Human body balance ability obtaining method and system, computer device and medium
阅读说明：本技术 人体平衡能力获取方法及系统、计算机设备及介质 (Human body balance ability obtaining method and system, computer device and medium ) 是由 贾凡 姜立 汤丹 周莉 何盛一 于 2019-11-19 设计创作，主要内容包括：本发明公开一种人体平衡能力获取方法及系统、计算机设备及介质。该方法的一具体实施方式包括：根据用户行走时足底二维压力分布时序数据获取足底压力时间周期；将连续的至少一个时间周期的足底二维压力分布时序数据累加为三维压力数据；对所述三维压力数据进行处理得到足底压力立体数据；将所述足底压力立体数据输入已训练的体素回归网络进行特征提取,得到所述体素回归网络输出的所述用户的Berg平衡量表数值。该实施方式可准确地获取人体平衡能力。(The invention discloses a method and a system for acquiring human body balance capacity, computer equipment and a medium. One embodiment of the method comprises: obtaining a plantar pressure time period according to plantar two-dimensional pressure distribution time sequence data when a user walks; accumulating the two-dimensional pressure distribution time sequence data of the sole in at least one continuous time period into three-dimensional pressure data; processing the three-dimensional pressure data to obtain plantar pressure three-dimensional data; inputting the three-dimensional data of the plantar pressure into a trained voxel regression network for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network. The embodiment can accurately acquire the human body balance ability.)
1. A method for acquiring human body balance ability is characterized by comprising the following steps:
obtaining a plantar pressure time period according to plantar two-dimensional pressure distribution time sequence data when a user walks;
accumulating the two-dimensional pressure distribution time sequence data of the sole in at least one continuous time period into three-dimensional pressure data;
processing the three-dimensional pressure data to obtain plantar pressure three-dimensional data;
inputting the three-dimensional data of the plantar pressure into a trained voxel regression network for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network.
2. The method of claim 1, wherein the obtaining the plantar pressure time period according to the plantar two-dimensional pressure distribution time sequence data when the user walks further comprises:
summing the plantar pressure distribution data p (x, y) at the moment t to obtain the pressure distribution sum at the moment tWherein x and y are coordinates of sampling points of the plantar pressure sampling region;
obtaining a local maximum value of the pressure distribution sum and a corresponding moment;
and obtaining the plantar pressure time period to be delta T/N according to the time length delta T between the N +1 adjacent local maximum values.
3. The method of claim 2, wherein said obtaining a local maximum of a sum of pressure distributions further comprises:
determining a local pressure detection threshold, wherein the local pressure detection threshold is k x a sum of pressures measured when the user is standing still, wherein k is an empirical factor;
deriving the pressure distribution sum P (t) at the moment t to obtain an extreme point of the pressure distribution sum;
if only the extreme point exists in the preset time range corresponding to one extreme point, the extreme point is the local maximum, otherwise, the extreme points existing in the time range are compared, and the maximum extreme point is selected as the local maximum.
4. The method of claim 3, wherein accumulating the time series data of the two-dimensional pressure distribution of the sole for at least one consecutive time period into three-dimensional pressure data further comprises:
accumulating the plantar two-dimensional pressure distribution time sequence data p (x, y) in m time periods according to the sampling frames to obtain three-dimensional pressure data V (x, y, z), wherein z represents the pressure accumulation sum in m time periods at the (x, y) coordinate,
5. The method of claim 4, wherein said processing said three-dimensional pressure data to obtain plantar pressure stereo data further comprises:
sampling pressure at p × q sampling points in a plantar pressure sampling region, wherein p is the number of sampling points in the x direction of the sampling region, q is the number of sampling points in the y direction of the sampling region, and p: q is determined according to the proportion of the sole size in the x and y directions;
the z is taken as the maximum value z of the accumulated summaxCarrying out normalization and rounding;
all z after normalization and rounding are multiplied by a partition parameter r,
and forming a p × q × r three-dimensional matrix, wherein if the sampling point (u, v) does not sample pressure, the corresponding element value in the matrix is 0, and if the sampling point (u, v) samples pressure, the corresponding element value in the matrix is 1, so as to obtain the plantar pressure stereo data, wherein u is more than or equal to 1 and less than or equal to p, and v is more than or equal to 1 and less than or equal to q.
