Hand action reference generation method, system and equipment based on surface myoelectricity and medium

Abstract

The invention relates to the technical field of man-machine interaction, and discloses a surface electromyography-based hand action benchmark generation method, system and device and a medium, the method comprises the following steps: collecting a multi-channel surface electromyography signal at a wrist, and combining a setting relation between a surface electromyography electrode and wrist tendon distribution to generate a multi-channel surface electromyography signal; multi-channel surface electromyographic signals are separated on the basis of signal transmission characteristic differences from different hand tendons to surface electromyographic electrodes, so that electromyographic signal components driven by different hand tendons are obtained, reference data of hand actions are generated accordingly, highly-coupled surface electromyographic signals can be effectively distinguished in a limited space, and the accuracy of the surface electromyographic signals is improved. Therefore, stable and repeatable hand action reference data can be obtained.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, and in particular to a method, system, device and medium for generating hand movement references based on surface electromyography. Background Technology

[0002] With the gradual development of human-computer interaction technology, non-contact gesture interaction has received widespread attention from the market due to its high degree of freedom and rich interactive experience.

[0003] In hand motion recognition and related applications, obtaining stable and reliable hand motion data is crucial for improving system reliability. Existing technologies largely rely on vision or inertial sensing to perceive external hand movements. However, these methods rely on indirect inference based on apparent motion, making it difficult to consistently and accurately reflect the true state of hand movements in complex environments, and particularly challenging to generate consistent and repeatable motion benchmark data. Therefore, improvements to existing technologies are necessary. Summary of the Invention

[0004] This invention provides a method, system, device, and medium for generating hand movement references based on surface electromyography, in order to solve the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for generating hand movement references based on surface electromyography includes:

[0007] Multi-channel surface electromyography (EMG) signals at predetermined locations on the hand are acquired using multiple surface EMG electrodes.

[0008] Based on the arrangement of the surface electromyography electrodes relative to the distribution of tendons at predetermined positions on the hand, the multi-channel surface electromyography signals are processed.

[0009] Based on the differences in signal transmission characteristics from different hand tendons to surface electromyography electrodes, the processed multi-channel surface electromyography signals are separated to obtain electromyography signal components driven by different hand tendons.

[0010] Based on the electromyographic signal components, baseline data for hand movements are generated.

[0011] Optionally, the multi-channel surface electromyography signals are acquired at the wrist.

[0012] Optionally, the surface electromyography electrodes are arranged in a ring array along the circumference of the wrist.

[0013] Optionally, the surface electromyography electrode is divided into at least four acquisition areas according to the anatomical division of the wrist tendons: the back of the hand, the palm, the radial side, and the ulnar side. Each acquisition area corresponds to the extensor tendons and flexor tendons of different anatomical divisions.

[0014] Optionally, the differences in signal transmission characteristics include: frequency domain characteristics and / or time-frequency distribution characteristics.

[0015] Optionally, before performing signal separation on the processed multichannel surface electromyography signals, the method further includes:

[0016] For each surface electromyography electrode, calculate at least one of the following parameters: signal-to-noise ratio parameter, signal amplitude stability parameter, and saturation state parameter;

[0017] Based on at least one of the signal-to-noise ratio parameter, signal amplitude stability parameter, and saturation state parameter, channel filtering and / or channel weighting processing are performed on multi-channel surface electromyography signals.

[0018] Optional, also includes:

[0019] The baseline data of the hand movements are synchronized and aligned with the results of hand movement recognition or reconstruction on the time axis.

[0020] Perform difference calculations or deviation statistical processing on the motion parameters or spatial pose parameters at the corresponding time points.

[0021] The present invention also provides a hand movement reference generation system based on surface electromyography, comprising:

[0022] The surface electromyography (EMG) acquisition module includes multiple surface EMG electrodes, which are used to acquire multi-channel surface EMG signals at predetermined locations on the hand.

[0023] The signal processing module is used to process the multi-channel surface electromyography signals based on the arrangement relationship of the surface electromyography electrodes relative to the tendon distribution at a predetermined position on the hand.

[0024] The signal separation module is used to separate the processed multi-channel surface electromyography (EMG) signals based on the differences in signal transmission characteristics from different hand tendons to surface EMG electrodes, so as to obtain the EMG signal components driven by different hand tendons.

[0025] The reference data generation module is used to generate reference data for hand movements based on the electromyographic signal components.

