A personalized dance healing mood analysis method and system based on real-time motion capture

Abstract

The application relates to the technical field of motion capture analysis, in particular to a personalized dance healing mood analysis method and system based on real-time motion capture, which comprises the steps of synchronizing a first motion capture node with a second motion capture node, acquiring a first data frame carrying a timestamp, analyzing a control header, if a state indication bit is set, suspending scheduling according to a basic period, and broadcasting preemption signaling, merging a second data frame and the first data frame with a timestamp difference less than a time tolerance or triggering preemption into a combined unit, extracting a single-dimensional scalar component sequence of a relative displacement difference vector in a main vibration direction axis in the combined unit, counting the number of zero-crossing alternations of the single-dimensional scalar component sequence in a sliding time window, outputting a motion mode label based on the number of zero-crossing alternations, and generating a mood analysis result. The application can overcome micro-motion cache overflow and time domain misplacement, preempts a channel according to a state indication, and is hard-bound, so that time sequence alignment is realized.

Description

Technical Field

[0001] This application relates to the technical field of motion capture analysis, and in particular to a personalized dance therapy emotion analysis method and system based on real-time motion capture. Background Technology

[0002] Motion capture technology has a broad application base in high-dynamic scenarios such as personalized dance therapy and emotion analysis. These systems typically use wireless body area network (BNB) sensor nodes deployed across various parts of the body to collect multi-axis inertial data in real time, enabling continuous monitoring of macroscopic motion displacement and microscopic limb tremor characteristics. The BNB media access control layer bears the core function of scheduling communication between nodes and allocating underlying channel resources. The time alignment accuracy and high-frequency response capability of its data scheduling mechanism directly determine the reliability and effectiveness of transient motion capture data streams.

[0003] Regarding media access control in wireless body area networks (WBANs), patent application CN117998458A discloses a priority-based MAC method for WBANs. This scheme divides network load into three categories—urgent data, periodic data, and ordinary data—based on data application characteristics, and resets the communication superframe structure into a beacon phase, a contention-based access phase, a non-contention-based access phase, and a sleep phase. The system dynamically calculates transmission priority based on node category, buffer occupancy rate, traffic generation rate, and packet length. Low-priority ordinary nodes preempt the channel during the contention-based access phase using the CSMA / CA mechanism, while high-priority nodes obtain dedicated time slots allocated by the coordinator in descending order of priority during the non-contention-based access phase using the TDMA mechanism. In this mechanism, if the physiological data collected by a node exceeds a set high / low threshold, the system adjusts the subsequent data transmission status by raising the node category by one level.

[0004] The scheduling mechanism essentially relies on whether the data collected by a node exceeds a preset static threshold to trigger a priority change. However, it has inherent limitations in its underlying scheduling when dealing with highly dynamic motion capture scenarios such as dance therapy. When the human body performs large-scale macroscopic movements, a massive amount of high-frequency sampling data of limb micro-movements typically bursts within a very short time window. At this time, the network faces high-frequency abrupt changes in the overall dynamic state of the system. Due to the limitations of existing technologies that rely on the solidified superframe timing flow at each stage, and the fact that time slot allocation follows a post-request and pre-planning pattern, when sudden high-frequency micro-movement data streams and macroscopic large movements are instantaneously superimposed, nodes lack a dynamic channel preemption intervention mechanism based on the underlying transient kinematic state. Conventional nodes must undergo lengthy backoff competition or passively wait for dedicated allocation in the next superframe cycle, causing local buffer congestion and overflow during the long-term competition for uplink authorization for high-frequency micro-motion data. At the same time, this non-correlated timing scheduling and lagging priority determination make it easy for co-originating data frames collected by sensor nodes in different locations to generate severe timestamp phase discrepancies when transmitted to the aggregation end. Conventional on-demand polling and fixed-time-period competition frameworks are difficult to overcome the time-domain misalignment and micro-physical feature omission problems caused by multiple nodes concurrently transmitting high-dynamic-change signals. Summary of the Invention

[0005] To overcome micro-motion buffer overflow and time domain misalignment, this application provides a personalized dance therapy emotion analysis method and system based on real-time motion capture, which uses state indicators to preempt the channel and hard-bind it to achieve time alignment.

[0006] Firstly, this application provides a personalized dance therapy emotion analysis method based on real-time motion capture, employing the following technical solution: A personalized dance therapy emotion analysis method based on real-time motion capture, comprising:

[0007] Send the first threshold, the second threshold, the time tolerance, and the base period to the preset first motion capture node and the preset second motion capture node, and broadcast a beacon frame carrying the clock reference;

[0008] The first motion capture node and the second motion capture node are synchronized according to the clock reference, and the first data frame carrying the timestamp and control header reported by the first motion capture node is obtained according to the basic period. The status indication bit in the control header is set by the first motion capture node when the magnitude of its calculated comprehensive motion feature integral vector is greater than the first threshold.

[0009] Parse the control header; if the status indicator bit is set, pause scheduling according to the basic cycle and broadcast a preemption signaling message containing a burst of resource blocks.

[0010] In response to the preemption signaling, when the magnitude of the relative displacement difference vector cached locally by the second motion capture node is greater than the second threshold, a second data frame carrying a timestamp is reported within the burst resource block at a frequency higher than the basic periodicity.

[0011] The second data frame and the first data frame whose timestamp difference is less than the time tolerance or which triggers preemption are merged into a combined unit;

[0012] Extract the one-dimensional scalar component sequence of the relative displacement difference vector in the combined unit on the main tremor direction axis, and count the number of zero-crossing alternations of the one-dimensional scalar component sequence within the sliding time window. Output the action modality label based on the number of zero-crossing alternations to generate sentiment analysis results.

[0013] Optionally, the status indicator bit in the control header is set by the first motion capture node when the magnitude of its calculated integrated motion feature vector is greater than the first threshold, including: the first motion capture node acquiring multi-axis acceleration and angular velocity within the base period; normalizing the multi-axis acceleration and angular velocity and integrating them along the time axis to generate the integrated motion feature vector; when the magnitude of the integrated motion feature vector is greater than the first threshold within consecutive preset sampling periods, the reserved bit in the control header is redefined as the status indicator bit and set.

[0014] Optionally, the generation method of the relative displacement difference vector in the local cache of the second motion capture node includes: acquiring the original sequence of the second motion capture node at a preset sampling rate and storing it in a preset cache time window; filtering the original sequence in the cache time window to remove components with frequencies lower than a preset cutoff frequency to extract jitter components; performing differential transformation on the jitter components to generate the relative displacement difference vector and updating it in the local cache so that its magnitude can be compared with the second threshold.

[0015] Optional. The broadcast includes preemption signaling for bursty resource blocks, including:

[0016] The preemption signaling, which is configured with the burst resource block and its duration, is broadcast within a preset downlink control time slot.