6. The method according to claim 5, wherein the inputting the plantar pressure stereo data into a trained voxel regression network for feature extraction, the obtaining the Berg balance scale value of the user output by the voxel regression network further comprises:
performing first convolution on the plantar pressure stereo data by using a cube with a convolution kernel of p '× q' × r 'to extract x-direction data characteristics, and obtaining sub-sampled primary three-dimensional data, wherein p', q ', r' are obtained by dividing p, q and r by a maximum common divisor respectively;
performing second convolution on the primary three-dimensional data by using a cube with a convolution kernel of p ' × q ' × r ' so as to extract data characteristics in the y direction and obtain sub-sampled secondary three-dimensional data;
performing third convolution on the secondary three-dimensional data by using a cube with a convolution kernel of p ' × q ' × r ', thereby extracting z-direction data characteristics and obtaining sub-sampled tertiary three-dimensional data;
performing maximum pooling on the three-level three-dimensional data to compress into one-dimensional data;
and outputting the Berg balance table value of the user by the one-dimensional data through a dense connection layer.
7. The method of claim 1, wherein the user's Berg balance scale value is the user's 2-ary Berg balance scale value.
8. The method of claim 1, wherein prior to inputting the plantar pressure stereo data into a trained voxel regression network for feature extraction, the method further comprises:
constructing a training set and a testing set according to the plantar pressure three-dimensional data of a plurality of sample users marked with the Berg balance scale values;
and training a neural network model according to the training set and testing according to the testing set to obtain the trained voxel regression network.
9. The method of claim 8, wherein training a neural network model according to the training set further comprises: solving the neural network model parameters by using an Adam algorithm.
10. A computer device comprising a processor, wherein the processor implements the method of any one of claims 1-9 when executing a program.
11. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-9.
12. A system for acquiring the balance ability of human body is characterized by comprising
The computer device of claim 10 and an array of pressure sensors distributed on a user's sole;
the pressure sensor array is used for acquiring sole two-dimensional pressure distribution time sequence data when a user walks and sending the data to the computer equipment.
Technical Field
The invention relates to the technical field of artificial intelligence. And more particularly, to a human balance ability acquisition method and system, a computer device, and a medium.
Background
The human body balance ability has extremely high value for preventing human body from falling down and clinical prognosis evaluation. At present, the evaluation mode of human body balance ability is mainly based on two-dimensional plantar pressure images acquired in the human body walking process, and gait features contained in the images are extracted through a neural network, so that an evaluation result is obtained. The gait features extracted in the above way can only reflect static gait features in the walking process of the human body, and the accuracy of the evaluation result is not sufficient.
Therefore, it is desirable to provide a new human balance ability acquisition method and system, computer device, and medium.
Disclosure of Invention
An object of the present invention is to provide a method and system for acquiring a human balance ability, a computer device, and a medium, which solve at least one of the problems of the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention provides a method for acquiring human body balance capacity in a first aspect, which comprises the following steps:
obtaining a plantar pressure time period according to plantar two-dimensional pressure distribution time sequence data when a user walks;
accumulating the two-dimensional pressure distribution time sequence data of the sole in at least one continuous time period into three-dimensional pressure data;
processing the three-dimensional pressure data to obtain plantar pressure three-dimensional data;
inputting the three-dimensional data of the plantar pressure into a trained voxel regression network for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network.
The method for acquiring the human body balance ability provided by the first aspect of the invention converts the time sequence data of two-dimensional pressure distribution of the sole formed by the two-dimensional sole pressure image into three-dimensional sole pressure stereo data, and performs characteristic extraction on the three-dimensional sole pressure data through the voxel regression network with the three-dimensional convolution kernel, so that the change process of the sole pressure in the human body walking process can be acquired based on the time sequence data, namely, the dynamic gait characteristics in the human body walking process contained in the three-dimensional sole pressure data are extracted through the voxel regression network, the dynamic balance ability of the human body in the motion process can be accurately, quantitatively and objectively evaluated, different dynamic gait characteristics can be accurately classified, and the balance ability of the human body can be accurately acquired.
