Gait health monitoring and analysis method and system

Abstract

The present invention relates to a gait health monitoring and analysis method and system, and relates to the field of gait health monitoring. The method includes: when the current monitored person completes walking in a walking passage with a set passage length, a posture analysis model is used to intelligently analyze the gait type of the current monitored person based on various visual data during the walking process of the current monitored person, various configuration information of gait health monitoring, and multiple physiological information of the current monitored person; the gait type of the current monitored person is converted into the corresponding disease type of the current monitored person. The present invention also relates to a gait health monitoring and analysis system. Through the present invention, a specific gait health monitoring scenario can be built, and the gait-related disease type of the monitored person can be directly given by designing an artificial intelligence model with a customized structure, so as to facilitate the direct use of the relevant technical solutions in various health monitoring departments, and facilitate the barrier-free use of the relevant technical solutions in industrial operations.

Description

Technical Field

[0001] The present invention relates to the field of gait health monitoring, and in particular to a gait health monitoring and analysis method and system. Background Art

[0002] Although everyone has their own unique walking posture, some common gaits can still reflect people's physical problems, that is, a person's health status can be seen from their walking posture. For example, when a person walks with a limp, in addition to congenital "long and short legs", acquired diseases such as scoliosis may cause this limp. At this time, this person needs to be treated as a suspected patient of scoliosis for in-depth screening and exclusion of subsequent diseases.

[0003] For example, the Chinese invention patent publication CN 117379038A proposes a gait health monitoring and analysis system, which includes a basic data layer for storing user information and system data; a user interaction layer for receiving user interaction instructions, sending a processing request to the business response layer according to the interaction instructions, and visually displaying the processing results returned by the business response layer; the business response layer for calling the core function layer according to the processing request to perform gait health monitoring and analysis on user information and system data to obtain processing results; the core function layer for collecting user information by searching and connecting to the gait monitoring device, inputting the user information into the gait abnormality judgment model and the disease probability prediction model, and generating a health report; the gait abnormality judgment model and the disease probability prediction model update the model parameters based on the learning method. The present invention optimizes the internal model of the system based on the learning method, can improve privacy protection, and can be widely used in the field of smart medical technology.

[0004] For example, the Chinese invention patent publication CN 111436940A proposes a gait health assessment method and device. The method includes: obtaining a gait pressure signal, identifying the gait phase based on the gait pressure signal using fuzzy logic reasoning rules; recording the phase sequence of the gait phase; calculating the gait phase duration and the overall gait cycle according to the phase sequence interval; based on the Perry model standard gait sequence, counting the abnormal gait cycle that deviates from the standard gait according to the medical standard phase duration; obtaining the abnormal cycle proportion based on the abnormal gait cycle and the overall gait cycle, and using the abnormal cycle proportion as the basis for gait health evaluation. The present invention applies a fuzzy logic reasoning system in the gait phase detection process to achieve smooth and continuous gait phase detection, uses trapezoidal and triangular membership functions in the output membership function, does not preset the formation of an individual gait phase sequence, fully considers the internal differences of individual gait, and is suitable for gait assessment needs of various groups of people.

[0005] Although the above-mentioned technical solutions all use artificial intelligence models to analyze personal gait data to provide health reports or gait health evaluation data, the above-mentioned artificial intelligence models are too biased towards mathematical algorithms, and have not built a specific gait health monitoring scenario. The health assessment data provided is also preliminary, and it is impossible to directly provide the type of gait-related diseases of the monitored person. As a result, the technical solutions in the existing technology cannot be directly put into use and cannot be effectively industrialized. Summary of the invention

[0006] In order to solve the technical defects in the prior art, the present invention provides a gait health monitoring and analysis method and system, which can build a specific gait health monitoring scenario. By designing an artificial intelligence model with a customized structure, on the basis of comprehensive and sufficient basic data, and most importantly, on the basis of introducing multiple visual basic data, the type of gait-related diseases of the monitored person is directly given, thereby facilitating the direct use of relevant technical solutions in various health monitoring departments, thereby enhancing the industrialization value of relevant technical solutions.

