the database unit 120.

The motion detection unit 110 detects a subject's body motion and extracts the subject's gait pattern information from the subject's body motion information. The motion detector 110 may be configured as a wearable motion capture system including an inertial measurement sensor (IMU), but is not limited thereto. The subject can freely perform actual walking in an environment in which the motion detector 110 is operated. The motion detection unit 110 detects and extracts a person's walking characteristics in a real environment, not a walking environment provided in a virtual space. The walking characteristic data extracted from the motion detector 110 may be sample data for generating standard walking pattern information.

The motion detection unit 110 may detect a subject's movement when actually walking, and may extract a subject's gait pattern information from the detected subject's body motion information. The gait pattern information of the extracted subject may include temporal parameters and spatial parameters. Specifically, as shown in FIG. 4, the motion detection unit 110 is a time parameter, a stride time (time between one HS time and the next HS time), and a step time (Step time) Time between time and HS time of other foot), Stance time (time between HS time of one foot and TO time), Swing time (time between TO time of one foot and HS time), single rim support Time (Single limb support time, time supported by one leg), Double limb support time (time supported by both feet), Cadence (Cadence, time between steps and steps), etc. Can be extracted. In addition, the motion detector 110 may extract stride length, step length, gait velocity, and the like as spatial parameters.

The walking pattern information of the subject in the motion detection unit 110 may be provided to the database unit 120.

The database unit 120 may store the gait pattern information of the subject according to the foot characteristic information of the subject.

Through the data collection process of the foot characteristic information acquisition unit 100 and the motion detection unit 110 described above, sample data related to the gait pattern information of the targeted subject can be constructed in the database 120.

The standard walking pattern information generation unit 130 analyzes a relationship between the subject's walking pattern information and the machine's foot characteristic information, and performs machine learning to generate standard walking pattern information corresponding to the target's foot characteristic information.

The standard gait pattern information generation unit 130 analyzes the relationship between the gait pattern information of the subject and the machine's gait pattern information based on the foot gait information and the actual gait pattern information built in the database unit 120. Through learning, standard walking pattern information corresponding to the subject's foot characteristic information may be generated. At this time, it is possible to generate a standard gait pattern for each subject's foot characteristics by considering all the factors of the above-mentioned foot characteristics, but in the case of a feature point or feature factor that is highly correlated with the extracted walking pattern, weight may be further assigned. In other words, according to the degree of correlation (correlation) with the gait pattern, standard gait pattern information may be generated after different weights are assigned to each of the factors of the subject's foot characteristic information.

The foot supports 50% to 60% of the load exerted by body weight, provides propulsive force during walking, and controls posture, so foot property information can have a close influence on human walking. In particular, the structural abnormality of the foot may cause abnormal pressure distribution when walking, thereby reducing walking ability and causing injury. Accordingly, it may be possible to predict an individual's gait pattern sufficiently with only foot characteristic information.

The standard walking pattern information generating unit 130 may perform machine learning by using the foot characteristic information as an input value and the subject's walking pattern information as an output value. The learning of the walking pattern may use an existing machine learning model method. In other words, SVR (Support Vector Machine Regression), which is a learning model that finds the answer with an efficient optimization problem using dimensional transformation, or Gaussian Mixture Regression, which is a learning model based on probability that predicts data distribution by assuming a data density distribution as a multivariate Gaussian. ) Or the EM technique, which is a learning model that finds the answer to the optimization problem by converting the data dimension to the essential low dimension (Principle Component Analysis), can be used.

In order to generate the above-described standard walking pattern database, the individual foot characteristics information and extracted walking patterns from N people are stored in the database unit 120. At this time, the foot characteristic information of N people is divided into M groups. Here, the exact M value can be determined by a clustering technique that divides into groups. When each group is determined in this way, foot characteristic information and a representative walking pattern are generated for each group. The foot characteristic information characteristic points of each group may be defined as an average value of the foot characteristic information characteristic points belonging to the group. The representative walking pattern of each group can be obtained through a function prediction model of walking patterns of foot characteristic information belonging to the group. The function prediction technique is largely divided into a linear technique and a nonlinear technique. Which method is used can select a model that is most suitable for each group of actual data.