[0017] The method further includes:

[0018] While broadcasting the preemption signaling, a resource protection timer with a preset timeout threshold is simultaneously started to monitor the occupancy status of the sudden resource block.

[0019] Optionally, monitoring the occupancy status of the burst resource block includes: during the process of acquiring the second data frame, if the magnitude of the relative displacement differential vector cached locally by the second motion capture node is continuously not greater than the second threshold within a preset state maintenance period, and a short control frame containing a single-bit silent flag is sent within the burst resource block, then in response to the short control frame, a channel release confirmation instruction is issued to instruct the second motion capture node to revert to polling scheduling according to the basic period, and the resource protection timer is reset; if the resource protection timer reaches the preset timeout threshold and the short control frame is not received, then the burst resource block is forcibly reclaimed.

[0020] Optionally, it also includes: extracting the uplink channel quality indicator of the second capture node during the broadcast of the preemption signaling; if the uplink channel quality indicator is lower than a preset link disconnection threshold, triggering a link adaptive mode, reconfiguring the allocated burst resource block as a sideline time slot, and issuing a sideline relay scheduling signaling to the first capture node and the second capture node; based on the sideline relay scheduling signaling, instructing the second capture node to pass the second data frame through the sideline time slot to the first capture node; obtaining the dual-node data frame reported by the first capture node, and parsing out the first data frame and the passed-through second data frame from it for merging into the combining unit.

[0021] Optionally, when there are multiple second capture nodes, the preemption signaling is configured with multiple allocation indexes of the burst resource blocks that correspond one-to-one with each of the second capture nodes and do not overlap;

[0022] Within the burst resource block, acquiring the second data frame carrying a timestamp reported by the second motion capture node at a frequency higher than the basic cycle frequency specifically involves:

[0023] Within their respective burst resource blocks, each of the second motion capture nodes concurrently acquires the second data frames carrying the timestamps reported at a frequency higher than the base period.

[0024] Optionally, the step of merging the second data frame with the first data frame whose timestamp difference is less than the time tolerance or which triggers preemption into a combination unit includes:

[0025] Compare the timestamps of the first data frame and the second data frame;

[0026] Extract the first payload of the first data frame whose timestamp difference is less than the time tolerance or triggers preemption, and the second payload of the second data frame;

[0027] The first payload and the second payload are hard-bonded and encapsulated to generate the combined unit.

[0028] Optionally, the step of extracting the single-dimensional scalar component sequence of the relative displacement difference vector within the combined unit along the main tremor direction axis, and counting the number of zero-crossing alternations of the single-dimensional scalar component sequence within the sliding time window, and outputting action modality labels based on the number of zero-crossing alternations to generate sentiment analysis results, includes:

[0029] Within the sliding time window, the relative displacement difference vector sequence within the second effective load of the consecutive combined units is extracted, and the one-dimensional scalar component sequence of the relative displacement difference vector sequence on the main flutter direction axis is extracted;

[0030] The number of zero-crossing alternations is obtained by counting the number of times the single-dimensional scalar component sequence crosses a preset zero baseline;

[0031] If the number of zero-crossing alternations is greater than a preset alternation threshold, the first payload in the corresponding combination unit is extracted, and the first payload is bound to the number of zero-crossing alternations to generate the action modality label; based on the preset alternation threshold range where the number of zero-crossing alternations is located, the action modality label is mapped to the corresponding specific healing emotion and written into the emotion analysis result.

[0032] Secondly, this application provides a personalized dance therapy emotion analysis system based on real-time motion capture, which adopts the following technical solution: A personalized dance therapy emotion analysis system based on real-time motion capture, the system comprising:

[0033] The basic scheduling module is used to send a first threshold, a second threshold, a time tolerance, and a basic period to a preset first motion capture node and a preset second motion capture node, and broadcast a beacon frame carrying a clock reference; synchronize the first motion capture node and the second motion capture node according to the clock reference, and acquire the first data frame carrying a timestamp and a control header reported by the first motion capture node according to the basic period. The status indication bit in the control header is set by the first motion capture node when the magnitude of its calculated integrated motion feature integral vector is greater than the first threshold.

[0034] The status triggering and preemption module is used to parse the control header. If the status indicator bit is set, the scheduling according to the basic cycle is paused, and a preemption signaling containing a burst of resource blocks is broadcast.

[0035] A burst acquisition module is used to, in response to the preemption signaling, acquire a second data frame carrying a timestamp that is reported within the burst resource block at a frequency higher than the basic period when the magnitude of the relative displacement difference vector cached locally by the second motion capture node is greater than the second threshold.

[0036] The fusion analysis module is used to merge the second data frame and the first data frame whose timestamp difference is less than the time tolerance or triggers preemption into a combined unit; extract the single-dimensional scalar component sequence of the relative displacement difference vector in the combined unit on the main tremor direction axis, and count the number of zero-crossing alternations of the single-dimensional scalar component sequence within the sliding time window, and output the action modality label based on the number of zero-crossing alternations to generate sentiment analysis results.

[0037] In summary, this application includes the following beneficial technical effects:

[0038] 1. The first motion capture node calculates the integral vector magnitude of the comprehensive motion characteristics and sets the status indicator bit when the magnitude exceeds the threshold. After parsing, the scheduling end suspends the basic cycle scheduling and broadcasts the preemption signaling containing the burst resource block. This enables the second motion capture node to report the second data frame at a frequency higher than the basic cycle when the relative displacement differential vector magnitude in the local buffer exceeds the threshold, thus overcoming the buffer overflow problem of micro-motion data in channel contention. The second data frame and the first data frame whose timestamp difference is less than the time tolerance or triggers preemption are merged into a combination unit, eliminating the timestamp phase misalignment of multi-node data and realizing the timing alignment of macroscopic actions and micro-judders.

[0039] 2. By synchronously starting a resource protection timer when broadcasting preemptive signaling, and dynamically releasing or forcibly reclaiming burst resource blocks based on the single-bit silent flag reported by the second capture node or the timer timeout threshold, the channel resources are prevented from being invalidally occupied for a long time due to signaling loss, which effectively improves the channel utilization efficiency and scheduling reliability of the wireless body area network.

[0040] 3. By extracting the single-dimensional scalar component sequence of the relative displacement difference vector within the combined unit on the main tremor direction axis, and counting the number of zero-crossing alternations within the sliding time window, the action modality label is mapped to the corresponding specific healing emotion based on the preset alternation number threshold range, thereby realizing the quantitative identification of human emotion-related micro-tremor features and improving the accuracy of emotion analysis results in dance therapy scenarios. Attached Figure Description

[0041] Figure 1 This is a logical flowchart of the method in the embodiments of this application;

[0042] Figure 2 This is a schematic diagram of the physical space deployment and network architecture of the system in the embodiment of this application. Detailed Implementation

[0043] The following combination Figures 1-2 This application will be described in further detail.

