Human body motion data model construction method, device, equipment and medium

Abstract

The present invention relates to a method, device, equipment and medium for constructing a human motion data model, and belongs to the technical field of database construction. The method comprises obtaining a video stream containing human activities, and then preprocessing the video stream to obtain human motion information; constructing a standard human motion model, mapping the human motion information through the standard human motion model, and obtaining initial motion unit data that meets the standard; setting an adaptive filtering unit, the adaptive filtering unit includes a state transfer module and an inter-frame relationship module, and constructing a gain module according to the inter-frame data relationship in the inter-frame relationship module; in the process of dynamic evolution of the standard human motion model, the initial motion unit data is optimized through the state transfer module, the inter-frame relationship module and the gain module, and clear and stable human motion unit data is output; and a human motion database is constructed through the human motion unit data. The present invention can provide low-cost, high-efficiency and high-quality human motion unit data.

Description

Technical Field

[0001] The present invention relates to the technical field of database construction, and in particular to a method, device, equipment and medium for constructing a human motion data model. Background Art

[0002] As a typical application platform of embodied intelligence, intelligent control of humanoid robots has attracted extensive attention. In the study of embodied intelligence, motion data for machine learning imitation and training is a key part of the infrastructure, and human motion database is of great significance to the study of intelligent control of humanoid robots.

[0003] In order to obtain human motion data, wearable devices or motion capture systems are usually used to record the body movements of people when performing tasks. These devices record the movements of various parts of the human body and convert them into standard motion recording units based on the kinematic model of the human body. For example, a motion recording unit may include the position and posture of the body, speed and angular velocity, the angles and angular velocities of each joint of the limbs, the forces at the contact points, etc. The motion recording units at different times constitute the motion data stream, and the data streams under different tasks constitute the human motion database. In the training of the intelligent control algorithm of the humanoid robot, the motion sequences in the human motion database will be used to help the robot imitate and learn the movements of the human body when performing tasks.

[0004] However, due to the high price of wearable devices and motion capture systems, the amount of motion data that can be collected is limited, usually only for specific demonstration tasks, making it difficult to form a comprehensive motion database. In addition, due to the high cost and low efficiency of these devices, experimenters spend a lot of time and effort in collecting data, and it is almost impossible to achieve rich motion data collection under a large number of tasks. These defects increase the difficulty and cost of building a human motion database, hindering the development of research on intelligent control of humanoid robots. Summary of the invention

[0005] Based on this, it is necessary to provide a low-cost, high-efficiency, and high-quality method, device, equipment, and medium for constructing a human motion data model in response to the above-mentioned technical problems.

[0006] A method for constructing a human motion data model, the method comprising:

[0007] Acquire a video stream containing human activities, and then pre-process the video stream to obtain human movement information;

[0008] Constructing a standard human motion model, mapping the human motion information through the standard human motion model, and obtaining initial motion unit data that meets the standard;

[0009] Setting an adaptive filtering unit, the adaptive filtering unit comprising a state transfer module and an inter-frame relationship module, and constructing a gain module according to the inter-frame data relationship in the inter-frame relationship module;

[0010] During the dynamic evolution of the standard human motion model, the initial motion unit data is optimized through the state transfer module, the inter-frame relationship module and the gain module to output clear and stable human motion unit data; a human motion database is constructed through the human motion unit data.

[0011] A human body motion data model construction device, the device comprising:

[0012] An external data acquisition module is used to acquire a video stream containing human activities, and then pre-process the video stream to obtain human movement information;

[0013] An initial motion unit data acquisition module is used to construct a standard human motion model, map the human motion information through the standard human motion model, and obtain initial motion unit data that meets the standard;

[0014] An adaptive filtering construction module is used to set an adaptive filtering unit, wherein the adaptive filtering unit includes a state transfer module and an inter-frame relationship module, and a gain module is constructed according to the inter-frame data relationship in the inter-frame relationship module;

[0015] The database construction module is used to optimize the initial motion unit data through the state transfer module, the inter-frame relationship module and the gain module during the dynamic evolution of the standard human motion model, and output clear and stable human motion unit data; and construct a human motion database through the human motion unit data.

[0016] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method for constructing a human motion data model when executing the computer program.

[0017] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for constructing a human motion data model.