In addition, the standard walking pattern information generation unit 130 imitates an information processing method of classifying objects after the human brain finds patterns in a large amount of data, and deep learning (deep learning) that the computer performs machine learning to discriminate objects. learning) to create a standard walking pattern database. That is, the abstract walking model using the standard walking pattern information generating unit 130 foot characteristic information as an input value and the subject's walking pattern information as an output value can be constructed. The standard gait pattern information generating unit 130 is a deep neural network model in which multiple hidden layers exist between the input layer and the output layer, and forms a connection pattern between neurons similar to the structure of the animal visual cortex. A deep learning model of one of the convolutional neural network model, a recurrent neural network model that builds up the neural network every moment over time, and a restricted Boltzmann machine that can learn the probability distribution for a set of inputs. Can be used. However, the above-described method is only an example, and the machine learning method according to an embodiment of the present invention is not limited thereto.

Through this machine learning learning process, the standard walking pattern information generating unit 130 may generate standard walking pattern information corresponding to foot characteristic information. The standard gait pattern information corresponding to the foot characteristic information may be newly accumulated in the database 120 or newly updated according to a certain period.

The walking ability prediction unit 140 predicts the walking ability of a new subject by using the generated standard walking pattern information.

The new subject is acquired only the foot characteristic information through the foot characteristic information acquisition unit 100. That is, the new subject does not need to acquire gait pattern information according to actual body movement through the motion detector 110. The walking ability prediction unit 140 may predict the walking ability of the new subject by substituting foot characteristic information of the new subject into the generated standard walking pattern information.

The walking ability prediction method according to an embodiment of the present invention can predict a person's walking ability and walking characteristics only by a test for acquiring simple foot characteristic information. That is, by using standard walking pattern information corresponding to the constructed foot characteristic information, it is possible to predict the actual walking ability of a new subject with a simple foot characteristic information acquisition test. Accordingly, input of equipment and manpower required for walking prediction may be reduced. In addition, since the patient proceeds with a simple test, it is possible to obtain real gait prediction data by performing a foot characteristic information acquisition test even for patients with many difficulties.

Hereinafter, a method for predicting walking ability using foot characteristic information according to an embodiment of the present invention will be described.

5 is a flowchart of a method for predicting walking ability using foot characteristic information according to an embodiment of the present invention. The method may be performed in the systems of FIGS. 1 to 4 described above, and FIGS. 1 to 4 may be referred to for description in this embodiment.

Referring to FIG. 5, in the method for predicting a walking ability using foot characteristic information according to an embodiment of the present invention, constructing standard walking pattern information corresponding to the foot characteristic information (S100), obtaining foot characteristic information of an evaluation target It includes a step (S110) and a step (S120) of predicting the walking ability of the evaluation target.

First, standard walking pattern information corresponding to foot characteristic information is constructed (S100).

Foot characteristic information is acquired from a plurality of subjects, and actual gait pattern information is obtained from the plurality of subjects to be databased. The foot characteristic information acquisition unit 100 acquires the foot depth image of the foot of the measurement target located on the scaffold, and acquires the foot characteristic information of the subject from the acquired foot depth image. The motion detection unit 110 detects a subject's body motion and extracts the subject's gait pattern information from the subject's body motion information. Through the data collection process of the foot characteristic information acquisition unit 100 and the motion detection unit 110, sample data related to the gait pattern information of each subject can be constructed in the database unit 120.

The standard walking pattern information generating unit 130 analyzes a relationship between the subject's walking pattern information and the machine's foot characteristic information, and performs machine learning to generate standard walking pattern information corresponding to the target's foot characteristics. The standard walking pattern information generating unit 130 may perform machine learning by using the foot characteristic information as an input value and the subject's walking pattern information as an output value. After the machine learning process, the standard walking pattern information generation unit 130 may generate standard walking pattern information for each foot characteristic information.

Next, foot characteristic information of the evaluation target is acquired (S110).