[0026] The present invention also provides a computer device, including a processor and a memory, wherein the memory stores a computer program, characterized in that, when the processor executes the computer program, it is used to implement the hand movement reference generation method based on surface electromyography as described in any of the preceding claims.

[0027] The present invention also provides a computer-readable storage medium storing computer-executable instructions thereon, characterized in that, when the computer-executable instructions are executed by a processor, they are used to implement the hand movement reference generation method based on surface electromyography as described in any of the preceding claims.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] This invention provides a method, system, device, and medium for generating hand movement references based on surface electromyography (SEMG). By combining the arrangement relationship between SEMG electrodes and tendon distribution, and based on the differences in signal transmission characteristics from different hand tendons to SEMG electrodes, multi-channel SEMG signals are separated to obtain SEMG signal components driven by different hand tendons. This allows for the effective differentiation of highly coupled SEMG signals within a limited space, thereby obtaining stable and repeatable hand movement reference data.

[0030] The present invention has other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of the invention. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart of a method for generating hand movement references based on surface electromyography provided in an embodiment of the present invention;

[0033] Figure 2 A diagram showing the distribution of tendons from the palm to the wrist;

[0034] Figure 3 This is a schematic diagram of a wristband used in a method for generating hand movement references based on surface electromyography provided in an embodiment of the present invention;

[0035] Figure 4This is an electrode diagram on a wristband in a method for generating hand movement references based on surface electromyography provided in an embodiment of the present invention;

[0036] Figure 5 This is a specific electrode distribution diagram on the wristband in a hand movement reference generation method based on surface electromyography provided in an embodiment of the present invention;

[0037] Figure 6 This is a flowchart illustrating the alignment and comparative analysis of motion parameters or spatial pose parameters of a motion reference model and a three-dimensional hand motion model at corresponding time points in a hand motion reference generation method based on surface electromyography provided in an embodiment of the present invention.

[0038] Figure 7 This is a structural block diagram of a hand movement reference generation system based on surface electromyography provided in an embodiment of the present invention;

[0039] Figure 8 This invention provides a hand movement reference generation system based on surface electromyography that is associated with VR display to achieve high-precision projection of hand movements.

[0040] Figure reference numerals: 10, surface electromyography acquisition module; 20, signal processing module; 30, signal separation module; 40, reference data generation module. Detailed Implementation

[0041] To illustrate the possible application scenarios, technical principles, implementable specific solutions, and achievable objectives and effects of this application in detail, the following description, in conjunction with the listed specific embodiments and accompanying drawings, provides a detailed explanation. The embodiments described herein are merely illustrative of the technical solutions of this application and are therefore intended to limit the scope of protection of this application.

[0042] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0043] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit this application.

[0044] In the description of this application, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" logical relationship.

[0045] In this application, terms such as “first” and “second” are used only to distinguish one entity or operation from another, and do not necessarily require or imply any actual quantity, hierarchy or order relationship between these entities or operations.

[0046] Unless otherwise specified, the use of terms such as “comprising,” “including,” “having,” or other similar expressions in this application is intended to cover non-exclusive inclusion, which does not exclude the presence of additional elements in a process, method, or product that includes the stated elements, such that a process, method, or product that includes a list of elements may include not only those defined elements but also other elements not expressly listed, or elements inherent to such a process, method, or product.

[0047] Similar to the understanding in the Examination Guidelines, in this application, expressions such as "greater than," "less than," and "exceeding" are understood to exclude the stated number; expressions such as "above," "below," and "within" are understood to include the stated number. Furthermore, in the description of the embodiments in this application, "multiple" means two or more (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups" and "multiple times," unless otherwise explicitly specified.

[0048] In the description of the embodiments of this application, the space-related expressions used, such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "vertical," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," indicate the orientation or positional relationship based on the orientation or positional relationship shown in the specific embodiments or drawings. They are only for the purpose of describing the specific embodiments of this application or for the reader's understanding, and do not indicate or imply that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0049] Unless otherwise expressly specified or limited, the terms "installation," "connection," "linking," "fixing," and "setting," as used in the description of the embodiments of this application, should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral setting; it can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be the internal connection of two components or the interaction between two components. For those skilled in the art to which this application pertains, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0050] Please refer to Figure 1 This invention provides a method for generating hand movement references based on surface electromyography, comprising:

[0051] S1. Multi-channel surface electromyography (EMG) signals are acquired at predetermined locations on the hand using multiple surface EMG electrodes.