[0018] The above-mentioned human motion data model construction method, device, equipment and medium obtain human motion information by acquiring a video stream containing human activities and then preprocessing the video stream; constructing a standard human motion model, mapping the human motion information through the standard human motion model, and obtaining initial motion unit data that meets the standard; setting an adaptive filtering unit, the adaptive filtering unit includes a state transfer module and an inter-frame relationship module, and constructing a gain module according to the inter-frame data relationship in the inter-frame relationship module; in the process of dynamic evolution of the standard human motion model, the initial motion unit data is optimized through the state transfer module, the inter-frame relationship module and the gain module, and clear and stable human motion unit data is output; and a human motion database is constructed through the human motion unit data.

[0019] The present invention uses the human motion information obtained from the video stream as external data, and the data source is easy to obtain and the cost is low. After the standard human motion model is used to convert it into initial motion unit data that meets the preservation standard, due to the existence of occlusion in the video stream, the initial motion unit data may be distorted, jump, etc. The present invention constructs an adaptive filtering unit, and by setting a state transfer module, the standard human motion model can be dynamically evolved to form a self-loop; the image frame in the evolution process is evaluated by the inter-frame relationship module, which is helpful to check the consistency of the predicted motion unit data; the response to the error is enhanced by the gain module, the error accumulation is reduced, and the evolution process is smoother and more natural; through the comprehensive processing of the state transfer module, the inter-frame relationship module and the gain module, it is possible to generate unobstructed, complete data, high-quality, clear and stable human motion unit data, and a standardized human motion database is constructed through these human motion unit data to provide high-quality human motion data support for robot intelligent control, human-computer interaction and other fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0021] Figure 1 A schematic diagram of a process of constructing a human motion data model in one embodiment;

[0022] Figure 2 A schematic diagram of the structure of a standard human motion model in one embodiment;

[0023] Figure 3 A schematic diagram of a self-circulation demonstration in an embodiment;

[0024] Figure 4 A structural block diagram of a device for constructing a human motion data model in one embodiment;

[0025] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment.

[0026] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] It should be noted that the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0029] The following will describe the implementation of the present invention in detail with reference to the accompanying drawings in the embodiment diagram of the present invention.

[0030] Example 1

[0031] This embodiment discloses a method for constructing a human motion data model. By using the human motion information obtained from the video stream as external data, the data source is easy to obtain and the cost is low. After the standard human motion model is used to convert it into initial motion unit data that meets the preservation standard, due to the occlusion in the video stream, the initial motion unit data may be distorted, jump, etc. The present invention constructs an adaptive filtering unit, and by setting a state transfer module, the standard human motion model can be dynamically evolved to form a self-loop; the image frame in the evolution process is evaluated by the inter-frame relationship module, which helps to check the consistency of the predicted motion unit data; the response to the error is enhanced by the gain module, the error accumulation is reduced, and the evolution process is smoother and more natural; through the comprehensive processing of the state transfer module, the inter-frame relationship module and the gain module, it is possible to generate human motion unit data without occlusion, complete data, high quality, clear and stable, and a standardized human motion database is constructed through these human motion unit data to provide high-quality human motion data support for robot intelligent control, human-computer interaction and other fields.

[0032] like Figure 1As shown, the method for constructing a human motion data model in this embodiment includes the following steps:

[0033] Step 201, obtaining a video stream containing human activities, and then preprocessing the video stream to obtain human body movement information.

[0034] It can be understood that in addition to human motion information, the video stream data also contains other interference data. Through preprocessing, the noise, artifacts and other interference factors in the video stream data can be removed, and only key features such as human contours, joints, postures, etc. are detected and extracted to obtain clearer and more accurate human motion information for subsequent use.

[0035] Step 202: construct a standard human motion model, map the human motion information through the standard human motion model, and obtain initial motion unit data that meets the standard.

[0036] It can be understood that human motion information comes from video stream data, and the methods of obtaining video stream data are relatively diverse. In addition, there are various differences in the target human bodies to be identified (height, weight, limb length, etc.). By mapping the human motion information through a standard human motion model, motion information from different sources and in different formats can be unified into a standardized representation, which is convenient for the subsequent construction of a standardized human motion database and can make the human motion database universal.

[0037] Step 203: Setting an adaptive filtering unit, the adaptive filtering unit includes a state transfer module and an inter-frame relationship module, and constructing a gain module according to the inter-frame data relationship in the inter-frame relationship module.

[0038] Step 204, during the dynamic evolution of the standard human motion model, the initial motion unit data is optimized through the state transfer module, the inter-frame relationship module and the gain module, and clear and stable human motion unit data is output; a human motion database is constructed through the human motion unit data.