[0007] According to a first aspect of the present invention, a gait health monitoring and analysis method is provided, the method comprising:

[0008] Acquire various configuration information of gait health monitoring, wherein the gait health monitoring adopts a walking passage with a set passage length and two monitoring cameras with the same camera parameters are respectively arranged on the side and in front of the walking passage, and the various configuration information of the gait health monitoring includes the set passage length, the set walking passage width, the front monitoring camera distance, the side monitoring camera distance, and multiple camera parameters of the monitoring camera;

[0009] Obtain visual monitoring content corresponding to a frame of monitoring screen output by each monitoring camera device, wherein the visual monitoring content corresponding to each frame of monitoring screen is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value, and blue channel value of each pixel point in the area occupied by the human body in the imaging screen of the frame;

[0010] Performing multiple training on the deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer;

[0011] When the currently monitored person completes walking in a walking passage of a set length, the visual monitoring contents corresponding to the respective frames of imaging images captured by the two monitoring cameras during the currently monitored person's walking, the various configuration information of gait health monitoring, and the multiple physiological information of the currently monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the currently monitored person output by the posture analysis model;

[0012] Converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person;

[0013] Among them, the various configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, including: the front monitoring camera device distance is the distance from the monitoring camera device set in front of the walking channel to the end of the walking channel, and the side monitoring camera device distance is the minimum value of the two distances from the monitoring camera device set on the side of the walking channel to the two sides of the walking channel respectively;

[0014] The gait type of the current monitored person output by the posture analysis model is one of unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait, and scissor-step walking gait;

[0015] Among them, the gait type of the current monitored person output by the received posture analysis model is converted into the corresponding disease type of the current monitored person, including: gait types of unilateral hand not swinging, swaying left and right, small steps and slow speed, drunken gait, stumbling gait, unilateral crooked gait and scissors step walking gait, which correspond to the disease types of Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian artery steal syndrome and precursor of cerebral infarction one by one.

[0016] According to a second aspect of the present invention, a gait health monitoring and analysis system is provided, the system comprising:

[0017] An information analysis device, used to obtain various configuration information of gait health monitoring, wherein the gait health monitoring adopts a walking channel with a set passage length and two monitoring cameras with the same camera parameters are respectively arranged on the side and in front of the walking channel, and the various configuration information of the gait health monitoring includes the set passage length, the set walking channel width, the front monitoring camera distance, the side monitoring camera distance and multiple camera parameters of the monitoring camera;

[0018] A visual input device is used to obtain visual monitoring content corresponding to a frame of monitoring screen output by each monitoring camera device, and the visual monitoring content corresponding to each frame of monitoring screen is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the area occupied by the human body in the imaging screen of the frame;

[0019] A network training device, used to perform multiple training on a deep neural network to obtain a deep neural network after multiple training and output it as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer;

[0020] The intelligent analysis device is respectively connected with the information analysis device, the visual input device and the network training device, and is used to synchronously input the visual monitoring contents corresponding to the frames of imaging images taken by the two monitoring cameras during the walking process of the current monitored person, the various configuration information of gait health monitoring and the multiple physiological information of the current monitored person into the posture analysis model when the current monitored person completes walking in the walking passage of the set passage length, and execute the posture analysis model to obtain the gait type of the current monitored person output by the posture analysis model;

[0021] a disease conversion device, connected to the intelligent analysis device, for converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person;

[0022] Among them, the various configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, including: the front monitoring camera device distance is the distance from the monitoring camera device set in front of the walking channel to the end of the walking channel, and the side monitoring camera device distance is the minimum value of the two distances from the monitoring camera device set on the side of the walking channel to the two sides of the walking channel respectively;

[0023] The gait type of the current monitored person output by the posture analysis model is one of unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait, and scissor-step walking gait;

[0024] Among them, the gait type of the current monitored person output by the received posture analysis model is converted into the corresponding disease type of the current monitored person, including: gait types of unilateral hand not swinging, swaying left and right, small steps and slow speed, drunken gait, stumbling gait, unilateral crooked gait and scissors step walking gait, which correspond to the disease types of Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian artery steal syndrome and precursor of cerebral infarction one by one.

[0025] It can be seen that the present invention has at least the following four outstanding substantive features:

[0026] First: When the current monitored person completes walking in the walking channel for gait health monitoring, the posture analysis model is used to intelligently analyze the walking gait type of the current monitored person based on the targeted screened visual contents and various configuration information, and the current disease type of the current monitored person is determined based on the walking gait type of the current monitored person obtained by intelligent analysis, thereby improving the automation and intelligence level of gait health monitoring;

[0027] Second, the deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, wherein the multiple hidden layers are located between the single output layer and the single input layer, the number of times the deep neural network is trained is monotonically positively correlated with the number of pixels of the imaging screen of the monitoring camera device, and the number of hidden layers of the deep neural network is proportional to the set travel length, thereby completing the structural customization of the posture analysis model and ensuring the effectiveness and stability of the intelligent analysis results;

[0028] The third part: the various visual contents and configuration information selected in a targeted manner are specifically the visual monitoring contents corresponding to the frames of imaging images taken by the two monitoring cameras during the walking process of the current monitored person, the various configuration information of gait health monitoring, and multiple physiological information of the current monitored person, wherein the visual monitoring contents corresponding to each frame of imaging image are the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value, and blue channel value of each pixel point in the area occupied by the human body in the frame of imaging image, the various configuration information of gait health monitoring includes the set passage length, the set walking channel width, the front monitoring camera device distance, the side monitoring camera device distance, and multiple camera parameters of the monitoring camera device, and the multiple physiological information of the current monitored person are the age information, gender information, height information, and weight information of the current monitored person;

