A three-dimensional immersive anti-sickness method and system

Abstract

The application provides a three-dimensional immersive anti-dizziness method and system, the method comprising: generating an individual dizziness sensitivity coefficient and a resting heart rate variability baseline based on an individualized baseline calibration procedure; generating a dynamic physiological risk level and an environmental overload vector evaluation in combination with a nonlinear regression model; generating a hierarchical dimension reduction execution strategy using a cross-dimension sensory breathing control algorithm module; generating visual impact compensation using a motion vector conservation algorithm; and finally gradually restoring the picture size and parallax after the physiological indicators recover to the safety baseline in combination with the hierarchical dimension reduction execution strategy and the visual impact compensation. The application evaluates the dynamic risk through a standardized test and a nonlinear regression model, generates a hierarchical dimension reduction strategy based on the results, compensates for the visual impact using a motion vector conservation algorithm; and matches the recovery speed with the individual tolerance through a centrifugal dynamic recovery model to avoid secondary dizziness. The method realizes full-process elastic adjustment, suppresses motion sickness, and retains the three-dimensional scene realism and interaction accuracy.

Description

Technical Field

[0001] This invention relates to the field of anti-vertigo technology, and in particular to a three-dimensional immersive anti-vertigo method and system. Background Technology

[0002] In virtual reality (VR), remote control (drone piloting, robot teleoperation), and highly dynamic interactive scenarios, the pursuit of ultimate 3D immersion is the mainstream development direction. However, under extreme physical environments (such as severe turbulence, ultra-high G-forces, and long-range low-frequency oscillations), existing technologies have significant drawbacks: the intense motion in the visual image and the stationary state perceived by the vestibular system create a severe perceptual conflict, triggering severe motion sickness—a physiological mechanism similar to the "stationary elevator illusion," leading to dizziness, shock, and even cumulative nervous fatigue in the operator. Current solutions often adopt a "one-size-fits-all" strategy: either directly reducing the complexity of the image or forcibly exiting the immersive state, which is equivalent to stopping surgery directly without individualized drug administration during anesthesia. This lack of step-by-step adjustment capabilities based on real-time physiological feedback leads to an imbalance between the immersive experience and the anti-vertigo effect. Specifically, this manifests in two ways: First, when users have not yet experienced obvious dizziness, excessive simplification of the screen can significantly reduce the realism of the 3D scene and the accuracy of interaction. For example, in a drone inspection scenario, reducing the field of view may cause the operator to miss key equipment details. Second, when users have experienced moderate dizziness, a single dimensionality reduction strategy (such as simply switching to a 2D display) cannot effectively alleviate the vestibular-visual conflict. Instead, the sudden loss of depth information can trigger a secondary dizziness shock. Summary of the Invention

[0003] This invention aims to at least solve the technical problems existing in the prior art, and in particular, it innovatively proposes a three-dimensional immersive anti-vertigo method and system.

[0004] To achieve the above-mentioned objectives of the present invention, the present invention provides a three-dimensional immersive anti-vertigo method, the method comprising: S1. A biofeedback-based individualized baseline calibration procedure that uses a standardized vestibular-visual coupling test to generate an individual’s vertigo sensitivity coefficient and resting heart rate variability baseline. S2. Based on the individual's vertigo sensitivity coefficient and resting heart rate variability baseline, a dynamic physiological risk level and environmental overload vector assessment are generated using a nonlinear regression model and real-time monitored physiological signals and environmental physical vectors. S3. Based on the dynamic physiological risk level and environmental overload vector assessment, a graded dimensionality reduction execution strategy is generated using the cross-dimensional sensory breathing control algorithm module. S4. Based on the hierarchical dimensionality reduction execution strategy, the motion vector conservation algorithm is used to generate a synchronous rendering mechanism and vector continuation effect at the moment of dimension switching. By extracting the motion vector of the previous frame and superimposing radial blur, the visual impact caused by the disappearance of the sense of depth is compensated, and the final visual impact compensation is generated. S5. Based on the hierarchical dimensionality reduction execution strategy and the final visual impact compensation, the centrifugal dynamic recovery model is used to gradually restore the screen size and parallax according to the exponential recovery function after the physiological indicators are restored to the safe baseline.