The foot characteristic information acquisition unit 100 acquires the foot depth image of the foot of the measurement target located on the scaffold, and acquires the foot characteristic information of the subject from the acquired foot depth image. Here, the foot characteristic information is arbitrary information related to the sole, and may include arbitrary information about the shape, the overall shape, and characteristics of each part of the sole. In one embodiment, the foot characteristic information may include information about at least one of a sole shape, a sole width, a position of a foot arch, a width of a foot arch, and an angle of a foot arch curve. Specifically, the foot characteristic information acquisition unit 100 may extract the MLA line and the LLA line from the foot depth image. In addition, the foot characteristic information acquisition unit 100 may acquire the depth information of the sole contact surface and the sole from the foot depth image, extract the MLA curve from the MLA line based on the information, and extract the LLA curve from the LLA line Can. The foot characteristic information acquisition unit 100 may extract foot characteristic information (foot arch height, foot length, foot width, arch curve angle, etc.) based on at least one of the MLA curve and the LLA curve. In addition, the foot characteristic information may further include body information such as age information of the subject, gender information, walking-related disease information, and leg shape characteristics. The new subject is acquired only the foot characteristic information through the foot characteristic information acquisition unit 100. That is, the new subject does not need to acquire gait pattern information according to actual body movement through the motion detector 110.

The walking ability of the evaluation target is predicted (S120).

The walking ability prediction unit 140 predicts the walking ability of a new subject by using the generated standard walking pattern information. The walking ability prediction unit 140 may predict the walking ability of the new subject by substituting foot characteristic information of the new subject into the generated standard walking pattern information. That is, in the method for predicting gait ability according to an embodiment of the present invention, the gait ability and gait characteristics of a subject may be predicted only by a simple foot characteristic information acquisition test. That is, by using standard walking pattern information corresponding to the constructed foot characteristic information, it is possible to predict the actual walking ability of a new subject with a simple foot characteristic information acquisition test. Accordingly, input of equipment and manpower required for walking prediction may be reduced. In addition, since it is a simple test, even a patient with a lot of difficulty in moving can obtain actual walking prediction data through performing a foot test.

6 is a graph showing experimental results of a walking prediction method according to an embodiment of the present invention.

FIG. 6 is a graph examining errors in gait prediction data and motion data when a subject actually walks according to an embodiment of the present invention. Using the constructed standard gait pattern information, the error between the gait prediction data predicting the gait ability of the evaluator and the actual gait data detecting the actual motion of the evaluator was examined using only the foot characteristic information obtained from the evaluator.

Here, Only LLA is the first embodiment in which the foot characteristic information is extracted only from the LLA curve and the standard walking pattern information is constructed by using it as an input value. Only MLA extracts the foot characteristic information only from the MLA curve and uses it as an input value. This is Embodiment 2 in which standard walking pattern information is constructed. LLA + MLA is Example 3 in which standard gait pattern information is constructed using both the LLA curve and MLA curve, and Foot + Body Parameters are constructed using standard gait pattern information using both foot characteristic information and subject's body information. Example 4. Foot + Body Parameters + Label classifies subjects into two groups, the general or athlete, and builds standard walking pattern information by using foot characteristics and body information of subjects as well as labels for the groups of subjects. This is Example 5.

The X-axis disclosed in FIGS. 6(a) to 6(f) shows each spatiotemporal parameter observed when the subject is walking (stride length (Stride_L), step length (Step_L), single rim support time (SLS_T), swing time (Swing_T) ), stance time (Stance_T, walking speed (Gait V)], and the Y-axis represents the error between the predicted data and the actual data for each space-time parameter. In addition, the last delimiter in the space-time parameter indicates the speed (F:fast, N:normal, S:slow) at which the subject operated.

As the input variable in modeling is varied through machine learning, more accurate modeling is possible, and it can be seen that the error gradually decreases according to modeling from Example 1 to Example 5. However, as the number of input variables increases, additional devices and personnel may be required to acquire them. When the MLA parameter and the LLA parameter were used as in Examples 1 and 2, respectively, it was observed that the accuracy was slightly lower compared to Examples 4 and 5. However, it was confirmed that Example 3, in which both the MLA parameter and the LLA parameter indicating the foot characteristics were applied, had a large error range compared to Example 4 and Example 5. That is, it can be confirmed that even if only the foot characteristic information acquired by the simple test is used, it is possible to sufficiently predict the walking ability of the subject. The walking ability prediction method according to an embodiment of the present invention enables prediction of the actual walking ability of a new subject by simply acquiring foot characteristic information, and accordingly, input of equipment and manpower required for walking prediction can be reduced. have.