[0029] Fourthly, in each training of the deep neural network, the gait type of a certain monitored person known after the monitored person completed walking in a walking passage of a set length in history is used as the output content of the deep neural network, and each frame of imaging images taken by two monitoring cameras during the walking process of the certain monitored person corresponding to each visual monitoring content, each configuration information of gait health monitoring and multiple physiological information of the certain monitored person are used as the input content of the deep neural network to perform this training, thereby ensuring the training effect of each training of the deep neural network. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The embodiments of the present invention will be described below with reference to the accompanying drawings, wherein:

[0031] Figure 1 It is a technical flow chart of the gait health monitoring and analysis method and system according to the present invention.

[0032] Figure 2 This is a flowchart of the steps of the gait health monitoring and analysis method according to Embodiment 1 of the present invention.

[0033] Figure 3 This is a flowchart of the steps of the gait health monitoring and analysis method according to Embodiment 2 of the present invention.

[0034] Figure 4 This is a flowchart of the steps of the gait health monitoring and analysis method according to Embodiment 3 of the present invention.

[0035] Figure 5 1 is an internal structure diagram of a gait health monitoring and analysis system according to Embodiment 4 of the present invention.

[0036] Figure 6 1 is an internal structure diagram of a gait health monitoring and analysis system according to Embodiment 5 of the present invention.

[0037] Figure 7 1 is an internal structure diagram of a gait health monitoring and analysis system according to Embodiment 6 of the present invention. DETAILED DESCRIPTION

[0038] like Figure 1 As shown, a technical flow chart of the gait health monitoring and analysis method and system according to the present invention is given.

[0039] like Figure 1 As shown, the specific technical process of the present invention is as follows:

[0040] Step 1: Building a specific gait health monitoring scenario, which is used to install relevant monitoring equipment and perform intelligent gait analysis on each user who enters the gait health monitoring scenario;

[0041] Specifically, in the constructed gait health monitoring scene, a walking passage with a set passage length and a set walking passage width is laid, and two monitoring cameras with the same camera parameters are respectively set on the side and front of the walking passage, and the installation position of each monitoring camera is fixed;

[0042] Step 2: Design a customized posture analysis model for the constructed gait health monitoring scenario. Different gait health monitoring scenarios correspond to posture analysis models with different structures.

[0043] For example, the structural customization of the posture analysis model corresponding to the gait health monitoring scenario is as follows:

[0044] Aspect A: performing multiple training on the deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, wherein the number of times the deep neural network is trained is monotonically positively correlated with the number of pixels of the imaging screen of the monitoring camera device of the gait health monitoring scene;

[0045] Aspect B: The number of hidden layers of the deep neural network is proportional to the set travel length of the gait health monitoring scenario;

[0046] Aspect C: Each training process of the deep neural network is designed specifically to ensure the training effect of each deep neural network;

[0047] Specifically, in each training of the deep neural network, the gait type of a certain monitored person after the monitored person completes walking in a walking passage of a set length in history is used as the output content of the deep neural network, and the visual monitoring contents corresponding to each frame of imaging images taken by two monitoring camera devices during the walking process of the certain monitored person, various configuration information of gait health monitoring, and multiple physiological information of the certain monitored person are used as the input content of the deep neural network to execute this training process;

[0048] In this way, the structural customization of the posture analysis model is completed, thereby ensuring the effectiveness and stability of the intelligent analysis results;

[0049] Step 3: Targeted selection of various visual contents and configuration information for intelligent analysis of gait types to ensure the adequacy and comprehensiveness of basic data;

[0050] Specifically, the various visualization contents and configuration information are the various visual monitoring contents corresponding to the imaging frames captured by the two monitoring cameras during the walking process of the current monitored person, the various configuration information of gait health monitoring, and the multiple physiological information of the current monitored person;

[0051] The visual monitoring content corresponding to each frame of the imaging picture is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the area occupied by the human body in the imaging picture of the frame, and the various configuration information of gait health monitoring includes the set passage length, the set walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, and multiple physiological information of the current monitoring person, including the age information, gender information, height information and weight information of the current monitoring person;

[0052] Therefore, if Figure 1 As shown, sufficient and comprehensive basic data are obtained for intelligent analysis of gait types;

[0053] Step 4: Using the customized posture analysis model designed in step 2, based on the various visualization contents and configuration information selected in step 3, complete the intelligent gait type analysis of the user in the gait health monitoring scenario currently built in step 1;

[0054] Step 5: Determine the user's current disease type based on the user's gait type obtained by intelligent analysis in step 4;

[0055] Specifically, the gait types of unilateral hand not swinging, swaying from side to side, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait and scissor-step walking gait correspond to the disease types of Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian artery steal syndrome and cerebral infarction precursor respectively;

[0056] It can be seen that through the coordinated operation of the above five steps, the automation and intelligence levels of gait health monitoring have been improved. At the same time, the type of gait-related diseases of the monitored person can be directly given, which facilitates the direct use of relevant technical solutions in various health monitoring departments.