[0052] In this embodiment, firstly, multi-channel surface electromyography signals generated at predetermined locations on the hand are acquired using multiple surface electromyography electrodes.

[0053] Among them, surface electromyography (SEMG) signals originate from the bioelectrical signals generated on the skin surface during the contraction and relaxation of hand tendons. These signals can reflect the intrinsic driving state of hand movements from a physiological perspective. By placing multiple SEMG electrodes at predetermined locations on the hand, electromyographic activity information from different tendon regions can be acquired simultaneously in space, laying the foundation for subsequent differentiation of movements driven by different tendons.

[0054] S2. Based on the arrangement of surface electromyography electrodes relative to the distribution of tendons at predetermined positions on the hand, multi-channel surface electromyography signals are processed.

[0055] After acquiring multi-channel surface electromyography (EMG) signals, the acquired multi-channel EMG signals are further processed by combining the setting relationship of the surface EMG electrodes relative to the tendon distribution at a predetermined position on the hand.

[0056] It is understandable that different hand tendons have distinct anatomical distribution characteristics, and the relative positional relationship between the electrodes and tendons directly affects the electromyographic signals sensed by the electrodes. In this step, by introducing the aforementioned relationship information during signal processing, it is possible to avoid treating multi-channel electromyographic signals as independent or randomly distributed data, making the final recognized gestures closer to actual hand movements.

[0057] S3. Based on the differences in signal transmission characteristics from different hand tendons to surface electromyography electrodes, the processed multi-channel surface electromyography signals are separated to obtain electromyography signal components driven by different hand tendons.

[0058] Furthermore, this method separates the processed multi-channel surface electromyography signals based on the differences in signal transmission characteristics from different hand tendons to surface electromyography electrodes.

[0059] It should be noted that in the designated area of ​​the hand, especially at the wrist, multiple hand tendons are highly concentrated anatomically. Different tendons are spatially adjacent or even partially overlapping. Furthermore, during actual movement, multiple tendons often participate in the same hand movement in a coordinated manner, resulting in high coupling of the corresponding surface electromyographic signals during acquisition. Particularly, tendons located at different anatomical depths exhibit different attenuation and response characteristics when their electromyographic signals propagate to the surface electrodes. Deep tendon signals are relatively weak and easily masked by superficial tendon signals, significantly increasing the difficulty of distinguishing signals from adjacent tendons.

[0060] Based on the above characteristics, in this step, the multi-channel surface electromyography signals are separated by considering the differences in transmission characteristics of different hand tendons during signal propagation, so as to avoid the signals fed back by deep tendons being covered by superficial tendon signals.

[0061] Specifically, the electromyographic (EMG) signals corresponding to different tendons differ in spectral energy distribution, dominant frequency band, and time-varying characteristics. Furthermore, their activation sequence may vary during movement. These differences in activation sequence stem from variations in tendon anatomical depth, neural control pathways, and mechanical transmission methods. By comprehensively analyzing frequency domain characteristics, time-frequency distribution characteristics, and activation sequence characteristics, it is possible to distinguish the EMG signal components driven by different hand tendons within highly coupled multi-channel surface EMG signals, thereby achieving effective decoupling of signals from anatomically adjacent tendons.

[0062] To address the challenge of signal separation between the flexor digitorum superficialis (FDS) and flexor digitorum profundus (FDP) tendons, which are anatomically adjacent, this invention establishes an electromyographic signal transmission model. Specifically, the mixed electromyographic signal S_i(t) received by electrode i can be expressed as S_i(t) = α_i・S_FDS(t)*h_FDS(τ) + β_i・S_FDP(t)*h_FDP(τ) + n_i(t).

[0063] Where S_FDS(t) and S_FDP(t) are the actual electromyographic signals of FDS and FDP, respectively, h_FDS(τ) and h_FDP(τ) represent the impulse response functions from the tendon to the electrode, α_i and β_i are the attenuation coefficients related to the anatomical depth of the tendon, n_i(t) is the noise term, and * indicates convolution operation.

[0064] The key to this invention lies in the discovery of a significant difference between the impulse response functions of FDS and FDP. The impulse response function of superficial tendon FDS is h_FDS(τ)=exp(-τ / τ_s) (τ_s is the superficial attenuation coefficient, which is small, and the signal has no oscillation and decays rapidly), while the impulse response function of deep tendon FDP is h_FDP(τ)=exp(-τ / τ_d)*cos(ω_d・τ) (τ_d is the deep attenuation coefficient, which is large, and the signal has oscillating characteristics and decays slowly). This difference provides the core basis for the accurate separation of the two signals.