[0039] It can be understood that there may be occlusion in the initial motion unit data. Therefore, the initial motion unit data is optimized by designing an adaptive filtering unit, and the motion state of the occluded part is predicted according to the motion state of the adjacent frame images, so that the generated human motion unit data is complete, clear and stable, and is not affected by occlusion in the video stream. Among them, the state transfer module is mainly used for the dynamic evolution of the standard human motion model to enhance the adaptability of the standard human motion model, so that it can simulate more complex motions, so as to better predict the future motion state; the inter-frame relationship module is used to evaluate the motion of two adjacent frames of images to assist in checking the consistency between the predicted motion state and the actual motion state; the gain module is mainly used to enhance the response to errors, reduce error accumulation, and make the evolution process smoother and more natural. By optimizing the initial motion unit data through the state transfer module, the inter-frame relationship module and the gain module, the distortion and jump problems of the data can be suppressed, so that the obtained human motion unit data is effective, complete, stable and clear.

[0040] In the specific implementation process of step 201, firstly, video streams containing human activities under various task conditions are collected and screened, and the data sources include but are not limited to databases, the Internet, open source platforms, etc. In addition, when acquiring video streams, in addition to considering the diversity of sources, attention should also be paid to the quality, resolution and shooting angle of the video.

[0041] Then, the image frames containing human activities in the video stream are processed in sequence such as denoising, motion segmentation, and motion recognition to obtain human motion information. Among them, the image frames containing human activities in the video stream are denoised and enhanced to obtain enhanced images to improve contrast and enhance the recognizability of the video stream. In addition to traditional methods, denoising can also use algorithms such as convolutional neural networks (CNN) in deep learning to perform more refined image denoising and enhancement. Then, different human motion parts in the enhanced image are effectively segmented to obtain segmented images; when performing motion segmentation, a human posture estimation model (such as OpenPose) can be used to accurately identify and segment different parts of the human body to improve the accuracy of segmentation. Finally, the human motion pattern in the segmented image is identified to obtain human motion information; when performing motion recognition, the motion data of different parts of the human body or the mutual motion data between different parts of the human body can be obtained; the human motion pattern can be efficiently identified by combining machine learning algorithms (such as random forests, support vector machines) and deep learning technologies (such as long short-term memory networks LSTM).

[0042] The human motion information obtained includes the position, posture, speed, angular velocity of the body, and the angles and angular velocities of each joint of the limbs, etc. However, due to the problem of occlusion in the video stream, the obtained human motion information may be distorted, jumpy, etc. If it is not processed, the validity of the data will be destroyed, and the quality of the generated human motion unit data will be low. Therefore, when mapping and converting the human motion information, it is also necessary to remove the occlusion.

[0043] During the specific implementation of step 202, a standard human motion model is constructed, which includes motion unit data; human motion information is input into the standard human motion model, and the difference between the human motion information and the motion unit data and the first-order derivative of the human motion information are amplified and then summed to map them into initial motion unit data that meets the standard.

[0044] As a preferred embodiment, the standard human motion model is as follows: Figure 2 As shown in Figure 1, it contains complete motion unit data and is not affected by occlusion in the video stream. The motion unit data mainly includes a floating body with 6 degrees of freedom. and the angles and angular velocities of each joint ,Therefore, the human motion information is mapped through the standard human motion ,model, that is, the human motion information is converted into a standardized format ,according to the motion unit data, and the initial motion unit data is generated.

[0045] Specifically, the human motion information is mapped through the standard human motion model, mainly by adding the difference between the human motion information and the motion unit data and the first-order derivative of the human motion information after gaining, and the expression is:

[0046] ;

[0047] In the formula, Indicates the mapping adjustment amount; represents position gain; Represents human body movement information; represents the first time derivative of the motion unit data; Indicates speed gain; The first-order time derivative representing the human motion information; Represents the solution of the kinematic model based on the motion unit data.

[0048] After the preliminary mapping is completed, the generated initial motion unit data expression is:

[0049] ;

[0050] In the formula, represents three rotational degrees of freedom; represents three translational degrees of freedom; Represents the angle and angular velocity of each joint.

[0051] In the specific implementation process of step 203, an adaptive filtering unit is designed to dynamically evolve the standard human motion model to form an internal self-loop. In this way, the input external data - human motion information is accurately identified and optimized to generate high-quality human motion unit data that is not affected by occlusion in the video stream.