Optionally, the obtaining of the plantar pressure time period according to the plantar two-dimensional pressure distribution time series data when the user walks further includes:
summing the plantar pressure distribution data p (x, y) at the moment t to obtain the pressure distribution sum at the moment t
obtaining a local maximum value of the pressure distribution sum and a corresponding moment;
and obtaining the plantar pressure time period to be delta T/N according to the time length delta T between the N +1 adjacent local maximum values.
According to the alternative mode, the plantar pressure time period in the walking process of the user can be rapidly and accurately acquired or divided.
Optionally, the obtaining a local maximum of the sum of the pressure distributions further comprises:
determining a local pressure detection threshold, wherein the local pressure detection threshold is k x a sum of pressures measured when the user is standing still, wherein k is an empirical factor;
deriving the pressure distribution sum P (t) at the moment t to obtain an extreme point of the pressure distribution sum;
if only the extreme point exists in the preset time range corresponding to one extreme point, the extreme point is the local maximum, otherwise, the extreme points existing in the time range are compared, and the maximum extreme point is selected as the local maximum.
This alternative way, the accuracy of the time period of the acquired plantar pressure can be guaranteed.
Optionally, the accumulating the time series data of the two-dimensional pressure distribution of the sole for at least one continuous time period into three-dimensional pressure data further comprises:
for m time periodsAccumulating the plantar two-dimensional pressure distribution time sequence data p (x, y) according to the sampling frame to obtain three-dimensional pressure data V (x, y, z), wherein z represents the pressure accumulation sum in m time periods at the (x, y) coordinate,wherein m is an integer of 1 or more.
Optionally, the processing the three-dimensional pressure data to obtain plantar pressure stereo data further includes:
sampling pressure at p × q sampling points in a plantar pressure sampling region, wherein p is the number of sampling points in the x direction of the sampling region, q is the number of sampling points in the y direction of the sampling region, and p: q is determined according to the proportion of the sole size in the x and y directions;
the z is taken as the maximum value z of the accumulated summaxCarrying out normalization and rounding;
all z after normalization and rounding are multiplied by a partition parameter r,
and forming a p × q × r three-dimensional matrix, wherein if the sampling point (u, v) does not sample pressure, the corresponding element value in the matrix is 0, and if the sampling point (u, v) samples pressure, the corresponding element value in the matrix is 1, so as to obtain the plantar pressure stereo data, wherein u is more than or equal to 1 and less than or equal to p, and v is more than or equal to 1 and less than or equal to q.
Optionally, the inputting the plantar pressure stereo data into a trained voxel regression network for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network further includes:
performing first convolution on the plantar pressure stereo data by using a cube with a convolution kernel of p '× q' × r 'to extract x-direction data characteristics, and obtaining sub-sampled primary three-dimensional data, wherein p', q ', r' are obtained by dividing p, q and r by a maximum common divisor respectively;
performing second convolution on the primary three-dimensional data by using a cube with a convolution kernel of p ' × q ' × r ' so as to extract data characteristics in the y direction and obtain sub-sampled secondary three-dimensional data;
performing third convolution on the secondary three-dimensional data by using a cube with a convolution kernel of p ' × q ' × r ', thereby extracting z-direction data characteristics and obtaining sub-sampled tertiary three-dimensional data;
performing maximum pooling on the three-level three-dimensional data to compress into one-dimensional data;
and outputting the Berg balance table value of the user by the one-dimensional data through a dense connection layer.
By adopting the optional mode, the voxel regression network can accurately and efficiently extract dynamic gait characteristics of the input plantar pressure three-dimensional data, accurately and efficiently classify the data and output Berg balance scale values capable of directly reflecting the human body balance capacity.
Optionally, the Berg balance scale value of the user is a 2-ary Berg balance scale value of the user.
This alternative may improve the accuracy of the retrieved Berg balance scale values for the user.