[0057] The key points of the present invention are: the specific construction of the gait health monitoring scene, the customized structural design of the posture analysis model, the targeted screening of various visual contents and configuration information for performing intelligent analysis of user gait types, and the one-to-one correspondence between various gait types and various disease types.

[0058] The gait health monitoring and analysis method and system of the present invention will be specifically described below by way of embodiments.

[0059] Example 1

[0060] Figure 2 This is a flowchart of the steps of the gait health monitoring and analysis method according to Embodiment 1 of the present invention.

[0061] like Figure 2 As shown, the gait health monitoring and analysis method comprises the following steps:

[0062] Step S11: Acquire various configuration information of gait health monitoring, wherein the gait health monitoring adopts a walking channel with a set passage length and two monitoring cameras with the same camera parameters are respectively set on the side and front of the walking channel, and the various configuration information of the gait health monitoring includes the set passage length, the set walking channel width, the front monitoring camera distance, the side monitoring camera distance, and multiple camera parameters of the monitoring camera;

[0063] For example, multiple different information acquisition units may be selected to respectively obtain a set passage length, a set walking channel width, a front monitoring camera device distance, a side monitoring camera device distance, and multiple camera parameters of the monitoring camera device;

[0064] Step S12: Obtain visual monitoring content corresponding to a monitoring picture frame output by each monitoring camera device, wherein the visual monitoring content corresponding to each monitoring picture frame is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the human body occupied area in the imaging picture frame;

[0065] Specifically, the visual monitoring content corresponding to each frame of monitoring picture is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the human body occupied area in the frame of imaging picture, including: each pixel point has an R channel value, i.e., a red channel value, a G channel value, i.e., a green channel value and a B channel value, i.e., a blue channel value in the RGB color space;

[0066] Step S13: performing multiple training on the deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer, and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer;

[0067] For example, a numerical simulation mode may be selected to implement simulation and testing of a data processing process of performing multiple trainings on a deep neural network to obtain a deep neural network after multiple trainings and outputting the deep neural network as a posture analysis model;

[0068] Step S14: when the currently monitored person completes walking in the walking passage of the set passage length, the visual monitoring contents corresponding to the respective frames of imaging images captured by the two monitoring cameras during the walking process of the currently monitored person, the various configuration information of gait health monitoring, and the multiple physiological information of the currently monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the currently monitored person output by the posture analysis model;

[0069] Step S15: converting the gait type of the currently monitored person output by the received posture analysis model into the corresponding disease type of the currently monitored person;

[0070] Among them, the various configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, including: the front monitoring camera device distance is the distance from the monitoring camera device set in front of the walking channel to the end of the walking channel, and the side monitoring camera device distance is the minimum value of the two distances from the monitoring camera device set on the side of the walking channel to the two sides of the walking channel respectively;

[0071] The gait type of the current monitored person output by the posture analysis model is one of unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait, and scissor-step walking gait;

[0072] The gait type of the current monitored person output by the received posture analysis model is converted into the corresponding disease type of the current monitored person, including: unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait and scissor-step walking gait, which correspond to Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian steal syndrome and precursor of cerebral infarction respectively;

[0073] The deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The number of times the deep neural network is trained is monotonically positively correlated with the number of pixels in the imaging picture of the monitoring camera device;

[0074] For example, the number of times the deep neural network is trained is monotonically positively correlated with the number of pixels of the imaging screen of the surveillance camera device, including: when the number of pixels of the imaging screen of the surveillance camera device is 1024×720, the corresponding number of times the deep neural network is trained is 100, when the number of pixels of the imaging screen of the surveillance camera device is 1920×1080, the corresponding number of times the deep neural network is trained is 150, when the number of pixels of the imaging screen of the surveillance camera device is 3840×2160, the corresponding number of times the deep neural network is trained is 200, and so on;

[0075] The deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The number of hidden layers of the deep neural network is proportional to the set travel length;

[0076] For example, the number of hidden layers of the deep neural network is proportional to the set travel length, including: when the travel length is set to 30 meters, the number of hidden layers of the deep neural network is 3, when the travel length is set to 40 meters, the number of hidden layers of the deep neural network is 4, when the travel length is set to 50 meters, the number of hidden layers of the deep neural network is 5, and so on;

[0077] And wherein, the deep neural network is trained multiple times to obtain a deep neural network after multiple trainings and output as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The following include: in each training performed on the deep neural network, the gait type of a certain monitoring person known after the monitoring person completes walking in a walking passage with a set passage length in history is used as the output content of the deep neural network, and the visual monitoring contents corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the certain monitoring person, various configuration information of gait health monitoring, and multiple physiological information of the certain monitoring person are used as the input content of the deep neural network to perform this training.