[0044] like Figure 1 As shown, this application relates to the fields of motion capture analysis and wireless body area network (WBN) communication technology, specifically a personalized dance-based therapy emotion analysis method based on real-time motion capture. The system upon which this method relies comprises four core entities: a preset first motion capture node, a preset second motion capture node, a scheduling gateway, and a processing server. All entities follow a unified wireless body area network media access control layer protocol for data interaction and timing synchronization.

[0045] The first motion capture node is pre-deployed in the human torso area and has a built-in nine-axis inertial measurement unit and a wireless microcontroller. The nine-axis inertial measurement unit is used to collect multi-axis acceleration and angular velocity data during human movement, and the wireless microcontroller is used to perform data normalization, integration calculation, status indicator configuration and data reporting functions.

[0046] The second motion capture node is pre-deployed in the extremity region of the human limb. It has a built-in six-axis inertial measurement unit, a wireless microcontroller and a local cache module. The six-axis inertial measurement unit is used to collect the raw motion sequence data of the human limb extremity. The wireless microcontroller is used to perform data filtering, differential conversion, threshold comparison and data reporting functions. The local cache module is used to store the raw motion sequence and the processed relative displacement differential vector data within a preset time period.

[0047] The scheduling gateway is the central control entity of the wireless body area network. It has a built-in radio frequency transceiver module, timing control module and data fusion module. The radio frequency transceiver module is used to broadcast beacon frames, issue control signaling and receive data frames reported by each node. The timing control module is used to perform global clock synchronization, basic periodic scheduling, burst resource block allocation and channel state management. The data fusion module is used to perform timestamp comparison of multi-source data frames and combination unit encapsulation.

[0048] The solution server is the backend data processing entity, with built-in modality recognition module and emotion analysis module. The modality recognition module is used to count the number of zero-crossing alternations of the relative displacement difference vector and generate action modality labels. The emotion analysis module is used to generate corresponding personalized dance therapy emotion analysis results based on the action modality labels.

[0049] Reference Figure 2In the physical deployment configuration, a first motion capture node is pre-attached to the torso region of the target object, and a second motion capture node is pre-attached to the extremities of the human limbs. Both the first and second motion capture nodes are configured to establish a bidirectional data interaction connection with the scheduling gateway via a wireless communication link. The scheduling gateway, as the central aggregation node of the wireless body area network, after acquiring data frames aligned with micro and macro data, transmits the combined unit to the backend-configured processing server via a configured high-speed data transmission link for further analysis. The following steps will be described in detail:

[0050] S1 Global Clock Synchronization and System Operating Parameter Initialization

[0051] This step is the initial execution step of the method. Before execution, the initial system state is that the scheduling gateway has completed the initialization of the hardware circuit and the wireless body area network media access control layer protocol stack, and has entered the master control ready state. The preset first motion capture node and the preset second motion capture node have completed power-on self-test and sensor initialization, and have entered the wireless channel idle listening state. All nodes have not established a unified time base, have not been configured with unified operating parameters, and have not started the data acquisition and reporting process.

[0052] Based on the kinematic characteristics of the human body in a dance therapy scenario and the wireless body area network communication standard, the scheduling gateway generates four core operating parameters: a preset first threshold, a preset second threshold, a preset time tolerance, and a basic cycle. The scheduling gateway sends these four core operating parameters to all preset first and preset second motion capture nodes via dedicated downlink control frames. Upon receiving the downlink control frames, the preset first and preset second motion capture nodes extract the operating parameters from the frames and write them to their local non-volatile storage units, completing the local configuration of the operating parameters.

[0053] The base period is set to 50ms, corresponding to a base reporting frequency of 20Hz. The base period is determined based on the main frequency range of macroscopic limb movements in the dance therapy scenario. The main frequency of macroscopic limb movements during human dance is distributed in the range of 0.5Hz to 5Hz. According to the Nyquist sampling theorem, the sampling frequency must be higher than twice the highest main frequency of the movement, corresponding to a sampling period of no more than 100ms. This value also takes into account the low power consumption requirements of the wireless body area network node, balancing data acquisition accuracy and node endurance.

[0054] The preset first threshold is 10 times the average magnitude of the integral vector of comprehensive motion features in the normal static state of the human body. The preset first threshold is determined based on the statistical results of the human kinematics database. In the dance therapy scenario, the normalized integral vector of comprehensive motion features corresponding to large-amplitude limb movements of the human body is 8 to 12 times the magnitude corresponding to normal static or small-amplitude movements. This value can filter out false triggering interference caused by environmental vibration and sensor noise while ensuring the sensitivity of macroscopic movement recognition.

[0055] The preset second threshold value is 0.05 mm. The preset second threshold value is determined based on the displacement characteristics of muscle micro-tremors related to human emotions. The micro-tremors of the limbs generated during the release of human emotions correspond to a relative displacement distribution in the range of 0.05 mm to 0.2 mm. This value can effectively filter out random sampling noise with an amplitude of less than 0.05 mm, and retain only the effective micro-tremor data components related to human emotional state.

[0056] The preset time tolerance is 3ms. The preset time tolerance is determined based on the communication delay characteristics of the wireless body area network and the node sampling synchronization error. The maximum delay jitter of data transmission within the wireless body area network does not exceed 2ms, and the sampling synchronization error between nodes does not exceed 1ms. This value can cover the timing deviation in the transmission and sampling process, taking into account both the accuracy of data matching and the retention rate of effective data.

[0057] While distributing operating parameters, the scheduling gateway generates a global clock reference based on its local high-precision hardware clock and periodically broadcasts beacon frames carrying this clock reference. The broadcast period of the beacon frames is consistent with the base period. The local high-precision hardware clock used by the scheduling gateway has an accuracy of no less than 1μs, providing a stable and unified time reference for the entire network. Preset first and second motion capture nodes continuously monitor the beacon frames broadcast by the scheduling gateway. Upon receiving each beacon frame, they extract the global clock reference carried within the frame, calculate the offset between the local clock and the global clock reference, and correct the timing frequency and timing phase of the local clock based on this offset.

[0058] If the preset first motion capture node or the preset second motion capture node fails to receive a beacon frame from the scheduling gateway for three consecutive basic cycles, it is determined to be out of sync. The node suspends data reporting and enters a continuous beacon frame listening state. If no valid beacon frame is received for 30 consecutive basic cycles, it enters a low-power sleep state and wakes up to listen for beacon frames every 10 basic cycles. After receiving a beacon frame again, clock offset correction and synchronization are performed again. After synchronization is completed, normal working state is restored.

[0059] Through continuous beacon frame monitoring and clock offset correction, the synchronization error between the local clocks of the preset first motion capture node and the preset second motion capture node and the global clock reference of the scheduling gateway is controlled within 1μs, so that all nodes in the network maintain a unified time reference, thus avoiding the accumulation of timing deviations in the multi-node data acquisition process from the root.