[0065] S4. Generate baseline data for hand movements based on electromyographic signal components.

[0066] In this step, based on the electromyographic signal components driven by different hand tendons obtained in step S3, reference data for hand movements are further generated based on these electromyographic signal components.

[0067] Since the reference data obtained in this embodiment comes directly from the physiological signals of hand tendon activity and does not rely on indirect inference from external action appearances, it can provide a consistent and repeatable reference representation for hand movements.

[0068] In summary, the overall process of generating hand movement reference data based on surface electromyography in this embodiment is as follows: starting from signal acquisition, it includes signal preprocessing, feature extraction and movement recognition, and hand movement reconstruction in sequence.

[0069] Specifically, the signal acquisition module acquires multi-channel surface electromyography (EMG) signals by placing multiple surface EMG electrodes at predetermined positions on the hand; the signal preprocessing module performs basic processing such as filtering and noise reduction on the acquired EMG signals to obtain signals suitable for subsequent analysis; based on this, the feature extraction and action recognition module analyzes the processed EMG signals and further generates representations of hand movements; finally, the hand movement reconstruction module forms the corresponding hand movement model.

[0070] In a further embodiment, the acquisition location for the multichannel surface electromyography signal is limited to the wrist.

[0071] Understandably, the wrist is a crucial area where multiple tendons related to fine hand movements converge and extend towards the fingers. Electromyography (EMG) data acquisition in this area allows for the simultaneous acquisition of EMG activity information from multiple tendons without interfering with free finger movement. By focusing data acquisition on the wrist, the ability of EMG signals to represent overall hand movements can be improved while maintaining ease of wear.

[0072] Furthermore, the surface electromyography electrodes are arranged in a ring array along the circumference of the wrist. This ring array configuration creates continuous acquisition coverage around the wrist, allowing for the effective sensing of tendon signals from different directions and anatomical locations. Compared to unilateral or linear arrangements, the ring array offers higher uniformity in spatial sampling, which enhances the complementarity between multi-channel signals, thereby improving the reliability of subsequent signal separation.

[0073] Furthermore, the surface electromyography electrodes are divided into multiple acquisition areas according to the anatomical divisions of the wrist tendons, with each acquisition area corresponding to the extensor and flexor tendons of different functional zones. Through this regional arrangement based on anatomical divisions, the major tendons in both the dorsal and palmar sides of the wrist can be systematically covered, thus forming a comprehensive acquisition of the major motor tendons of the hand in the wrist area.

[0074] This embodiment optimizes the hardware acquisition process, reducing the aliasing of signals from different functional tendons and providing favorable conditions for subsequent differentiation of fine hand movements.

[0075] Based on the above implementation method, in practical applications, this embodiment further limits the material, number, partition layout and acquisition parameters of the surface electromyography electrodes in order to achieve comprehensive and high-fidelity acquisition of electromyography signals of 17 major tendons in the hand.

[0076] The surface electromyography (EMG) electrodes are made of silver / silver chloride, a material with good biocompatibility and signal transmission stability. This reduces impedance interference at the skin-electrode interface, ensuring accurate transmission of EMG signals. In this embodiment, please refer to... Figure 2 This is a diagram showing the distribution of tendons from the palm to the wrist. The arrangement of the electrodes is based on the following anatomical concept of human tendon distribution:

[0077] The "nine tendons" on the back of the wrist are the core tendons that drive hand extension movements. They correspond to thumb abduction, extension, distal extension, radial / ulnar extension of the wrist, extension of the 2nd to 5th fingers, and independent extension of the index and little fingers. The "six tubes" are the tendon sheaths of the back tendons, which wrap around the corresponding tendons according to the principle of functional association, and play a role in fixation and protection.

[0078] Therefore, in this embodiment, please refer to Figures 3 to 5 The diagrams show a wristband for wearing on the wrist, an electrode diagram on the wristband, and a detailed electrode distribution diagram. The preferred number of electrodes is 24, arranged in a tight circular array along the circumference of the wrist. The wearing position is strictly limited to 2-3 cm proximal to the wrist crease, a position that precisely corresponds to the distribution area of ​​the wrist tendons, enabling signal acquisition without affecting hand movement.