[0052] The adaptive filtering unit includes a state transfer module and an inter-frame relationship module, wherein the transfer module transforms the state at the previous moment into the state at the next moment according to the state transfer matrix, and the expression is:

[0053] ;

[0054] The inter-frame relationship module is constructed according to the optimization principle based on the state transfer matrix and the inter-frame relationship of the previous moment, combined with the evaluation matrix of the current image and the next frame image, and the expression is:

[0055] ;

[0056] In the formula, It represents the state estimation information transfer matrix from the current frame image to the next frame image. This is because the time interval between the human body movement in two adjacent frames of the video stream is very small, and the target change can be considered to be uniform motion, so it is represented as a linear matrix; Indicates the state estimation information corresponding to the next frame image; Indicates the state estimation information corresponding to the current frame image; Indicates the current frame image; Indicates the next frame image; represents the optimization matrix for updating the next estimation based on the state transition law and the relationship between frames; It represents the inter-frame relationship matrix corresponding to the filtering calculation of the current frame image. The number of rows and columns of the matrix represent the amount of information of the current frame and the next frame respectively. represents the evaluation matrix between the current frame image and the next frame image, and the evaluation matrix is ​​obtained by matching the initial motion unit data with the dynamic evolution data of the standard human motion model; Represents the transpose of the state estimation information transfer matrix from the current frame image to the next frame image.

[0057] Inter-frame data relationship matrix When updating, it includes two processes: one is the time update in the previous prediction cycle; the other is the observation evaluation update related to external data. Therefore, the gain module is also constructed according to the inter-frame data relationship in the inter-frame relationship module. The gain module is constructed according to the updated inter-frame relationship matrix and the default gain compensation, and the expression is:

[0058] ;

[0059] In the formula, Represents the gain matrix for gaining the next frame of image; represents the optimization matrix for updating the next estimation based on the state transition law and the relationship between frames; Represents the basic gain matrix when estimating the next frame image.

[0060] The gain matrix is ​​used to automatically optimize and adjust the relevant filter parameters to make the evolution process smoother and more natural, effectively suppressing the distortion and jump problems of the data.

[0061] In the specific implementation process of step 203, an adaptive filtering unit is introduced into the standard human motion model, thereby realizing the self-loop optimization of the standard human motion model. Figure 3 As shown, human motion information is input into the standard human motion model as external data. The state transfer module dynamically evolves according to the initial motion unit data output by the standard human motion model to generate predicted motion unit data corresponding to the next frame image. The inter-frame relationship module evaluates the motion of the current frame and the next frame image, and automatically optimizes and adjusts the relevant parameters of the inter-frame relationship module through the gain module to make the predicted motion state of the next frame image consistent with the actual motion state, and finally outputs the human motion unit data through the standard human motion model.

[0062] The output data of each human motion unit mainly include: the position and posture of the body, the speed and angular velocity of the body, the angles of each joint of the limbs, the angular velocity of each joint of the limbs and the force of the contact point. All the human motion unit data obtained after optimization are saved separately according to the task type, and then together form a standardized human motion database, providing high-quality human motion data support for the fields of robot intelligent control, human-computer interaction, etc., and supporting the research and development in the fields of humanoid robot intelligent control, human-computer interaction, etc. The database is opened to researchers in an open source or authorized manner as an important infrastructure for robot intelligent control research.

[0063] In the method proposed in the present invention, the motion unit data in the constructed standard human motion model is complete and is not affected by data missing when there is occlusion in the video stream; at the same time, under the joint action of the state transfer module, the inter-frame relationship module and the gain module, the human motion unit data output by the standard human motion model can optimally match the human motion in the video stream.

[0064] It should be understood that although this embodiment Figure 1 The steps in the process are shown in sequence as indicated by the arrows, but the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of the steps, and the steps can be executed in other orders. Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0065] Example 2

[0066] Based on the method for constructing a human motion data model in Example 1, this embodiment discloses a device for constructing a human motion data model, such as Figure 4 As shown, the device includes an external data acquisition module 401, an initial motion unit data acquisition module 402, an adaptive filter construction module 403 and a database construction module 404. Among them:

[0067] The external data acquisition module 401 is used to acquire a video stream containing human activities, and then pre-process the video stream to obtain human movement information.

[0068] The initial motion unit data acquisition module 402 is used to construct a standard human motion model, and to map human motion information through the standard human motion model to obtain initial motion unit data that meets the standard.

[0069] The adaptive filtering construction module 403 is used to set an adaptive filtering unit. The adaptive filtering unit includes a state transfer module and an inter-frame relationship module. The gain module is constructed according to the inter-frame data relationship in the inter-frame relationship module.

[0070] The database construction module 404 is used to optimize the initial motion unit data during the dynamic evolution of the standard human motion model through the state transfer module, the inter-frame relationship module and the gain module, and output clear and stable human motion unit data; and construct a human motion database through the human motion unit data.