[0060] After the second motion capture node completes parameter configuration and clock synchronization, it immediately starts a continuous background data preprocessing process. The second motion capture node (whose six-axis inertial measurement unit has a built-in displacement calculation engine, and its output raw motion sequence is the spatial displacement coordinate sequence) continuously collects the raw motion sequence at a preset sampling rate of 1000Hz and stores it in a preset buffer time window.

[0061] The preset buffer time window adopts a cyclic first-in-first-out structure with a length of 100ms. When the buffer time window is full, it automatically overwrites the earliest sampled data. At the same time, an overflow flag is set in the local buffer. If an overflow occurs for 10 consecutive sampling cycles, the overflow flag will be carried in the second data frame reported in the subsequent report.

[0062] The second motion capture node first performs a pre-processing fourth-order Butterworth low-pass anti-aliasing filter on the original sequence within the buffer time window, with the cutoff frequency set to 500Hz and the stopband attenuation greater than 60dB. Then, the filtered sequence is subjected to high-pass filtering to remove components with frequencies lower than the preset cutoff frequency in order to extract the jitter components.

[0063] The preset high-pass filter is a third-order Butterworth high-pass filter with a preset cutoff frequency of 8Hz. The passband ripple is less than 1dB and the stopband attenuation is greater than 40dB. It can completely eliminate macroscopic limb displacement components with frequencies below 8Hz and physically block the DC spatial coordinate drift caused by the low-frequency double integration of the inertial sensor. This ensures that the extracted tremor components can strictly represent the real local high-frequency AC amplitude, ensuring the effectiveness of the 0.05mm preset second threshold comparison and retaining only the microscopic muscle tremor components related to human emotional state.

[0064] The preset second motion capture node performs a first-order backward differential transformation on the extracted tremor components (using a delay subtraction algorithm with a span of 50 sampling points to equivalently extract peak-to-peak values), generating a relative displacement difference vector and updating it in the local cache for subsequent comparison with a preset second threshold. The unit of the relative displacement difference vector is mm, used to characterize the relative displacement change of the limb end under a set delay span.

[0065] After this step is completed, all preset first motion capture nodes and preset second motion capture nodes in the entire network will complete global clock synchronization and local configuration of core operating parameters. The preset first motion capture node will enter the basic cycle data acquisition and reporting ready state, and the preset second motion capture node will enter the continuous data acquisition and preprocessing state.

[0066] S2 Basic Cycle Macroscopic Motion State Sensing and Data Reporting

[0067] This step begins with the final output state of S1. The scheduling gateway and all preset first motion capture nodes maintain microsecond-level global clock synchronization. The preset first motion capture nodes have completed the local configuration of the preset first threshold and basic cycle, and enter the ready state for basic cycle data acquisition and reporting.

[0068] The scheduling gateway allocates fixed uplink time slots to the preset first motion capture node according to the synchronized global clock and a preset basic cycle, and receives the first data frame carrying a timestamp and control header reported by the preset first motion capture node within the corresponding time slot. Within each basic cycle, the preset first motion capture node continuously acquires multi-axis acceleration and angular velocity data at a preset sampling rate of 1000Hz. Multi-axis acceleration includes linear acceleration in the X, Y, and Z axes of a Cartesian coordinate system, with units of m / s². 2 Angular velocity includes angular rate data corresponding to the three coordinate axes, measured in rad / s. A sampling rate of 1000Hz fully covers the transient dynamic changes of large-amplitude limb movements in dance therapy scenarios, while matching the conventional sampling performance of the inertial measurement unit, ensuring that the collected data can accurately reproduce the complete process of macroscopic human limb movements.

[0069] The preset first motion capture node normalizes the acquired multi-axis acceleration and angular velocity data. It divides the linear acceleration and angular velocity within each sampling period by the corresponding sensor's full-scale output value to obtain dimensionless normalized linear acceleration and angular velocity components. This normalization process eliminates calculation biases caused by differences in sensor hardware batches and measurement ranges, ensuring that the kinematic parameters output by different preset first motion capture nodes have a unified judgment standard.

[0070] If the multi-axis acceleration and angular velocity data collected by the preset first motion capture node exceeds the sensor's full range, or if the data is zero for five consecutive sampling cycles, it is determined to be abnormal. Subsequent calculations of the integrated motion feature vector and status indicator bit settings will not be performed. Simultaneously, an abnormal flag will be set in the control header of the first data frame and reported to the scheduling gateway. If the data from the preset first motion capture node recovers to the normal range for 10 consecutive sampling cycles, the abnormal state will be exited, and kinetic energy integration calculation and status indicator bit reporting will resume. If the abnormality persists for five consecutive basic cycles, the scheduling gateway will send a sensor calibration command to the node, triggering hardware calibration.

[0071] When there are no data anomalies, the preset first motion capture node performs a resultant vector calculation on the normalized linear acceleration component and normalized angular velocity component at the same sampling time. Then, through dynamic weight fusion based on human kinematics characteristics, a comprehensive motion feature quantity for a single sampling point is obtained. The basic weights are 0.6 for linear acceleration and 0.4 for angular velocity. When the magnitude of the linear acceleration resultant vector is less than 0.05 times the full scale, the angular velocity weight is automatically increased to 0.7; when the magnitude of the angular velocity resultant vector is less than 0.05 times the full scale, the linear acceleration weight is automatically increased to 0.7. The preset first motion capture node then performs a definite integral operation along the time axis on the comprehensive motion feature quantity obtained from continuous sampling within a single basic cycle, generating a comprehensive motion feature integral vector. The upper and lower limits of the definite integral are from the start time of the current basic cycle to the current sampling time, the integration step size is 1 ms corresponding to a single sampling cycle, and the integration result is a dimensionless comprehensive motion feature integral vector with a magnitude ranging from 0 to 1. The integration interval is exactly the same as the time length of the base cycle, ensuring that a unique integrated motion feature vector is generated within each base cycle, thereby achieving a quantitative representation of the macroscopic motion state of the human body within a single cycle.

[0072] Within a single base cycle time window, multi-axis data fusion presents a dynamic evolution process based on the time axis. The resultant vectors of normalized linear acceleration and normalized angular velocity are output as discrete amplitude sequences with a set sampling step size of 1ms. A preset first motion capture node performs dynamic weight fusion and definite integral calculations based on these two sequences, generating a comprehensive motion feature integral vector exhibiting a step-up pattern. When the magnitude of this comprehensive motion feature integral vector touches and crosses a constant preset first threshold horizontal line in the coordinate system during the dynamic accumulation process, a status indicator position trigger point is formed in the time domain. The system's underlying logic uses this trigger point to adjust the corresponding status indicator position in the MAC layer control header.

[0073] The first motion capture node compares the magnitude of the integrated motion feature vector with a pre-configured first threshold in real time. When the magnitude of the integrated motion feature vector is greater than the first threshold for consecutive preset sampling periods, the reserved bit in the MAC layer control header is redefined as a status indicator bit and set.