Although described above with reference to the embodiments, the present invention should not be construed as being limited by these embodiments or the drawings, and those skilled in the art will think and scope of the present invention described in the following claims It will be understood that various modifications and changes can be made to the present invention without departing from the scope.

100: foot characteristic information acquisition unit
101: scaffolding
102: camera walking motion device
103: foot characteristic information generating unit
110: motion detector
120: database unit
130: standard walking pattern information generating unit
140: walking ability prediction unit

Claims (12)

A foot characteristic information acquisition unit acquiring a foot depth image of a subject located on the scaffold and acquiring foot characteristic information from the acquired foot depth image;
A motion detector for sensing the subject's body motion and extracting the subject's gait pattern information from the subject's body motion information;
A database unit for storing the walking pattern information of the target person according to the foot characteristic information of the target person;
A standard walking pattern information generation unit that analyzes a relationship between the target's foot characteristic information and the target's walking pattern information and performs machine learning to generate standard walking pattern information corresponding to the target's foot characteristics; And
A walking ability prediction system including a walking ability prediction unit that predicts the walking ability of the new subject by substituting foot characteristic information obtained from a new subject into the standard walking pattern information. According to claim 1,
The foot characteristic information includes a foot shape, a foot width, a foot arch position, a foot arch width, and a walking ability prediction system including an angle of the foot arch curve. According to claim 2,
The foot characteristic information acquisition unit,
MLA curve, LLA curve is extracted from the foot depth image,
A walking capability prediction system that acquires the foot characteristic information based on at least one of the MLA curve and the LLA curve. According to claim 1,
The subject's gait pattern information includes time parameters and spatial parameters,
The time parameters include stride time, step time, stance time, swing time, single rim support time, double rim support time and cadence,
The spatial parameter includes a stride length, a step length, and a walking speed prediction system. According to claim 4,
The standard walking pattern information generating unit uses the foot characteristic information as an input value, and a walking ability prediction system that performs machine learning using the target's walking pattern information as an output value. The method of claim 5,
The standard walking pattern information generating unit is a walking capability prediction system that generates the standard walking pattern information through deep learning. A walking ability prediction method performed in a walking ability prediction system,
The walking ability prediction system acquires a foot depth image of the subject located on the scaffold, acquires foot characteristic information from the acquired foot depth image, detects the subject's body motion, and subjects from the subject's body motion information By extracting the gait pattern information of the subject, and storing the gait pattern information of the target by foot characteristics information of the subject, analyze the relationship between the gait pattern information of the subject with respect to the foot characteristic information of the subject, machine learning to the subject Constructing standard walking pattern information corresponding to the characteristics of the foot;
The walking ability prediction system obtaining foot characteristic information from a new subject; And
And a step of estimating the gait ability of the new subject by substituting foot characteristic information of the new subject into the standard gait pattern information. The method of claim 7,
The foot characteristic information includes a foot shape, a foot width, a position of a foot arch, a width of a foot arch, and an angle of a foot arch curve. The method of claim 8,
The foot characteristic information acquisition unit,
MLA curve, LLA curve is extracted from the foot depth image,
A walking capability prediction method for acquiring the foot characteristic information based on at least one of the MLA curve and the LLA curve. The method of claim 7,
The subject's gait pattern information includes time parameters and spatial parameters,
The time parameters include stride time, step time, stance time, swing time, single rim support time, double rim support time and cadence,
The spatial parameter includes a stride length, step length, and walking speed. The method of claim 10,
The step of constructing the standard walking pattern information,
A method for predicting walking ability using the foot characteristic information as an input value and performing machine learning by using the subject's walking pattern information as an output value. The method of claim 11,
The step of constructing the standard walking pattern information,
A walking ability prediction method for generating the standard walking pattern information through deep learning.

Images