[0079] The 24-electrode ring array layout, based on the functional distribution of "nine tendons" and the anatomical grouping of "six tubes", fully covers 17 major tendons in four regions: dorsal side, palmar side, radial side, and ulnar side, to achieve precise acquisition of electromyographic signals.

[0080] Furthermore, depending on the actual acquisition accuracy requirements, the number of electrodes can be expanded to 32 to achieve denser tendon coverage.

[0081] To achieve comprehensive coverage of the 17 major tendons, the 24 surface electromyography electrodes were divided into four acquisition areas according to the anatomical divisions of the wrist tendons. The number of electrodes and coverage area in each area are as follows:

[0082] Back of hand (extensor area): 8 electrodes (numbered D1-D8) are arranged, mainly covering the "nine tendons" on the back of the wrist, including the abductor pollicis longus tendon, extensor pollicis brevis tendon, extensor carpi radialis longus tendon, extensor carpi radialis brevis tendon, extensor pollicis longus tendon, extensor digitorum tendon, extensor index finger tendon, extensor digiti minimi tendon, and extensor carpi ulnaris tendon, corresponding to extensor movements such as thumb abduction, finger extension, and radial / ulnar deviation of the wrist;

[0083] Palmar side (flexor area): 8 electrodes (numbered P1-P8) are arranged, mainly covering the palmar flexor tendons, including the flexor pollicis longus tendon, flexor digitorum superficialis tendon (FDS), flexor digitorum profundus tendon (FDP), palmaris longus tendon, flexor carpi radialis tendon, and flexor carpi ulnaris tendon, corresponding to flexor movements such as thumb flexion, finger flexion, and wrist flexion.

[0084] Radial side (thumb side): Four electrodes (numbered R1-R4) are placed to assist in covering the brachioradialis tendon, flexor carpi radialis tendon, abductor pollicis longus tendon, and extensor pollicis brevis tendon, corresponding to auxiliary movements such as forearm rotation, radial deviance of the wrist, and thumb abduction and extension.

[0085] Ulnar side (little finger side): Four electrodes (numbered U1-U4) are placed to assist in covering the ulnar extensor carpi ulnaris tendon, ulnar flexor carpi ulnaris tendon, extensor digiti minimi tendon, and abductor digiti minimi tendon, corresponding to auxiliary movements such as ulnar deviance of the wrist, extension and abduction of the little finger.

[0086] The correspondence between each electrode and the tendon, and their signal characteristics, are shown in the table below:

[0087]

[0088] Meanwhile, the acquisition parameters for the multi-channel surface electromyography (EMG) signals are set as follows:

[0089] The sampling frequency is 1000-2000Hz, which can fully capture the dynamic changes of electromyography (EMG) signals; the signal acquisition range is 50-500μV, which is suitable for the amplitude characteristics of surface EMG signals; a bandpass filter of 20-500Hz is used during the acquisition process to effectively remove interference signals and ensure that the acquired raw EMG signals have high fidelity.

[0090] In one alternative implementation, signal separation is based on the differences in frequency domain characteristics and / or time-frequency distribution characteristics of electromyographic signals corresponding to different hand tendons, in order to distinguish electromyographic signal components driven by hand tendons with adjacent anatomical locations.

[0091] It is understandable that due to inherent differences in anatomical depth, muscle fiber type, and nerve control pathway, even if the anatomically adjacent tendons (such as the flexor digitorum superficialis tendon FDS and the flexor digitorum profundus tendon FDP) result in high signal coupling, the frequency distribution and time-frequency dynamic changes of their electromyographic signals still have quantifiable differences.

[0092] Based on this, the separation of frequency domain features based on Fast Fourier Transform (FFT) in this embodiment can be achieved by matching and classifying the energy components in different frequency ranges according to the preset tendon frequency domain feature database, thus completing the initial separation of the signal.

[0093] Considering the non-stationary nature of electromyographic signals, wavelet transform technology can be used to analyze the time-frequency distribution characteristics. This not only solves the time-frequency resolution contradiction of traditional Fourier transform, but also accurately captures the dynamic characteristics of the signal in the two-dimensional time-frequency space.