[0005] In another aspect, the present invention also provides a three-dimensional immersive anti-vertigo system, characterized in that the system comprises: processor; Memory used to store processor-executable instructions; The processor is configured to implement the three-dimensional immersive anti-vertigo method according to any one of claims 1 to 7 when executing the executable instructions.

[0006] The beneficial effects of this invention are as follows: This invention effectively solves the core problem of the imbalance between immersive experience and anti-vertigo effect in existing technologies by sequentially implementing a full-link, step-by-step, elastic interactive protocol consisting of individualized baseline calibration, dynamic risk assessment, hierarchical dimensionality reduction execution, visual compensation, and progressive recovery. Specifically, the method generates individual vertigo sensitivity coefficients and heart rate variability baselines through standardized vestibular-visual coupling tests, and combines a nonlinear regression model to fuse physiological signals and environmental physical vectors in real time, achieving accurate assessment of dynamic physiological risk levels and environmental overload vectors. Based on the assessment results, the cross-dimensional sensory breathing control algorithm module generates a hierarchical dimensionality reduction execution strategy that includes dimensionality reduction ratio, field of view compression, and color saturation adjustment. While avoiding "one-size-fits-all" simplification of the screen, it extracts the motion vector of the previous frame at the moment of dimensional switching and superimposes radial blur through a motion vector conservation algorithm to compensate for the visual impact caused by the loss of depth perception and maintain the recognizability of key scene details. Finally, after the physiological indicators are restored to a safe baseline, the eccentric dynamic recovery model gradually restores the screen size and parallax according to an exponential recovery function, ensuring that the recovery speed matches the individual's tolerance and avoiding secondary vertigo. This method achieves flexible adjustment throughout the entire process, from physical perception conflict defense to immersion restoration. In extreme physical environments, it can effectively suppress motion sickness while maximizing the preservation of the realism and interaction accuracy of the 3D scene, breaking through the bottleneck of existing technologies where immersive experience and anti-dizziness effects cannot be achieved simultaneously.

[0007] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0008] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart of a three-dimensional immersive anti-vertigo method according to the present invention. Detailed Implementation

[0009] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0010] Example 1 like Figure 1 As shown, a three-dimensional immersive anti-vertigo method is provided, the method comprising: S1. A biofeedback-based individualized baseline calibration procedure that uses a standardized vestibular-visual coupling test to generate an individual’s vertigo sensitivity coefficient and resting heart rate variability baseline. The standardized vestibular-visual coupling test described in step S1 adopts a rotating chair-eye tracking joint paradigm. Sine wave stimulation is performed within the angular acceleration range of 0.1-1.5Hz, and the slow phase velocity of nystagmus, subjective vertigo score, and skin conductance are recorded simultaneously. The vertigo sensitivity coefficient is determined by calculating the Pearson correlation coefficient between eye movement gain and vertigo score. At the same time, electrocardiogram data under n minutes of resting state are collected, and the high-frequency power spectral density is extracted by wavelet transform as the baseline of resting heart rate variability.

[0011] In step S1, it is necessary to explain in detail that n is set to 5-10 minutes to ensure the stability of heart rate variability data; eye movement gain is the ratio of slow-phase nystagmus velocity to rotational stimulus angular velocity, calculated in real time by an eye-tracking system; subjective vertigo scoring uses a 10-point visual analog scale, with subjects providing real-time feedback on their vertigo sensations during stimulation; electrodermal activity is collected using Ag / AgCl electrodes placed on the fingertips of the index and middle fingers, with a sampling frequency of no less than 256Hz to capture subtle changes in autonomic nervous system responses; wavelet transform uses Morlet wavelets for time-frequency analysis to extract the power spectral density of the 0.15-0.4Hz high-frequency band (HF), which corresponds to parasympathetic nervous system activity and can effectively reflect an individual's autonomic nervous system regulation ability; in the calculation of the vertigo sensitivity coefficient, the larger the absolute value of the Pearson correlation coefficient, the higher the degree of vestibular-visual coupling of the individual; the vertigo sensitivity coefficient ranges from 0 to 1, and the closer the coefficient is to 1, the stronger the individual's vertigo sensitivity to three-dimensional scene motion.