[0078] Example 2

[0079] Figure 3 This is a flowchart of the steps of the gait health monitoring and analysis method according to Embodiment 2 of the present invention.

[0080] like Figure 3As shown, after converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person, that is, after step S15, the gait health monitoring and analysis method further includes:

[0081] Step S16: Transmitting the disease type of the currently monitored person obtained through the conversion to a remote gait management server via a wireless communication link;

[0082] Specifically, the corresponding disease type of the currently monitored person obtained by conversion is transmitted to a remote gait management server through a wireless communication link, including: the gait management server is a cloud computing server, a big data server or a blockchain server.

[0083] Example 3

[0084] Figure 4 This is a flowchart of the steps of the gait health monitoring and analysis method according to Embodiment 3 of the present invention.

[0085] like Figure 4 As shown, after performing multiple training on the deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer, that is, after step S13, the method further includes:

[0086] Step S17: completing the model storage of the posture analysis model by storing various model parameters of the posture analysis model;

[0087] Specifically, completing the model storage of the posture analysis model by storing various model parameters of the posture analysis model includes: a FLASH storage device, an MMC storage device or a static storage device can be selected to complete the model storage of the posture analysis model by storing various model parameters of the posture analysis model.

[0088] In any of the above embodiments 1-3, optionally, in the gait health monitoring and analysis method:

[0089] The visual monitoring contents, various configuration information of gait health monitoring and multiple physiological information of the current monitored person corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the current monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the current monitored person output by the posture analysis model, which also includes: the multiple physiological information of the current monitored person is the age information, gender information, height information and weight information of the current monitored person;

[0090] The configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the distance of the front monitoring camera device, the distance of the side monitoring camera device, and multiple camera parameters of the monitoring camera device. The multiple camera parameters of the monitoring camera device also include: the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast, and exposure time of the monitoring camera device;

[0091] Specifically, the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device, including: the camera frame rate of the monitoring camera device is the number of pictures taken by the monitoring camera device per second.

[0092] Example 4

[0093] Figure 5 1 is an internal structure diagram of a gait health monitoring and analysis system according to Embodiment 4 of the present invention.

[0094] like Figure 5 As shown, the gait health monitoring and analysis system includes the following components:

[0095] An information analysis device, used to obtain various configuration information of gait health monitoring, wherein the gait health monitoring adopts a walking channel with a set passage length and two monitoring cameras with the same camera parameters are respectively arranged on the side and in front of the walking channel, and the various configuration information of the gait health monitoring includes the set passage length, the set walking channel width, the front monitoring camera distance, the side monitoring camera distance and multiple camera parameters of the monitoring camera;

[0096] For example, the information analysis device may include a plurality of different information acquisition units, and the plurality of different information acquisition units are used to respectively obtain a set passage length, a set walking channel width, a front monitoring camera device distance, a side monitoring camera device distance, and a plurality of camera parameters of the monitoring camera device;

[0097] A visual input device is used to obtain visual monitoring content corresponding to a frame of monitoring screen output by each monitoring camera device, and the visual monitoring content corresponding to each frame of monitoring screen is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the area occupied by the human body in the imaging screen of the frame;

[0098] Specifically, the visual monitoring content corresponding to each frame of monitoring picture is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the human body occupied area in the frame of imaging picture, including: each pixel point has an R channel value, i.e., a red channel value, a G channel value, i.e., a green channel value and a B channel value, i.e., a blue channel value in the RGB color space;

[0099] A network training device, used to perform multiple training on a deep neural network to obtain a deep neural network after multiple training and output it as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer;

[0100] For example, a numerical simulation mode may be selected to implement simulation and testing of a data processing process of performing multiple trainings on a deep neural network to obtain a deep neural network after multiple trainings and outputting the deep neural network as a posture analysis model;

[0101] The intelligent analysis device is respectively connected with the information analysis device, the visual input device and the network training device, and is used to synchronously input the visual monitoring contents corresponding to the frames of imaging images taken by the two monitoring cameras during the walking process of the current monitored person, the various configuration information of gait health monitoring and the multiple physiological information of the current monitored person into the posture analysis model when the current monitored person completes walking in the walking passage of the set passage length, and execute the posture analysis model to obtain the gait type of the current monitored person output by the posture analysis model;

[0102] a disease conversion device, connected to the intelligent analysis device, for converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person;