[0094] Furthermore, adjacent tendons exhibit differences in activation timing during synergistic movements. For instance, when the fingers flex, the FDS signal activates 10-20ms earlier than the FDP signal. Moreover, the energy of the extensor tendons is concentrated in the mid-to-high frequency range and is dispersed, while the energy of the flexor tendons is concentrated in the mid-to-low frequency range and is more concentrated. The energy of the accessory tendons decays faster. By extracting features such as the energy peak position and energy concentration frequency band from the time-frequency matrix and matching templates, accurate signal separation can be achieved.

[0095] In practical applications, frequency domain features and time-frequency distribution features can be fused to form a multi-dimensional separation strategy. First, the mixed signal is coarsely partitioned using frequency domain features, then fine-grained separation is carried out using time-frequency features, and finally the weights of the two types of features are dynamically adjusted and fused according to the signal quality to further improve the accuracy and robustness of the separation, so as to better adapt to the scenario of adjacent tendons working together in fine hand movements.

[0096] For example, using 24 surface electromyography (EMG) electrodes, the steps to distinguish EMG signal components driven by anatomically adjacent hand tendons based on differences in frequency domain characteristics and / or time-frequency distribution characteristics of EMG signals corresponding to different hand tendons include:

[0097] First, the 24-channel electromyography signals are preprocessed to remove noise, and then multi-dimensional features are extracted.

[0098] Then, these features are input into a specially trained deep learning model that optimizes the separation of FDS and FDP: by utilizing the inherent differences in their impulse response functions (shallow h_FDS(τ) is non-oscillating and decays rapidly, while deep h_FDP(τ) is oscillating and decays slowly), combined with the amplitude difference caused by signal depth (FDP is -20dB weaker than FDS) and the 10-20ms activation timing difference, the model's hierarchical learning ability decouples highly coordinated and overlapping mixed signals.

[0099] Finally, the 24-channel electromyography signals are filtered and feature extracted before being input into the deep learning model. The model uses the differences in impulse response functions, signal amplitude and activation timing of adjacent tendons (such as FDS and FDP) to solve the separation problem caused by their anatomical overlap and functional coordination, and finally outputs 3D coordinates of 21 key points of the hand to achieve accurate representation of fine movements.

[0100] The aforementioned deep learning model was trained on sample data from over 1000 people of different ages, genders, hand shapes, and movement habits. The samples covered various fine motor skills and everyday action scenarios, including grasping, extending, and pinching, ensuring the model had good generalization ability. During training, the 3D coordinates of 21 key points on the hand were used as supervision labels. These 21 key points specifically included: 5 for the thumb (covering key motor units such as the metacarpophalangeal joint, proximal interphalangeal joint, and distal interphalangeal joint), 4 each for the index to little fingers (corresponding to the metacarpophalangeal joint, proximal interphalangeal joint, middle interphalangeal joint, and distal interphalangeal joint, respectively), and 1 for the palm / wrist (as a motion reference anchor point). Through multiple rounds of iterative optimization of model parameters, the model was able to accurately map the correspondence between electromyographic signal components and hand spatial pose.

[0101] Furthermore, in practical applications, the electrodes are configured as follows: 24 silver / silver chloride electrodes are arranged in a ring on the inner side of the wristband, with a sampling frequency of 1000-2000Hz and a signal range of 50-500μV.

[0102] Because the acquisition effect of surface electromyography signals is easily affected by individual differences (such as variations in tendon anatomical position and muscle development), skin condition (such as skin humidity and stratum corneum thickness), and the tightness of electrode wearing, the signal quality acquired by different electrode channels varies.

[0103] To further optimize signal quality, in one optional embodiment, before performing signal separation on the processed multi-channel surface electromyography (EMG) signals, channel filtering and / or channel weighting processing are further included. This includes: calculating at least one of the signal-to-noise ratio (SNR) parameter, signal amplitude stability parameter, and saturation state parameter for each channel of the EMG signal, and performing channel filtering and / or channel weighting processing on the multi-channel EMG signals based on the aforementioned parameters. By reducing the influence of low-quality channels on the overall signal before signal separation, the stability and consistency of the separated EMG signal components are improved.

[0104] As attached Figure 6 As shown, in practical applications, the hand motion reference data generated based on the aforementioned steps serves as a motion reference model, reflecting the true driving state of hand movements from the perspective of tendon physiological activity. Meanwhile, the three-dimensional hand motion model generated based on the visual tracking mechanism of virtual reality devices is used to characterize the hand motion reconstruction results under external perception conditions. By aligning and comparing the motion parameters or spatial pose parameters of these two types of models at corresponding time points, the differences between them are obtained, thus providing a reference for performance evaluation of hand motion recognition or reconstruction schemes.