[0012] S2. Based on the individual's vertigo sensitivity coefficient and resting heart rate variability baseline, a dynamic physiological risk level and environmental overload vector assessment are generated using a nonlinear regression model and real-time monitored physiological signals and environmental physical vectors. S3. Based on the dynamic physiological risk level and environmental overload vector assessment, a graded dimensionality reduction execution strategy is generated using the cross-dimensional sensory breathing control algorithm module. S4. Based on the hierarchical dimensionality reduction execution strategy, the motion vector conservation algorithm is used to generate a synchronous rendering mechanism and vector continuation effect at the moment of dimension switching. By extracting the motion vector of the previous frame and superimposing radial blur, the visual impact caused by the disappearance of the sense of depth is compensated, and the final visual impact compensation is generated. S5. Based on the hierarchical dimensionality reduction execution strategy and the final visual impact compensation, the centrifugal dynamic recovery model is used to gradually restore the screen size and parallax according to the exponential recovery function after the physiological indicators are restored to the safe baseline.

[0013] The expression for the exponential recovery function is: in, Indicates time The degree of recovery at that time This indicates the maximum degree of restoration, usually set to 1 (meaning complete restoration to a fully immersive 3D state). The recovery slope is inversely proportional to the individual's cumulative vertigo value (i.e., the longer the vertigo lasts, the more severe the deterioration of physiological indicators). The smaller the value, the slower the recovery speed.

[0014] In step S5, the centrifugal dynamic recovery model needs to be explained in detail. This model achieves dynamic control of the recovery process through a real-time closed-loop feedback mechanism: First, it continuously monitors the user's heart rate variability and skin conductance. When both physiological indicators are stable within ±5% of the resting baseline for 5 consecutive seconds, the recovery process is initiated. Second, using centrifugal recovery logic, it prioritizes the recovery of the 3D parallax and detail accuracy of the central area of ​​the image (such as key components of equipment in drone inspections or operational targets in virtual surgery), while the edge areas are temporarily maintained in a low-load state after dimensionality reduction to avoid global recovery. The visual load increases dramatically during recovery; furthermore, the recovery slope is dynamically adjusted every 2 seconds based on real-time physiological data: if there are no abnormal fluctuations in physiological indicators, the recovery slope is gradually increased by 5%; if the heart rate variability rate decreases by more than 10% or the peak skin conductance increases, recovery is paused for 1 second and the recovery slope is reduced by 10% to ensure that the recovery speed matches individual tolerance; finally, after the central area is fully recovered, the recovery range is expanded outward from the center with increasing radius, and the expansion speed is proportional to the current recovery slope, until the entire screen is restored to a fully immersive 3D state. During the recovery process, the cross-dimensional sensory breathing control algorithm module synchronously outputs breathing guidance signals that match the recovery rhythm (such as slow abdominal breathing prompts of 6 breaths / minute in the early stage of recovery, gradually increasing to 12 breaths / minute as recovery progresses), further strengthening the coordinated regulation of the vestibular-visual system and reducing the risk of secondary dizziness.

[0015] As an optional embodiment of the present invention, optionally, in step S2, based on the individual's vertigo sensitivity coefficient and resting heart rate variability baseline, a dynamic physiological risk level and environmental overload vector assessment are generated using a nonlinear regression model and real-time monitored physiological signals and environmental physical vectors, including: S201. Based on the individual's vertigo sensitivity coefficient and resting heart rate variability baseline, a nonlinear regression model is used to generate individualized basic parameters for physiological risk assessment. In step S201, it is necessary to explain in detail that, The nonlinear regression model uses a bivariate input LSTM network structure. The input layer contains the vertigo sensitivity coefficient (S) and the resting heart rate variability baseline (HRVbase). Feature mapping is performed through three hidden layers (with 64, 32, and 16 neurons per layer, respectively). The output layer generates a set of basic parameters including risk threshold offset (ΔT), physiological signal weight coefficients (WECG, WGSR), and environmental vector sensitivity (Kenv). Among them, the risk threshold offset ΔT is positively correlated with the dizziness sensitivity coefficient S (ΔT=0.3×S+0.1), and is used to dynamically adjust the threshold for subsequent risk level determination; the physiological signal weight coefficients WECG and WGSR are inversely distributed according to the size of HRVbase. When HRVbase is less than 50ms², WECG (0.6) is higher than WGSR (0.4), giving priority to heart rate variability changes. When HRVbase is greater than 100ms², WGSR (0.6) is higher than WECG (0.4), focusing on sympathetic nerve activation reflected by skin conductance; the environmental vector sensitivity Kenv is inversely proportional to the product of S and HRVbase (Kenv=1 / (S×HRVbase)×0.5), indicating that highly sensitive individuals have a lower tolerance threshold for changes in environmental physical vectors. The model was trained using the Adam optimizer with a learning rate of 0.001. The training set included vestibular function test data and physiological signal samples from 200 subjects of different ages (18-65 years old). Five-fold cross-validation was used to ensure that the model could maintain its evaluation accuracy even with large individual differences. The mean absolute error on the validation set was controlled within 0.05.