[0074] The range of the continuously preset sampling period is 2 to 5 sampling periods, with a preferred value of 3. Each sampling period is 1ms corresponding to a 1000Hz sampling rate. Continuous judgment across multiple periods can filter out single-impact noise caused by environmental vibrations and random sensor jumps, ensuring that the status indicator bit is set only when the human body performs continuous large-amplitude macroscopic movements, avoiding invalid status markings. This reserved bit adopts an undefined bit reserved in the IEEE 802.15.6 standard MAC layer control header. Redefining the bit does not require modifying the overall length of the MAC layer frame structure, ensuring compatibility with existing wireless body area network communication protocol frameworks, and simultaneously realizing direct cross-layer mapping from human kinematic state to MAC layer signaling. The preset first motion capture node reports the first data frame carrying a global timestamp and control header to the scheduling gateway within the specified uplink time slot corresponding to each basic period. The timestamp is the global clock value collected by the preset first motion capture node at the start time of the corresponding data in this frame, synchronized with the global clock reference of the scheduling gateway, with an accuracy of 1μs, providing a unified time reference for subsequent timing matching of multi-source data.

[0075] S3 event-triggered preemptive channel scheduling and high-frequency data acquisition

[0076] This step begins with the final output state of S2. The scheduling gateway has completed the reception of the first data frame and the parsing of the MAC layer control header. The first preset motion capture node and the second preset motion capture node in the entire network maintain microsecond-level global clock synchronization with the scheduling gateway. The second preset motion capture node has completed the local configuration of core parameters such as the second preset threshold and the basic period, and has entered the state of continuous raw data acquisition and preprocessing.

[0077] The scheduling gateway continuously parses the MAC layer control header of the received first data frame. If the status indicator bit is set and no abnormal flag bit is detected, it immediately suspends the polling scheduling based on the basic cycle and broadcasts a preemption signaling message containing the burst resource block within a preset downlink control time slot. The downlink control time slot is a fixed downlink signaling transmission time slot reserved within each basic cycle. Control signaling can be directly issued without waiting for a complete superframe cycle, eliminating the scheduling lag problem caused by existing technologies relying on fixed superframe timing. The preemption signaling message is configured with the time slot index and duration of the burst resource block. The time slot index is determined based on a global clock reference, and the duration is set to 500ms to ensure that the preset second motion capture node can accurately lock the time window of the burst transmission, avoiding transmission timing conflicts. While broadcasting the preemption signaling message, the scheduling gateway simultaneously starts a resource protection timer with a preset timeout threshold to monitor the occupancy status of the burst resource block. The preset timeout threshold is set to 520ms, which corresponds to the length of more than 10 basic cycles. This value covers the duration of micro-tremors accompanying macroscopic movements in dance therapy scenarios, and at the same time can avoid the problem of long-term occupation of channel resources due to signaling loss, thus ensuring the orderly use of wireless channel resources.

[0078] The scheduling gateway continuously listens to the uplink channel during the effective duration of the burst resource block, acquires various data frames reported by the preset second dynamic capture node, and synchronously monitors the occupancy status of the burst resource block. If a short control frame containing a single-bit silent flag is sent by the preset second dynamic capture node within the effective duration of the corresponding burst resource block when the magnitude of the relative displacement difference vector buffered locally by the preset second dynamic capture node is not greater than a preset second threshold for a preset state maintenance period, then in response to the short control frame, a channel release confirmation instruction is issued to instruct the preset second dynamic capture node to fall back to polling scheduling based on the basic period and reset the resource protection timer.

[0079] The short control frame is transmitted no later than one-third of the effective duration of the burst resource block. This short control frame includes a preamble and a synchronization word at the physical layer to maintain clock synchronization and frame start positioning at the receiver. Its MAC layer payload is simplified to a single-bit silence flag with a logic low level. This mechanism, while meeting the demodulation requirements of the underlying RF protocol, minimizes the air interface duration, requiring only minimal channel resources to complete node status feedback, avoiding meaningless channel resource occupation, and improving resource utilization efficiency. The channel release confirmation command uses the MAC layer control element format, which can be directly parsed and executed by the media access control layer of the preset second capture node without processing through the upper-layer protocol stack, thus shortening the response latency for state switching.

[0080] If the resource protection timer reaches the preset timeout threshold and the short control frame is not received, the burst resource block is forcibly reclaimed, the basic cycle polling scheduling is restored, and a resource reclamation notification is broadcast to all preset second motion capture nodes, instructing the nodes to revert to the basic cycle polling scheduling. After the forced resource reclamation, the scheduling gateway still performs the normal timing matching and combination unit encapsulation process on the received second data frames to avoid loss of valid data.

[0081] During the broadcast preemption signaling period, the scheduling gateway synchronously extracts the uplink channel quality indicator (HSI) of the preset second motion capture node. The HSI is calculated based on the signal-to-noise ratio (SNR) and packet error rate (PER) of the historically reported data from the preset second motion capture node, and is used to characterize the wireless link transmission quality between the preset second motion capture node and the scheduling gateway. If the HSI is lower than a preset link failure threshold, a link adaptive mode is triggered, reconfiguring the allocated burst resource blocks into side-link time slots and relay uplink time slots for reporting by the preset first motion capture node, and sending side-link relay scheduling signaling to both the preset first and preset second motion capture nodes. The preset link failure threshold corresponds to a signal strength with a PER greater than 80%, which can accurately identify direct link degradation scenarios caused by human torso rotation and limb obstruction in dance therapy scenarios.

[0082] The synchronization reference for the side-link time slot is a global clock reference. The side-link synchronization accuracy between the preset second motion capture node and the preset first motion capture node is controlled within 1μs. Near-field communication uses the 2.4GHz ISM band, the physical layer uses GFSK modulation, the symbol rate is 1Mbps, and the transmit power is 0dBm. Based on the side-link relay scheduling signaling, the scheduling gateway instructs the preset second motion capture node to transmit the second data frame to the preset first motion capture node via the side-link time slot. The second data frame transmitted by the preset second motion capture node adopts a simplified format, containing only 2 bytes of node address, 4 bytes of timestamp, and 8 bytes of relative displacement differential vector data, with a total frame length of 14 bytes. After receiving the transmitted second data frame, the preset first motion capture node aligns it with its own first data frame according to the timestamp, encapsulates it into a dual-node data frame using low-order BPSK modulation, and reports it to the scheduling gateway via the relay uplink time slot. The frame header of the dual-node data frame includes a node identifier bit.

[0083] After the scheduling gateway obtains the dual-node data frame reported by the preset first motion capture node via the relay uplink time slot, it parses the first data frame and the transparent second data frame based on the node identifier bit for subsequent merging into a combination unit. The side-link time slot adopts a near-field low-power transmission mode, which is not affected by human torso obstruction, ensuring reliable transmission of micro-motion data in link deterioration scenarios and eliminating the high-frequency data loss problem caused by body obstruction in the prior art.