Optionally, before inputting the plantar pressure stereo data into the trained voxel regression network for feature extraction, the method further includes:
constructing a training set and a testing set according to the plantar pressure three-dimensional data of a plurality of sample users marked with the Berg balance scale values;
and training a neural network model according to the training set and testing according to the testing set to obtain the trained voxel regression network.
This alternative may ensure the validity of the trained voxel regression network.
Optionally, training the neural network model according to the training set further comprises: solving the neural network model parameters by using an Adam algorithm.
This alternative may further ensure the validity of the trained voxel regression network.
A second aspect of the invention provides a computer apparatus comprising a processor which, when executing a program, performs the method provided by the first aspect of the invention.
A third aspect of the invention provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the method provided by the first aspect of the invention.
A fourth aspect of the present invention provides a system for acquiring a human body balance ability, comprising
The computer apparatus and the array of pressure sensors distributed over the user's sole are provided by a second aspect of the invention;
the pressure sensor array is used for acquiring sole two-dimensional pressure distribution time sequence data when a user walks and sending the data to the computer equipment.
The invention has the following beneficial effects:
the technical scheme of the invention converts the plantar two-dimensional pressure distribution time sequence data formed by a two-dimensional plantar pressure image into three-dimensional plantar pressure three-dimensional data, and performs characteristic extraction on the plantar pressure three-dimensional data through the voxel regression network with a three-dimensional convolution kernel, so that the change process of the plantar pressure in the human body walking process can be obtained based on the time sequence data, namely, the dynamic gait characteristics in the human body walking process contained in the plantar pressure three-dimensional data are extracted through the voxel regression network, the dynamic balance capability of the human body in the motion process can be accurately, quantitatively and objectively evaluated, different dynamic gait characteristics can be accurately classified, and the balance capability of the human body can be accurately obtained.
Drawings
The following describes embodiments of the present invention in further detail with reference to the accompanying drawings;
fig. 1 shows a flowchart of a method for acquiring a human balance ability according to an embodiment of the present invention.
Fig. 2 is a data trend diagram of a human balance ability obtaining method according to an embodiment of the present invention.
Fig. 3 is a schematic structural diagram of a computer device provided in an embodiment of the present invention.
Detailed Description
In order to more clearly illustrate the invention, the invention is further described below with reference to preferred embodiments and the accompanying drawings. Similar parts in the figures are denoted by the same reference numerals. It is to be understood by persons skilled in the art that the following detailed description is illustrative and not restrictive, and is not to be taken as limiting the scope of the invention.
As shown in fig. 1 and 2, an embodiment of the present invention provides a human body balance ability acquisition method, including:
obtaining a plantar pressure time period according to plantar two-dimensional pressure distribution time sequence data when a user walks;
accumulating the two-dimensional pressure distribution time sequence data of the sole in at least one continuous time period into three-dimensional pressure data;
processing the three-dimensional pressure data to obtain plantar pressure three-dimensional data;
inputting the plantar pressure stereo data into a trained Voxel Regression Network (VRN) for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network.
In a typical human walking cycle, plantar pressure is at the joints from the subtalar, tarsometatarsal and falcate joints, respectively. Plantar pressure is dynamically distributed in a quasi-periodic manner, and asynchronous features affect the time-sequential distribution of each period. The method for acquiring the human body balance ability provided by the embodiment converts the sequence data of the two-dimensional pressure distribution of the sole formed by the two-dimensional sole pressure image into three-dimensional sole pressure stereo data, and performs characteristic extraction on the three-dimensional sole pressure stereo data through the voxel regression network with the three-dimensional convolution kernel, so that the change process of the sole pressure in the human body walking process can be acquired based on the sequence data, namely, the dynamic gait characteristics in the human body walking process contained in the three-dimensional sole pressure data are extracted through the voxel regression network, and further, the dynamic balance ability of the human body in the motion process can be accurately, quantitatively and objectively evaluated, different dynamic gait characteristics can be accurately classified, and the balance ability of the human body can be accurately acquired. The Berg balance scale is an evaluation index for semi-quantitatively evaluating the human body balance capacity, which is the most common in clinic, and a large number of scientific research documents and actual clinical proofs show that the Berg balance scale can not only evaluate the balance capacity of a stroke patient, but also predict the risk of falling. Therefore, the numerical value of the Berg balance scale is adopted in the embodiment to quantify the gait characteristics, and then the human body balance capability is represented.