[0103] Among them, the various configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, including: the front monitoring camera device distance is the distance from the monitoring camera device set in front of the walking channel to the end of the walking channel, and the side monitoring camera device distance is the minimum value of the two distances from the monitoring camera device set on the side of the walking channel to the two sides of the walking channel respectively;

[0104] The gait type of the current monitored person output by the posture analysis model is one of unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait, and scissor-step walking gait;

[0105] The gait type of the current monitored person output by the received posture analysis model is converted into the corresponding disease type of the current monitored person, including: unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait and scissor-step walking gait, which correspond to Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian steal syndrome and precursor of cerebral infarction respectively;

[0106] The deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The number of times the deep neural network is trained is monotonically positively correlated with the number of pixels in the imaging picture of the monitoring camera device;

[0107] For example, the number of times the deep neural network is trained is monotonically positively correlated with the number of pixels of the imaging screen of the surveillance camera device, including: when the number of pixels of the imaging screen of the surveillance camera device is 1024×720, the corresponding number of times the deep neural network is trained is 100, when the number of pixels of the imaging screen of the surveillance camera device is 1920×1080, the corresponding number of times the deep neural network is trained is 150, when the number of pixels of the imaging screen of the surveillance camera device is 3840×2160, the corresponding number of times the deep neural network is trained is 200, and so on;

[0108] The deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The number of hidden layers of the deep neural network is proportional to the set travel length;

[0109] For example, the number of hidden layers of the deep neural network is proportional to the set travel length, including: when the travel length is set to 30 meters, the number of hidden layers of the deep neural network is 3, when the travel length is set to 40 meters, the number of hidden layers of the deep neural network is 4, when the travel length is set to 50 meters, the number of hidden layers of the deep neural network is 5, and so on;

[0110] And wherein, the deep neural network is trained multiple times to obtain a deep neural network after multiple trainings and output as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The following include: in each training performed on the deep neural network, the gait type of a certain monitoring person known after the monitoring person completes walking in a walking passage with a set passage length in history is used as the output content of the deep neural network, and the visual monitoring contents corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the certain monitoring person, various configuration information of gait health monitoring, and multiple physiological information of the certain monitoring person are used as the input content of the deep neural network to perform this training.

[0111] Example 5

[0112] Figure 6 1 is an internal structure diagram of a gait health monitoring and analysis system according to Embodiment 5 of the present invention.

[0113] like Figure 6 As shown, the gait health monitoring and analysis system also includes:

[0114] A wireless transmission device, connected to the disease conversion device, is used to transmit the disease type of the corresponding currently monitored person obtained through the conversion to a remote gait management server through a wireless communication link;

[0115] Specifically, the corresponding disease type of the currently monitored person obtained by conversion is transmitted to a remote gait management server through a wireless communication link, including: the gait management server is a cloud computing server, a big data server or a blockchain server.

[0116] Example 6

[0117] Figure 7 1 is an internal structure diagram of a gait health monitoring and analysis system according to Embodiment 6 of the present invention.

[0118] like Figure 7 As shown, the gait health monitoring and analysis system also includes:

[0119] A model storage device, connected to the network training device, and used to complete the model storage of the posture analysis model by storing various model parameters of the posture analysis model;

[0120] Specifically, completing the model storage of the posture analysis model by storing various model parameters of the posture analysis model includes: a FLASH storage device, an MMC storage device or a static storage device can be selected to complete the model storage of the posture analysis model by storing various model parameters of the posture analysis model.

[0121] And in any of the above embodiments 4-6, optionally, in the gait health monitoring and analysis system:

[0122] The visual monitoring contents, various configuration information of gait health monitoring and multiple physiological information of the current monitored person corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the current monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the current monitored person output by the posture analysis model, which also includes: the multiple physiological information of the current monitored person is the age information, gender information, height information and weight information of the current monitored person;

[0123] The configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the distance of the front monitoring camera device, the distance of the side monitoring camera device, and multiple camera parameters of the monitoring camera device. The multiple camera parameters of the monitoring camera device also include: the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast, and exposure time of the monitoring camera device;

[0124] Specifically, the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device, including: the camera frame rate of the monitoring camera device is the number of pictures taken by the monitoring camera device per second.