[0084] When multiple preset second motion capture nodes exist within the system, the preemption signaling is configured with allocation indexes for multiple burst resource blocks that correspond one-to-one with each preset second motion capture node and do not overlap. The system supports a maximum of 16 preset second motion capture nodes. In scenarios with 4 or fewer preset second motion capture nodes, burst resource blocks are allocated using TDMA time-division multiplexing. The burst resource blocks corresponding to each preset second motion capture node are arranged sequentially in the time domain according to the node address order. The minimum duration of each node's burst resource block is 1ms, and a 10μs protection interval is set between adjacent resource blocks to ensure non-overlap. When the number of nodes exceeds 4, a hybrid TDMA+FDMA method is used, with a frequency point interval of 2MHz and a total of 4 orthogonal frequency points. Each frequency point supports a maximum of 4 nodes' TDMA time slot allocation, and a 1MHz protection bandwidth is set between adjacent frequency points. Within their respective burst resource blocks, the scheduling gateway concurrently acquires the second data frames carrying timestamps reported by each preset second motion capture node at a frequency higher than the basic cycle frequency.

[0085] If the scheduling gateway detects a data collision within a burst resource block, it immediately sends a reallocation signal to the corresponding preset second motion capture node in the downlink control time slot, allocating a new idle burst resource block for it. The maximum number of reallocations is three. If a data collision is still detected after three consecutive reallocations, the scheduling gateway adjusts the node's burst resource block to an independent idle frequency point. If no frequency point can be allocated, a silence command is issued to the node, instructing it to fall back to the basic periodic polling scheduling. This allocation method can completely avoid channel collision problems when multiple nodes transmit concurrently, ensuring the synchronous acquisition and reporting of micro-motion data from multiple limbs, and providing a complete data source for subsequent data processing.

[0086] In response to the preemption signal broadcast by the scheduling gateway, the preset second motion capture node immediately wakes up the local data comparison process. Within the effective duration of the burst resource block, the out-of-synchronization judgment mechanism is suspended, and threshold comparison and reporting operations of the local data are performed. The preset second motion capture node compares the magnitude of the relative displacement difference vector in its local buffer with a pre-configured preset second threshold in real time. Only when the magnitude of the relative displacement difference vector exceeds the preset second threshold is a second data frame carrying a timestamp reported within the burst resource block at a frequency higher than the base period. The reporting frequency higher than the base period is set to 100Hz, corresponding to a reporting period of 10ms, which is 5 times the base reporting frequency. This value meets the Nyquist sampling requirements for the 8Hz to 12Hz frequency range of human muscle micro-tremors, allowing for complete capture of the complete waveform characteristics of the micro-tremors without excessively consuming channel resources.

[0087] The timestamp carried in the second data frame is the global clock value at the time of data acquisition, synchronized with the global clock reference of the scheduling gateway, with an accuracy of 1μs, ensuring the consistency of the timing reference for reported data. After the magnitude of the relative displacement differential vector triggers high-frequency reporting for the first time, a state maintenance timer covering the maximum period of a single micro-flicker is synchronously started (e.g., the timeout is set to 150ms, corresponding to the preset state maintenance period). Even if the magnitude falls below the preset second threshold within this window period, the transmission of single-bit silent frames is forcibly blocked and high-frequency data reporting is maintained to ensure the complete transmission of the zero-crossing waveform. If the magnitude of the relative displacement differential vector remains no greater than the preset second threshold within the preset state maintenance period, the preset second motion capture node transmits a single-bit silent frame within the effective duration of the corresponding burst resource block, without occupying subsequent channel resources.

[0088] S4 Multi-Source Data Temporal Fusion and Sentiment Analysis Result Generation

[0089] This step begins with the final output state of S3. The scheduling gateway has completed the full-process scheduling management of the burst resource block, and has obtained the first data frame reported by the preset first motion capture node and the second data frame reported by the preset second motion capture node. This includes the dual-node data frame reported by the preset first motion capture node in the side relay scenario. All data frames carry a microsecond-level precision timestamp synchronized with the global clock reference. Both the preset first motion capture node and the preset second motion capture node maintain clock synchronization with the scheduling gateway. The solution server and the scheduling gateway establish a stable high-speed data transmission link and enter the data reception and processing ready state.

[0090] The scheduling gateway continuously performs timing matching processing on the received first and second data frames, comparing the timestamps carried by the first and second data frames frame by frame. It extracts the first payload of the first data frame whose timestamp difference is less than a pre-configured preset time tolerance, and the second payload of the second data frame. During a sudden preemption period that pauses basic scheduling, the scheduling gateway by default reuses the first payload of the nearest first data frame that triggered the preemption (performing zero-order hold), using it as shared data to directly hard-bind and encapsulate it with the second payload of all high-frequency reported second data frames during the sudden period. The scheduling gateway then performs frame-payload-level hard binding and encapsulation of the extracted first and second payloads to generate a combined unit.

[0091] The hard-binding encapsulation format has a fixed total frame length of 128 bytes. The frame header is a 4-byte global timestamp reference. The first 60 bytes of the frame body are the first payload, the middle 2 bytes are a fixed separator (0xAA55), and the last 60 bytes are the second payload. The last 2 bytes use a CRC16-CCITT checksum, and the check range covers all data from the frame header to the frame body. Through precise matching of the same-source timestamps and multiplexing extraction during burst preemption, along with hard-binding of the frame payload, a unique temporal correlation is formed between the first payload representing macroscopic human movements and the second payload representing microscopic limb tremors. This completely eliminates the problems of multi-source data timestamp phase misalignment and data interruption during preemption caused by asynchronous network transmission, providing a time-accurately aligned data source for subsequent action feature recognition.

[0092] The scheduling gateway transmits the encapsulated combined units to the calculation server via a high-speed data interface. If the transmission link between the scheduling gateway and the calculation server is interrupted, the scheduling gateway stores the encapsulated combined units in local non-volatile storage and retransmits them in chronological order after the link is restored. If three consecutive transmission failures occur, a link alarm is triggered. After receiving the combined units, the calculation server extracts the relative displacement difference vector sequence within the second effective load of the consecutive combined units within a preset sliding time window. Before performing zero-crossing statistics, it first calculates the multidimensional relative displacement difference vector sequence within the sliding time window using principal component analysis (PCA), extracts the eigenvector direction with the largest variance as the main flutter direction axis, and then extracts the single-dimensional scalar component sequence of the multidimensional relative displacement difference vector on the main flutter direction axis.

[0093] The preset sliding time window length is set to 500ms, and the sliding step size is set to 100ms. That is, the effective data sequence within the sliding time window is updated every 100ms. This value matches the typical duration of human emotion-related micro-tremors in dance therapy scenarios, which can fully cover the complete process of a single effective micro-tremor, while avoiding the problem of confusion of different action stage characteristics caused by an excessively long time window.