In some optional implementations of this embodiment, the obtaining the plantar pressure time period according to the plantar two-dimensional pressure distribution time series data when the user walks further includes:
summing the plantar pressure distribution data p (x, y) at the moment t to obtain the pressure distribution sum at the moment t
obtaining a local maximum value of the pressure distribution sum and a corresponding moment;
the time period of the plantar pressure is obtained as delta T/N according to the time length delta T between the N +1 adjacent local maximum values, and it should be noted that the premise of the time period delta T/N is that the human body walks at a substantially constant speed, which is a common requirement when acquiring plantar pressure images for acquiring gait characteristics.
By the implementation mode, the plantar pressure time period in the walking process of the user can be rapidly and accurately acquired or divided.
In some optional implementations of this embodiment, the obtaining a local maximum of the sum of the pressure distributions further includes:
determining a local pressure detection threshold, wherein the local pressure detection threshold is k × the sum of pressures measured when a user stands still, where k is an empirical factor, in a specific example, the sum of sole pressures measured when a human body stands still is f, and the local pressure detection threshold can be set to 1.8 × f, that is, k is 1.8, it can be understood that the maximum sole pressure value when the user stands on one foot is 2f, and according to experience, the sole pressure can be determined as the local maximum value as long as the sole pressure exceeds 1.8f during walking;
deriving the pressure distribution sum P (t) at the moment t to obtain an extreme point of the pressure distribution sum;
if only the extreme point exists in a preset time range (for example, 200ms) corresponding to one extreme point, the extreme point is the local maximum, otherwise, the extreme points existing in the time range are compared, and the maximum extreme point is selected as the local maximum.
The accuracy of the obtained plantar pressure time period can be ensured based on the implementation mode of the peak detection algorithm.
In some optional implementations of this embodiment, the accumulating the temporal sequence data of the two-dimensional pressure distribution of the sole for at least one consecutive time period into the three-dimensional pressure data further includes:
accumulating the plantar two-dimensional pressure distribution time sequence data p (x, y) in m time periods according to the sampling frames to obtain three-dimensional pressure data V (x, y, z), wherein z represents the pressure accumulation sum in m time periods at the (x, y) coordinate,
In some optional implementations of this embodiment, the processing the three-dimensional pressure data to obtain plantar pressure stereo data further includes:
sampling pressure at p × q sampling points in a plantar pressure sampling region, wherein p is the number of sampling points in the x direction of the sampling region, q is the number of sampling points in the y direction of the sampling region, and p: q is determined by the ratio of the sole dimensions in the x and y directions, in one particular example, p: q is set to 5: 2, further, the value of p is 125, and the value of q is 50;
the z is taken as the maximum value z of the accumulated summaxGo on to unityMelting and rounding;
all z after normalization and rounding are multiplied by a partition parameter r, which in one specific example takes the value of 100,
forming a p × q × r three-dimensional matrix (in the following specific example, a 125 × 50 × 100 three-dimensional matrix is formed), wherein if the sampling point (u, v) does not sample the pressure, the corresponding element value in the matrix is 0, and if the sampling point (u, v) samples the pressure, the corresponding element value in the matrix is 1, thereby obtaining the plantar pressure stereo data, wherein u is greater than or equal to 1 and less than or equal to p, and v is greater than or equal to 1 and less than or equal to q.