[0125] In addition, in the gait health monitoring and analysis method and system according to the present invention:

[0126] The visual monitoring contents, various configuration information of gait health monitoring and multiple physiological information of the current monitored person corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the current monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the current monitored person output by the posture analysis model, which also includes: the multiple physiological information of the current monitored person is the age information, gender information, height information and weight information of the current monitored person;

[0127] Specifically, the multiple physiological information of the currently monitored person includes the age information, gender information, height information and weight information of the currently monitored person, including: the age information, gender information, height information and weight information of the currently monitored person are all in binary numerical representation, wherein the text information is converted into a binary numerical value through the ASCII code of each character, and the digital information is directly converted into a binary numerical value;

[0128] The configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the distance of the front monitoring camera device, the distance of the side monitoring camera device, and multiple camera parameters of the monitoring camera device. The multiple camera parameters of the monitoring camera device also include: the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast, and exposure time of the monitoring camera device;

[0129] Accordingly, the plurality of camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device, including: the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device are all in binary numerical representation;

[0130] Specifically, the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device, and also include: the horizontal resolution of the monitoring camera device is the total number of pixel columns in the monitoring screen of the monitoring camera device, and the vertical resolution of the monitoring camera device is the total number of pixel rows in the monitoring screen of the monitoring camera device.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them. Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present disclosure, and they should all be included in the scope of the claims and specification of the present disclosure.

Claims

1. A gait health monitoring and analysis method, characterized in that: The method comprises: Acquire various configuration information of gait health monitoring, wherein the gait health monitoring adopts a walking passage with a set passage length and two monitoring cameras with the same camera parameters are respectively arranged on the side and in front of the walking passage, and the various configuration information of the gait health monitoring includes the set passage length, the set walking passage width, the front monitoring camera distance, the side monitoring camera distance, and multiple camera parameters of the monitoring camera; Obtain visual monitoring content corresponding to a frame of monitoring screen output by each monitoring camera device, wherein the visual monitoring content corresponding to each frame of monitoring screen is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value, and blue channel value of each pixel point in the area occupied by the human body in the imaging screen of the frame; Performing multiple training on the deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer; When the currently monitored person completes walking in a walking passage of a set length, the visual monitoring contents corresponding to the respective frames of imaging images captured by the two monitoring cameras during the currently monitored person's walking, the various configuration information of gait health monitoring, and the multiple physiological information of the currently monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the currently monitored person output by the posture analysis model; Converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person; Among them, the various configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, including: the front monitoring camera device distance is the distance from the monitoring camera device set in front of the walking channel to the end of the walking channel, and the side monitoring camera device distance is the minimum value of the two distances from the monitoring camera device set on the side of the walking channel to the two sides of the walking channel respectively; The gait type of the current monitored person output by the posture analysis model is one of unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait, and scissor-step walking gait; Among them, the gait type of the current monitored person output by the received posture analysis model is converted into the corresponding disease type of the current monitored person, including: gait types of unilateral hand not swinging, swaying left and right, small steps and slow speed, drunken gait, stumbling gait, unilateral crooked gait and scissors step walking gait, which correspond to the disease types of Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian artery steal syndrome and precursor of cerebral infarction one by one.

2. The gait health monitoring and analysis method according to claim 1, characterized in that: Performing multiple training on a deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer, and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer, including: the number of times the deep neural network is trained is monotonically positively correlated with the number of pixels in the imaging picture of the monitoring camera device; The deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The number of hidden layers of the deep neural network is proportional to the set travel length; Among them, the deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The following include: in each training performed on the deep neural network, the gait type of a certain monitoring person known after the monitoring person completes walking in a walking passage with a set passage length in history is used as the output content of the deep neural network, and each visual monitoring content corresponding to each frame of imaging pictures taken by two monitoring cameras during the walking process of the certain monitoring person, various configuration information of gait health monitoring and multiple physiological information of the certain monitoring person are used as the input content of the deep neural network to perform this training.

3. The gait health monitoring and analysis method according to claim 2, characterized in that: After converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person, the method further includes: The corresponding disease type of the currently monitored person obtained through conversion is transmitted to a remote gait management server via a wireless communication link.

4. The gait health monitoring and analysis method according to claim 2, characterized in that: After performing multiple training on the deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer, the method further includes: The model storage of the posture analysis model is completed by storing various model parameters of the posture analysis model.

5. The gait health monitoring and analysis method according to any one of claims 2 to 4, characterized in that: The visual monitoring contents, various configuration information of gait health monitoring and multiple physiological information of the current monitored person corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the current monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the current monitored person output by the posture analysis model, which also includes: the multiple physiological information of the current monitored person is the age information, gender information, height information and weight information of the current monitored person; Among them, the various configuration information of gait health monitoring include setting the passage length, setting the walking channel width, the distance of the front monitoring camera device, the distance of the side monitoring camera device, and multiple camera parameters of the monitoring camera device. The multiple camera parameters of the monitoring camera device also include: the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device.