[0105] It should be noted that the appendix Figure 6 The model shown is for illustrative purposes only. This invention does not limit the benchmark data or comparison objects to a specific modeling method or output format.

[0106] Furthermore, the benchmark generation method provided in this embodiment also includes:

[0107] The generated hand motion reference data is synchronized with the hand motion recognition or reconstruction results on the time axis, and the difference calculation or deviation statistical processing is performed on the motion parameters or spatial pose parameters at the corresponding time points.

[0108] The motion parameters include tendon activation state and motion execution speed, while the spatial pose parameters include 3D coordinates of 21 key points of the hand and joint angles.

[0109] In this embodiment, the baseline data of hand movements generated based on surface electromyography is first synchronized with the results of hand movement recognition or reconstruction based on visual tracking on the time axis using a unified timestamp. Then, for the corresponding time points, the absolute difference is calculated for the movement parameters (including tendon activation state and movement execution speed) and spatial pose parameters (including 3D coordinates of 21 key points of the hand and joint angles), and the average deviation, maximum deviation and standard deviation of the deviation are statistically analyzed to provide objective data support for quantitatively evaluating the accuracy and stability of the hand movement recognition or reconstruction results.

[0110] By comparing and analyzing benchmark data obtained based on surface electromyography with motion results obtained based on other sensing methods, we can evaluate the differences in temporal consistency and motion accuracy among different hand motion acquisition schemes, thereby providing an objective reference for the performance verification of hand motion recognition systems.

[0111] Please refer to Figure 8 Based on the foregoing embodiments, this embodiment of the invention provides a hand movement reference generation system based on surface electromyography, including a surface electromyography acquisition module 10, a signal processing module 20, a signal separation module 30, and a reference data generation module 40.

[0112] The surface electromyography (EMG) acquisition module 10 is equipped with multiple surface EMG electrodes inside, which are used to acquire multi-channel surface EMG signals at predetermined locations on the hand.

[0113] Specifically, the surface electromyography (SEMG) acquisition module 10 is equipped with multiple SEMG electrodes for acquiring multi-channel SEMG signals from predetermined locations on the hand. The SEMG electrodes are made of silver / silver chloride and are arranged in a ring array along the circumference of the wrist. The preferred number is 24 (expandable to 32). They are worn 2-3 cm proximal to the wrist crease and divided into four acquisition areas: the back of the hand, the palm side, the radial side, and the ulnar side, fully covering the 17 major tendons of the wrist. The acquisition parameters are a sampling frequency of 1000-2000 Hz and a signal range of 50-500 μV. It also has a built-in 20-500 Hz bandpass filter to remove ECG and motion artifacts. In addition, it has an adaptive algorithm that can select the electrode combination with the best signal quality based on individual differences and skin conditions.

[0114] The signal processing module 20 is used to process multi-channel surface electromyography signals by combining the arrangement relationship of surface electromyography electrodes with the tendon distribution at a predetermined position on the hand.

[0115] Specifically, the signal processing module 20 is used to process multi-channel surface electromyography (SEMG) signals by combining the setting relationship of the surface electromyography electrodes with the tendon distribution at a predetermined position on the hand; specifically, based on the anatomical correspondence between the electrodes and the "nine tendons" and "six tubes" of the wrist, the acquired raw signals are filtered, denoised, amplified, and feature extracted, extracting multi-dimensional features in the time domain, frequency domain, etc., to provide accurate data support for subsequent signal separation.

[0116] The signal separation module 30 is used to separate the processed multi-channel surface electromyography signals based on the differences in signal transmission characteristics from different hand tendons to surface electromyography electrodes, so as to obtain electromyography signal components driven by different hand tendons.

[0117] Specifically, the signal separation module 30 is used to separate the processed multi-channel surface electromyography (EMG) signals based on the differences in signal transmission characteristics from different hand tendons to surface EMG electrodes, in order to obtain EMG signal components driven by different hand tendons. It focuses on tendons with adjacent anatomical locations (such as FDS and FDP), and utilizes their differences in impulse response function, signal amplitude (-20dB), and activation timing (10-20ms) to achieve precise decoupling of highly coupled signals through signal modeling and feature analysis.

[0118] The reference data generation module 40 is used to generate reference data for hand movements based on electromyographic signal components.