In some optional implementations of this embodiment, as shown in fig. 2, the inputting the plantar pressure stereo data into a trained voxel regression network for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network further includes:
performing first convolution on the plantar pressure stereo data by using a cube with a convolution kernel of p ' × q ' × r ' to extract x-direction data characteristics, and obtaining sub-sampled primary three-dimensional data, wherein p ', q ', r ' are obtained by dividing p, q, r by a maximum common divisor respectively, and are connected with a specific example of a 125 × 50 × 100 three-dimensional matrix, and values of p ', q ', r ' are 5, 2 and 4 respectively;
performing second convolution on the primary three-dimensional data by using a cube with a convolution kernel of p ' × q ' × r ' so as to extract data characteristics in the y direction and obtain sub-sampled secondary three-dimensional data;
performing third convolution on the secondary three-dimensional data by using a cube with a convolution kernel of p ' × q ' × r ', thereby extracting z-direction data characteristics and obtaining sub-sampled tertiary three-dimensional data;
performing maximum pooling on the three-level three-dimensional data to compress into one-dimensional data;
and outputting the Berg balance table value of the user by the one-dimensional data through a dense connection layer.
By adopting the implementation mode, the voxel regression network can accurately and efficiently extract dynamic gait characteristics of the input three-dimensional data of the plantar pressure, accurately and efficiently classify the three-dimensional data, and output the Berg balance scale value which can directly reflect the human body balance capacity.
In some optional implementations of this embodiment, the Berg balance scale value of the user is a 2-ary Berg balance scale value of the user.
This implementation may improve the accuracy of the retrieved Berg balance scale values for the user. In one specific example, the user's Berg balance table value may be set to a 6-bit 2-ary value, where each bit 2-ary value corresponds to one output dimension of the dense connection layer.
In some optional implementations of this embodiment, before inputting the plantar pressure stereo data into a trained voxel regression network for feature extraction, the method further includes:
constructing a training set and a testing set according to the plantar pressure three-dimensional data of a plurality of sample users marked with the Berg balance scale values;
and training a neural network model according to the training set and testing according to the testing set to obtain the trained voxel regression network.
The implementation mode can ensure the effectiveness of the trained voxel regression network.
In some optional implementations of this embodiment, training the neural network model according to the training set further comprises: solving the neural network model parameters by using an Adam algorithm.
The implementation mode can further ensure the effectiveness of the trained voxel regression network.
In one particular example of the use of the invention,
the training data consisted of three-dimensional data of plantar pressure of 1500 sample users as subjects, to which Berg balance scale values were assigned, wherein the Berg balance scale values of the sample users were evaluated according to a standard Berg balance scale table, and the Berg balance scale values of the sample users were previously converted into 2-ary values.
After the training data are obtained, a three-dimensional voxel regression network is adopted to build a neural network model, data of 1000 sample users are randomly used as a training set, data of the rest 500 sample users are used as a test set, the neural network model is trained according to the training set, and the test is carried out according to the test set, so that the voxel regression network is obtained. In the training process, the Adam algorithm is used for solving the neural network model parameters, the training iteration number is 500, and the parameters are carried out in the iteration process so as to realize the performance. The Adam algorithm (Adaptive motion Estimation) can calculate an Adaptive learning rate of each parameter during the network training process, which not only stores the exponentially decaying average of the AdaDelta previous squared gradient, but also maintains the previous gradient.
As shown in fig. 2, the voxel regression network is a neural network structure having a stereo convolution kernel. The depth of the neural network is reduced with the increase of the convolutional layer, and the trend of gradual convergence is shown. And each layer of convolution nerve layer utilizes convolution kernel convolution to extract features in the information compression process. In this example, three convolutional neural layers are provided in total for extracting features in the x direction, the y direction, and the z direction, respectively. In this example, the 3D neural convolution kernel samples, with the three layers of the convolutional neural layer convolution kernels each arranged as a 5 x 2 x 4 cube, in accordance with the typical size and timing resolution of the plantar pressure image. Firstly, performing convolution by using a 5 multiplied by 2 multiplied by 4 cube convolution kernel of a first convolution nerve layer to obtain data characteristics in the x direction, finding a corresponding central point and establishing primary sub-sampled three-dimensional data; then, obtaining y-axis direction data characteristics by utilizing convolution of a 5 multiplied by 2 multiplied by 4 cube convolution kernel of a second convolution nerve layer, finding a corresponding central point and establishing sub-sampled secondary three-dimensional data; then, obtaining z-direction data characteristics by utilizing convolution of a 5 multiplied by 2 multiplied by 4 cube convolution kernel of a third convolution nerve layer, finding a corresponding central point and establishing sub-sampled three-level three-dimensional data; finally, the three-level three-dimensional data is compressed into one-dimensional data by utilizing a maximum pooling layer (Maxpooling), and the Berg balance table value of the 2-system of the user is output through a Dense Connection layer (Dense Connection).