[0094] The server calculates the number of times the single-dimensional scalar component sequence crosses a preset zero baseline within a preset sliding time window, obtaining the zero-crossing alternation count. The preset zero baseline is directly set as the real-time average of the single-dimensional scalar component sequence within the current sliding time window. The preset zero baseline corresponds to the dynamic baseline state of the relative displacement of the limb extremities. The zero-crossing alternation count is the total number of times the single-dimensional scalar component crosses the zero baseline from positive to negative and from negative to positive. Only when the absolute values ​​of both the positive and negative peak values ​​are greater than 1 / 2 of a preset second threshold when the single-dimensional scalar component crosses the zero baseline is it counted as a valid zero-crossing alternation. This can directly quantify the frequency and intensity of microtremors in the limb extremities.

[0095] In an optional embodiment, the extracted one-dimensional scalar component sequence is presented as an alternating amplitude signal characterizing physical micro-vibrations within a 500ms sliding time window. The system uses the calculated and extracted mean zero baseline as the central axis to perform a dynamic determination process for the scalar component sequence's penetration of the baseline. The system's configured peak / trough amplitude verification logic filters zero-crossing actions. Valid zero-crossing alternation points are characterized as legitimate penetration states where the absolute value amplitude meets the condition of being greater than half a preset second threshold. Penetrations that do not meet this amplitude constraint are treated as invalid noise and filtered out. The final accumulated number of valid zero-crossing alternations serves as the direct numerical basis for generating action mode labels.

[0096] The calculation server compares the statistically obtained number of zero-crossing alternations with a preset alternation threshold. If the number of zero-crossing alternations exceeds the preset threshold, the first payload within the corresponding combination unit is extracted, and the first payload is time-sequentially bound to the number of zero-crossing alternations to generate an action modality label. The system has a pre-set mapping table between the preset alternation threshold range and action modalities and therapeutic emotional states. Specifically, when the number of zero-crossing alternations is in the range [10, 20), corresponding to a high-frequency, intense micro-trembling action modality label, the system maps it to a specific therapeutic emotion of anxiety release or anger release; when the number of zero-crossing alternations is in the range [5, 10), corresponding to a soothing micro-trembling action modality label, the system maps it to a specific therapeutic emotion of joy or relaxation. Based on the specific therapeutic emotions output by this mapping, and combined with the scene characteristics of dance therapy, the calculation server integrates and outputs the emotion analysis results, quantifying the degree of emotional release and psychological state changes of the user during dance therapy.

[0097] This specific embodiment also provides a personalized dance therapy emotion analysis system based on real-time motion capture, used to implement the aforementioned personalized dance therapy emotion analysis method based on real-time motion capture. The system includes four core functional modules: a basic scheduling module, a state triggering and preemption module, a burst acquisition module, and a fusion analysis module.

[0098] The basic scheduling module is deployed within the scheduling gateway. It is used to send preset first thresholds, preset second thresholds, preset time tolerances, and a basic period to preset first and second motion capture nodes, and to broadcast beacon frames carrying a clock reference. The basic scheduling module is also used to synchronize the preset first and second motion capture nodes according to the clock reference, and to acquire the first data frame reported by the preset first motion capture node, carrying a timestamp and a control header, according to the basic period. The status indicator bit in the control header is set by the preset first motion capture node when the magnitude of its calculated integrated motion feature integral vector exceeds the preset first threshold.

[0099] The status triggering and preemption module is deployed within the scheduling gateway. It is used to parse the control header of the first data frame. If the status indicator bit is set, the scheduling based on the basic cycle is paused, and a preemption signaling containing the burst resource block is broadcast.

[0100] The burst acquisition module is deployed within the scheduling gateway. In response to preemption signaling, it acquires a second data frame carrying a timestamp within the burst resource block when the magnitude of the relative displacement difference vector cached locally by the preset second motion capture node is greater than a preset second threshold.

[0101] The fusion analysis module, deployed within the scheduling gateway and the calculation server, is used to merge the second data frame and the first data frame (whose timestamp difference is less than a preset time tolerance or triggers preemption) into a combined unit. The fusion analysis module also extracts the single-dimensional scalar component sequence of the relative displacement difference vector within the combined unit along the main flutter direction axis, and counts the number of zero-crossing alternations of the single-dimensional scalar component sequence within a preset sliding time window. Based on the number of zero-crossing alternations, it outputs action modality labels to generate sentiment analysis results.

[0102] The four modules interact and link status through internal interfaces. The basic scheduling module interacts with the status triggering and preemption module through the local bus, transmitting the parsed first data frame control header information to the status triggering and preemption module. The status triggering and preemption module links with the burst acquisition module through the timing control interface, synchronizing the preemption signaling configuration information to the burst acquisition module. The burst acquisition module transmits the received second data frame to the fusion analysis module. The fusion analysis module interacts with the basic scheduling module to exchange time base information, ensuring the consistency of the timing matching base.

[0103] This application's embodiments construct a complete technical closed loop, encompassing motion state triggering, preemptive resource scheduling, high-frequency data acquisition, precise temporal fusion, and emotional modality recognition. By using the integrated motion feature vector detection and cross-layer marking of the state indicator bits in the preset first motion capture node, a direct mapping from the macroscopic motion state of the human body to the medium access control layer signaling is achieved. Scheduling responses are triggered without waiting for nodes to initiate resource requests, eliminating the lag inherent in existing technologies that rely on static data thresholds for scheduling. Through preemptive resource reconfiguration by the scheduling gateway, a dedicated transmission channel is opened for micro-vibration data at the instant macroscopic movement occurs, enabling the preset second motion capture node to directly report data in high-frequency mode, avoiding buffer overflow and loss of high-frequency micro-vibration data during channel contention. Furthermore, the two-level thresholds of the preset second motion capture node... Filtering and conditional reporting ensure that channel resources are only occupied when macroscopic actions are triggered and microscopic tremors exceed a threshold, thus avoiding invalid channel occupation while ensuring effective data acquisition and improving the utilization efficiency of wireless resources. Through a time-domain fusion mechanism with hard-binding timestamps, macroscopic action data and microscopic tremor data from the same source time are fixedly bound, completely eliminating the problem of timestamp phase misalignment in multi-node data and providing a precisely aligned data source for action modality recognition. Through statistical analysis of the number of zero-crossing alternations, accurate identification of microscopic tremor features related to human emotions is achieved, significantly improving the accuracy and reliability of emotion analysis in dance therapy scenarios.