[0119] Specifically, the benchmark data generation module 40 is used to generate benchmark data for hand movements based on electromyographic signal components. The separated electromyographic signal components are input into a deep learning model trained with over 1000 samples, outputting the 3D coordinates of 21 key points of the hand. This benchmark data can be used for VR interaction accuracy verification or directly associated with VR displays to achieve high-precision projection of hand movements. For details on the projection style, please refer to... Figure 8 .

[0120] Based on the foregoing embodiments, this invention provides a computer device including a processor and a memory. The memory stores a computer program for executing the above-described method for generating hand movement references based on surface electromyography. When the processor executes the computer program, the corresponding method steps can be completed.

[0121] Based on the foregoing embodiments, this invention provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are executed by a processor, the above-described method for generating hand movement references based on surface electromyography can also be implemented.

[0122] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not be construed as limiting the scope of protection of this application. Any technical solutions resulting from equivalent structural or procedural substitutions or modifications made based on the essential concept of this application and utilizing the content described in the text and drawings of this application, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of protection of this application.

Claims

1. A surface myoelectric-based hand motion reference generation method, characterized by, The method comprises: acquiring, by a plurality of surface electromyography electrodes, multi-channel surface electromyography signals at predetermined positions of a hand; processing the multi-channel surface electromyography signals based on a setting relationship of the surface electromyography electrodes relative to tendon distribution at the predetermined positions of the hand; performing signal separation on the processed multi-channel surface electromyography signals based on signal transmission characteristic differences of different hand tendons to the surface electromyography electrodes to obtain electromyography signal components driven by different hand tendons; and generating reference data of a hand movement based on the electromyography signal components.

2. The sEMG-based hand motion reference generation method according to claim 1, wherein The multi-channel surface electromyography signals are acquired at a wrist of the hand.

3. The sEMG-based hand motion reference generation method according to claim 2, wherein The surface electromyography electrodes are arranged in a ring array along a circumference of the wrist.

4. The sEMG-based hand motion reference generation method according to claim 3, wherein The surface electromyography electrodes are divided into at least four acquisition regions of a dorsal side, a palm side, a radial side, and an ulnar side according to anatomical divisions of the wrist, and each of the acquisition regions corresponds to extensor tendons and flexor tendons covering different anatomical divisions.

5. The sEMG-based hand motion reference generation method according to claim 1, wherein The signal transmission characteristic differences include frequency domain characteristics and / or time-frequency distribution characteristics.

6. The sEMG-based hand motion reference generation method according to claim 1, wherein Before the signal separation on the processed multi-channel surface electromyography signals, the method further comprises: calculating at least one of a signal-to-noise ratio parameter, a signal amplitude stability parameter, and a saturation state parameter for each of the multi-channel surface electromyography signals corresponding to the surface electromyography electrodes; and performing channel screening processing and / or channel weighting processing on the multi-channel surface electromyography signals based on at least one of the signal-to-noise ratio parameter, the signal amplitude stability parameter, and the saturation state parameter.

7. The sEMG-based hand motion reference generation method according to claim 1, wherein The method further comprises: synchronizing and aligning the reference data of the hand movement with a hand movement recognition or reconstruction result on a time axis; and performing difference calculation or deviation statistical processing on movement parameters or spatial pose parameters at corresponding time points.

8. A surface myoelectric-based hand motion reference generation system, comprising: The method comprises: a surface electromyography acquisition module comprising a plurality of surface electromyography electrodes, the plurality of surface electromyography electrodes being configured to acquire multi-channel surface electromyography signals at predetermined positions of a hand; a signal processing module configured to process the multi-channel surface electromyography signals based on a setting relationship of the surface electromyography electrodes relative to tendon distribution at the predetermined positions of the hand; a signal separation module configured to perform signal separation on the processed multi-channel surface electromyography signals based on signal transmission characteristic differences of different hand tendons to the surface electromyography electrodes to obtain electromyography signal components driven by different hand tendons; and a reference data generation module configured to generate reference data of a hand movement based on the electromyography signal components. 9.A computer device, comprising a processor and a memory, wherein the memory stores a computer program, and the computer device is configured to execute the computer program to perform the method according to any one of claims 1-8. The processor executes the computer program to implement the surface electromyography-based hand movement reference generation method in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer- executable instructions, wherein, The computer executable instructions are executed by the processor to implement the surface electromyography-based hand movement reference generation method in any one of claims 1 to 7.

Images