[0071] The specific working process and working principle of the external data acquisition module 401, the initial motion unit data acquisition module 402, the adaptive filter construction module 403 and the database construction module 404 in this embodiment are the same as those in the method of embodiment 1, so they are not described in detail in this embodiment. Each unit module can be implemented in whole or in part by software, hardware and a combination thereof, and each unit module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above unit modules.

[0072] Example 3

[0073] This embodiment discloses a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data generated in Example 1. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the method in the above-mentioned Example 1 is implemented.

[0074] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0075] Example 4

[0076] This embodiment discloses a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the method in the above-mentioned embodiment 1 is implemented.

[0077] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0078] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0079] The above-mentioned embodiments only express several implementation modes of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A method for constructing a human motion data model, characterized in that: The method comprises: Acquire a video stream containing human activities, and then pre-process the video stream to obtain human movement information; Constructing a standard human motion model, mapping the human motion information through the standard human motion model, and obtaining initial motion unit data that meets the standard; Setting an adaptive filtering unit, the adaptive filtering unit comprising a state transfer module and an inter-frame relationship module, and constructing a gain module according to the inter-frame data relationship in the inter-frame relationship module; During the dynamic evolution of the standard human motion model, the initial motion unit data is optimized through the state transfer module, the inter-frame relationship module and the gain module to output clear and stable human motion unit data; a human motion database is constructed through the human motion unit data; The state transfer module is used to transfer the state between images. The expression of the state transfer is: ； The inter-frame data relationship of the image is represented by the inter-frame relationship module, and the expression is: ； The image is gained by the gain module, and the expression is: ； In the formula, Represents the state estimation information transfer matrix from the current frame image to the next frame image; Indicates the state estimation information corresponding to the next frame image; Indicates the state estimation information corresponding to the current frame image; Indicates the current frame image; Indicates the next frame image; represents the optimization matrix for updating the next estimation based on the state transition law and the relationship between frames; Represents the inter-frame relationship matrix corresponding to the current frame image filtering calculation; Represents the transpose of the state estimation information transfer matrix from the current frame image to the next frame image; Represents the evaluation matrix between the current frame image and the next frame image; Represents the gain matrix for gaining the next frame of image; Represents the basic gain matrix when estimating the next frame image.

2. The method for constructing a human motion data model according to claim 1, characterized in that: Preprocessing the video stream to obtain human body motion information includes: Acquire an image frame containing human activities in a video stream, and perform denoising and enhancement processing on the image frame to obtain an enhanced image; Segmenting different human motion parts in the enhanced image to obtain a segmented image; The human body motion pattern in the segmented image is identified to obtain human body motion information.

3. The method for constructing a human motion data model according to claim 1, characterized in that: Constructing a standard human motion model, mapping the human motion information through the standard human motion model, and obtaining initial motion unit data that meets the standard, including: Constructing a standard human motion model, wherein the standard human motion model includes motion unit data; The human motion information is input into the standard human motion model, and the difference between the human motion information and the motion unit data and the first-order derivative of the human motion information are respectively amplified and summed up to map into initial motion unit data that meets the standard.

4. The method for constructing a human motion data model according to claim 3, characterized in that: The human motion information is mapped through a standard human motion model, and the expression is: ； In the formula, Indicates the mapping adjustment amount; represents position gain; Represents human body movement information; represents the first time derivative of the motion unit data; Indicates speed gain; The first-order time derivative representing the human motion information; Represents the solution of the kinematic model based on the motion unit data.

5. A human body motion data model construction device, characterized in that: The method for constructing a human motion data model according to any one of claims 1 to 4 is adopted, wherein the device comprises: An external data acquisition module is used to acquire a video stream containing human activities, and then pre-process the video stream to obtain human movement information; An initial motion unit data acquisition module is used to construct a standard human motion model, map the human motion information through the standard human motion model, and obtain initial motion unit data that meets the standard; An adaptive filtering construction module is used to set an adaptive filtering unit, wherein the adaptive filtering unit includes a state transfer module and an inter-frame relationship module, and a gain module is constructed according to the inter-frame data relationship in the inter-frame relationship module; The database construction module is used to optimize the initial motion unit data during the dynamic evolution of the standard human motion model through the state transfer module, the inter-frame relationship module and the gain module, and output clear and stable human motion unit data; and construct a human motion database through the human motion unit data.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method for constructing a human motion data model described in any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for constructing a human motion data model described in any one of claims 1 to 4 are implemented.

Images