[0104] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A personalized dance therapy emotion analysis method based on real-time motion capture, characterized in that, include: Send the first threshold, the second threshold, the time tolerance, and the basic period to the preset first motion capture node and the preset second motion capture node, and broadcast a beacon frame carrying the clock reference; The first motion capture node and the second motion capture node are synchronized according to the clock reference, and the first data frame carrying the timestamp and control header reported by the first motion capture node is obtained according to the basic period. The status indication bit in the control header is set by the first motion capture node when the magnitude of its calculated comprehensive motion feature integral vector is greater than the first threshold. Parse the control header; if the status indicator bit is set, pause scheduling according to the basic cycle and broadcast a preemption signaling message containing a burst of resource blocks. In response to the preemption signaling, when the magnitude of the relative displacement difference vector cached locally by the second motion capture node is greater than the second threshold, a second data frame carrying a timestamp is reported within the burst resource block at a frequency higher than the basic periodicity. The second data frame and the first data frame whose timestamp difference is less than the time tolerance or which triggers preemption are merged into a combined unit; Extract the one-dimensional scalar component sequence of the relative displacement difference vector in the combined unit on the main tremor direction axis, and count the number of zero-crossing alternations of the one-dimensional scalar component sequence within the sliding time window. Output the action modality label based on the number of zero-crossing alternations to generate sentiment analysis results.

2. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 1, characterized in that, The status indicator bit in the control header is set by the first motion capture node when the magnitude of its calculated integrated motion feature vector is greater than the first threshold. This includes: the first motion capture node acquiring multi-axis acceleration and angular velocity within the base period; normalizing the multi-axis acceleration and angular velocity and integrating them along the time axis to generate the integrated motion feature vector; and when the magnitude of the integrated motion feature vector is greater than the first threshold for consecutive preset sampling periods, redefining the reserved bit in the control header as the status indicator bit and setting it.

3. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 1, characterized in that, The generation method of the relative displacement difference vector in the local cache of the second motion capture node includes: acquiring the original sequence of the second motion capture node according to a preset sampling rate and storing it in a preset cache time window; filtering the original sequence in the cache time window to remove components with frequencies lower than a preset cutoff frequency to extract jitter components; performing differential transformation on the jitter components to generate the relative displacement difference vector and updating it in the local cache so that its magnitude can be compared with the second threshold.

4. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 1, characterized in that, The broadcast includes preemption signaling for bursty resource blocks, including: The preemption signaling, which is configured with the burst resource block and its duration, is broadcast within the preset downlink control time slot. The method further includes: While broadcasting the preemption signaling, a resource protection timer with a preset timeout threshold is simultaneously started to monitor the occupancy status of the sudden resource block.

5. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 4, characterized in that, The monitoring of the occupancy status of the burst resource block includes: during the process of acquiring the second data frame, if the magnitude of the relative displacement differential vector cached locally by the second motion capture node is continuously not greater than the second threshold within a preset state maintenance period, and a short control frame containing a single-bit silent flag is sent within the burst resource block, then in response to the short control frame, a channel release confirmation instruction is issued to instruct the second motion capture node to fall back to the polling scheduling according to the basic period, and the resource protection timer is reset; if the resource protection timer reaches the preset timeout threshold and the short control frame is not received, then the burst resource block is forcibly reclaimed.

6. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 4, characterized in that, Also includes: Extract the uplink channel quality indication of the second capture node during the broadcast of the preemption signaling; If the uplink channel quality indicator is lower than the preset link failure threshold, the link adaptive mode is triggered, the allocated burst resource block is reconfigured as a sideline time slot, and a sideline relay scheduling signaling is sent to the first capture node and the second capture node. Based on the side-line relay scheduling signaling, the second motion capture node is instructed to transmit the second data frame to the first motion capture node via the side-line time slot; the dual-node data frame reported by the first motion capture node is obtained, and the first data frame and the transmitted second data frame are parsed from it for merging into the combined unit.

7. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 4, characterized in that, When there are multiple second capture nodes, the preemption signaling is configured with multiple allocation indexes of the burst resource blocks that correspond one-to-one with each of the second capture nodes and do not overlap with each other; Within the burst resource block, acquiring the second data frame carrying a timestamp reported by the second motion capture node at a frequency higher than the basic periodicity specifically involves: Within their respective burst resource blocks, each of the second motion capture nodes concurrently acquires the second data frames carrying the timestamps reported at a frequency higher than the base period.

8. The personalized dance therapy emotion analysis method based on real-time motion capture according to claim 1, characterized in that, The step of merging the second data frame with the first data frame whose timestamp difference is less than the time tolerance or which triggers preemption into a combination unit includes: Compare the timestamps of the first data frame and the second data frame; Extract the first payload of the first data frame whose timestamp difference is less than the time tolerance or triggers preemption, and the second payload of the second data frame; The first payload and the second payload are hard-bonded and encapsulated to generate the combined unit.

9. A personalized dance therapy emotion analysis method based on real-time motion capture according to claim 8, characterized in that, The process involves extracting the single-dimensional scalar component sequence of the relative displacement difference vector within the combined unit along the main tremor direction axis, counting the number of zero-crossing alternations of the single-dimensional scalar component sequence within a sliding time window, and outputting action modality labels based on the number of zero-crossing alternations to generate sentiment analysis results, including: Within the sliding time window, the relative displacement difference vector sequence within the second effective load of the consecutive combined units is extracted, and the one-dimensional scalar component sequence of the relative displacement difference vector sequence on the main flutter direction axis is extracted; The number of zero-crossing alternations is obtained by counting the number of times the single-dimensional scalar component sequence crosses a preset zero baseline; If the number of zero-crossing alternations is greater than a preset alternation threshold, the first payload in the corresponding combination unit is extracted, and the first payload is bound to the number of zero-crossing alternations to generate the action modality label; based on the preset alternation threshold range where the number of zero-crossing alternations is located, the action modality label is mapped to the corresponding specific healing emotion and written into the emotion analysis result.

10. A personalized dance therapy emotion analysis system based on real-time motion capture, used to implement the personalized dance therapy emotion analysis method based on real-time motion capture as described in any one of claims 1-9, characterized in that, The system includes: The basic scheduling module is used to send a first threshold, a second threshold, a time tolerance, and a basic period to a preset first motion capture node and a preset second motion capture node, and broadcast a beacon frame carrying a clock reference; synchronize the first motion capture node and the second motion capture node according to the clock reference, and acquire the first data frame carrying a timestamp and a control header reported by the first motion capture node according to the basic period. The status indication bit in the control header is set by the first motion capture node when the magnitude of its calculated integrated motion feature integral vector is greater than the first threshold. The status triggering and preemption module is used to parse the control header. If the status indicator bit is set, the scheduling according to the basic cycle is paused, and a preemption signaling containing a burst resource block is broadcast. A burst acquisition module is used to, in response to the preemption signaling, acquire a second data frame carrying a timestamp that is reported within the burst resource block at a frequency higher than the basic period when the magnitude of the relative displacement difference vector cached locally by the second motion capture node is greater than the second threshold. The fusion analysis module is used to merge the second data frame and the first data frame whose timestamp difference is less than the time tolerance or triggers preemption into a combined unit; extract the single-dimensional scalar component sequence of the relative displacement difference vector in the combined unit on the main tremor direction axis, and count the number of zero-crossing alternations of the single-dimensional scalar component sequence within the sliding time window, and output the action modality label based on the number of zero-crossing alternations to generate sentiment analysis results.

Images