Another embodiment of the present invention provides a human body balance ability acquisition system, including:
a computer device and an array of pressure sensors distributed on a user's sole;
the pressure sensor array is used for acquiring plantar two-dimensional pressure distribution time sequence data when a user walks and sending the plantar two-dimensional pressure distribution time sequence data to the computer equipment;
the computer device comprises a processor, and the processor executes a program to realize the human balance ability acquisition method provided by the embodiment.
Wherein a processor of the computer device may be described as being provided with a plurality of modules, for example, the processor includes a dividing module, a converting module, and a classifying module;
the dividing module is used for acquiring a plantar pressure time period according to plantar two-dimensional pressure distribution time sequence data when a user walks;
the conversion module is used for accumulating the two-dimensional pressure distribution time sequence data of the sole in at least one continuous time period into three-dimensional pressure data and processing the three-dimensional pressure data to obtain three-dimensional sole pressure data;
and the classification module is used for inputting the three-dimensional data of the plantar pressure into a trained voxel regression network for feature extraction to obtain the Berg balance scale value of the user output by the voxel regression network.
It should be noted that the principle and the working flow of the system for acquiring human body balance ability provided in this embodiment are similar to those of the method for acquiring human body balance ability, and reference may be made to the above description for relevant points, which are not described herein again.
As shown in fig. 3, a computer system suitable for implementing the computer device in the body balance ability acquisition system provided by the present embodiment includes a central processing module (CPU) that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) or a program loaded from a storage section into a Random Access Memory (RAM). In the RAM, various programs and data necessary for the operation of the computer system are also stored. The CPU, ROM, and RAM are connected thereto via a bus. An input/output (I/O) interface is also connected to the bus.
An input section including a keyboard, a mouse, and the like; an output section including a speaker and the like such as a Liquid Crystal Display (LCD); a storage section including a hard disk and the like; and a communication section including a network interface card such as a LAN card, a modem, or the like. The communication section performs communication processing via a network such as the internet. The drive is also connected to the I/O interface as needed. A removable medium such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive as necessary, so that a computer program read out therefrom is mounted into the storage section as necessary.
In particular, the processes described in the above flowcharts may be implemented as computer software programs according to the present embodiment. For example, the present embodiments include a computer program product comprising a computer program tangibly embodied on a computer-readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication section, and/or installed from a removable medium.
The flowchart and schematic diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to the present embodiments. In this regard, each block in the flowchart or schematic diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the schematic and/or flowchart illustration, and combinations of blocks in the schematic and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
On the other hand, the present embodiment also provides a nonvolatile computer storage medium, which may be the nonvolatile computer storage medium included in the apparatus in the foregoing embodiment, or may be a nonvolatile computer storage medium that exists separately and is not assembled into a terminal. The non-volatile computer storage medium stores one or more programs that, when executed by a device, cause the device to: obtaining a plantar pressure time period according to plantar two-dimensional pressure distribution time sequence data when a user walks; accumulating the two-dimensional pressure distribution time sequence data of the sole in at least one continuous time period into three-dimensional pressure data; processing the three-dimensional pressure data to obtain plantar pressure three-dimensional data; inputting the three-dimensional data of the plantar pressure into a trained voxel regression network for feature extraction, and obtaining the Berg balance scale value of the user output by the voxel regression network.
In the description of the present invention, it is to be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
It should be understood that the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention, and it will be obvious to those skilled in the art that other variations and modifications can be made on the basis of the above description, and all embodiments cannot be exhaustive, and all obvious variations and modifications belonging to the technical scheme of the present invention are within the protection scope of the present invention.
- 上一篇：一种医用注射器针头装配设备
- 下一篇：信号采集传感器阵列、电子设备及垫子