6. A gait health monitoring and analysis system, characterized in that: The system comprises: An information analysis device, used to obtain various configuration information of gait health monitoring, wherein the gait health monitoring adopts a walking channel with a set passage length and two monitoring cameras with the same camera parameters are respectively arranged on the side and in front of the walking channel, and the various configuration information of the gait health monitoring includes the set passage length, the set walking channel width, the front monitoring camera distance, the side monitoring camera distance and multiple camera parameters of the monitoring camera; A visual input device is used to obtain visual monitoring content corresponding to a frame of monitoring screen output by each monitoring camera device, and the visual monitoring content corresponding to each frame of monitoring screen is the depth of field value, horizontal coordinate value, vertical coordinate value, red channel value, green channel value and blue channel value of each pixel point in the area occupied by the human body in the imaging screen of the frame; A network training device, used to perform multiple training on a deep neural network to obtain a deep neural network after multiple training and output it as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer; An intelligent analysis device is connected to the information analysis device, the visual input device and the network training device respectively, and is used to synchronously input various visual monitoring contents corresponding to various frames of imaging images taken by two monitoring cameras during the walking process of the current monitored person, various configuration information of gait health monitoring, and multiple physiological information of the current monitored person into the posture analysis model when the current monitored person completes walking in the walking passage of the set passage length, and execute the posture analysis model to obtain the gait type of the current monitored person output by the posture analysis model; a disease conversion device, connected to the intelligent analysis device, for converting the gait type of the current monitored person output by the received posture analysis model into the corresponding disease type of the current monitored person; Among them, the various configuration information of the gait health monitoring includes setting the passage length, setting the walking channel width, the front monitoring camera device distance, the side monitoring camera device distance and multiple camera parameters of the monitoring camera device, including: the front monitoring camera device distance is the distance from the monitoring camera device set in front of the walking channel to the end of the walking channel, and the side monitoring camera device distance is the minimum value of the two distances from the monitoring camera device set on the side of the walking channel to the two sides of the walking channel respectively; The gait type of the current monitored person output by the posture analysis model is one of unilateral hand not swinging, swaying left and right, small stride and slow pace, drunken gait, stumbling gait, unilateral crooked gait, and scissor-step walking gait; Among them, the gait type of the current monitored person output by the received posture analysis model is converted into the corresponding disease type of the current monitored person, including: gait types of unilateral hand not swinging, swaying left and right, small steps and slow speed, drunken gait, stumbling gait, unilateral crooked gait and scissors step walking gait, which correspond to the disease types of Parkinson's disease, scoliosis, knee sclerosis, cerebellar atrophy, nerve damage, subclavian artery steal syndrome and precursor of cerebral infarction one by one.

7. The gait health monitoring and analysis system according to claim 6, characterized in that: Performing multiple training on a deep neural network to obtain a deep neural network after multiple training and outputting it as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer, and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer, including: the number of times the deep neural network is trained is monotonically positively correlated with the number of pixels in the imaging picture of the monitoring camera device; The deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, wherein the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The number of hidden layers of the deep neural network is proportional to the set travel length; Among them, the deep neural network is trained multiple times to obtain a deep neural network after multiple training and output as a posture analysis model, the deep neural network includes a single output layer, a single input layer and multiple hidden layers, and the multiple hidden layers are located between the single output layer and the single input layer. The following include: in each training performed on the deep neural network, the gait type of a certain monitoring person known after the monitoring person completes walking in a walking passage with a set passage length in history is used as the output content of the deep neural network, and each visual monitoring content corresponding to each frame of imaging pictures taken by two monitoring cameras during the walking process of the certain monitoring person, various configuration information of gait health monitoring and multiple physiological information of the certain monitoring person are used as the input content of the deep neural network to perform this training.

8. The gait health monitoring and analysis system according to claim 7, characterized in that: The system further comprises: The wireless transmission device is connected to the disease conversion device and is used to transmit the disease type of the corresponding currently monitored person obtained through the conversion to a remote gait management server through a wireless communication link.

9. The gait health monitoring and analysis system according to claim 7, characterized in that: The system further comprises: The model storage device is connected to the network training device and is used to complete the model storage of the posture analysis model by storing various model parameters of the posture analysis model.

10. The gait health monitoring and analysis system according to any one of claims 7 to 9, characterized in that: The visual monitoring contents, various configuration information of gait health monitoring and multiple physiological information of the current monitored person corresponding to each frame of imaging images taken by two monitoring cameras during the walking process of the current monitored person are synchronously input into the posture analysis model, and the posture analysis model is executed to obtain the gait type of the current monitored person output by the posture analysis model, which also includes: the multiple physiological information of the current monitored person is the age information, gender information, height information and weight information of the current monitored person; Among them, the various configuration information of gait health monitoring include setting the passage length, setting the walking channel width, the distance of the front monitoring camera device, the distance of the side monitoring camera device, and multiple camera parameters of the monitoring camera device. The multiple camera parameters of the monitoring camera device also include: the multiple camera parameters of the monitoring camera device are the camera frame rate, horizontal resolution, vertical resolution, contrast and exposure time of the monitoring camera device.

Images