[0016] S202. Based on the individualized physiological risk assessment parameters, a real-time individual physiological state assessment is generated using real-time monitored physiological signals. In step S202, it is necessary to explain in detail that the physiological signals monitored in real time include dynamic heart rate variability (HRVreal-time) and skin electrical activity (GSRreal-time). The dynamic heart rate variability is obtained by extracting the 0.15-0.4Hz high-frequency power spectral density from real-time acquired electrocardiogram data using the same wavelet transform method as in step S1. Skin electrical activity is continuously acquired using Ag / AgCl electrodes, and after noise removal via a 5Hz low-pass filter, the mean skin conductance level (SCL) and peak frequency of skin conductance response (SCR) are calculated per minute. Individual physiological state assessment is achieved by calculating the physiological deviation index (PSI), with the formula: PSI = WECG × |HRVreal-time - HRVbase| / HRVbase + WGSR × (GSRreal-time - GSRrest) / GSRrest, where GSRrest is the baseline value of resting-state skin electrical activity acquired synchronously in step S1. When the PSI value is greater than 0, it indicates that the physiological indicators deviate from the baseline, and the larger the value, the more significant the deviation, reflecting the degree of disorder in the individual's current autonomic nervous system regulation.

[0017] S203. Based on the real-time individual physiological state assessment, an environmental overload vector assessment is generated using environmental physical vectors monitored by airborne IMU data. In step S203, it is necessary to explain in detail that the environmental physical vector includes linear acceleration (ax, ay, az), angular acceleration (αx, αy, αz), and field of view angle change rate (dFOV / dt) in three-dimensional space, which are acquired in real time by the airborne IMU (inertial measurement unit) at a sampling frequency of 100Hz. The environmental overload vector assessment employs a vector synthesis algorithm. First, all physical parameters are standardized to the [-1, 1] range (linear acceleration at full scale of 3g, angular acceleration at full scale of 200° / s², and field-of-view angle change rate at full scale of 60° / s). Then, based on the environmental vector sensitivity Kenv generated in step S201, the weighted environmental overload index (EOI) is calculated: EOI = Kenv × (0.4 × |ax| + 0.4 × |αy| + 0.2 × |dFOV / dt|). The weight allocation is based on the main triggers of motion sickness in the 3D scene, determining that linear acceleration (especially vertical ay) and yaw angle acceleration (αy) have the strongest stimulation to the vestibular system, thus assigning them higher weights. The field-of-view angle change rate, as a supplementary indicator of visual flow stimulation, is assigned a lower weight. A higher EOI value indicates a stronger sensory stimulation of the user by the current environmental physical motion; a value exceeding 0.8 is considered an environmental overload state.

[0018] S204. Based on the real-time assessment of individual physiological state and the real-time assessment of environmental overload vector, a dynamic comprehensive assessment of physiological risk level and environmental overload vector is generated using a preset hierarchical decision-making logic.

[0019] In step S204, it is necessary to explain in detail that the graded decision-making logic adopts a two-factor matrix evaluation method, using the Physiological Deviation Index (PSI) and the Environmental Overload Index (EOI) as input variables to construct a 3×3 risk level matrix. The PSI is divided into three intervals: 0 ≤ PSI < 0.3 for mild deviation, 0.3 ≤ PSI < 0.6 for moderate deviation, and PSI ≥ 0.6 for severe deviation. Similarly, the EOI is divided into three intervals: 0 ≤ EOI < 0.4 for low overload, 0.4 ≤ EOI < 0.8 for moderate overload, and EOI ≥ 0.8 for high overload. The matrix elements correspond to physiological risk levels, divided into 5 levels: Level 0 (safe state) is defined as a slight deviation in PSI and a low overload in EOI; Level 1 (warning state) is defined as a slight deviation in PSI and a moderate overload in EOI, or a moderate deviation in PSI and a low overload in EOI; Level 2 (mild intervention state) is defined as a slight deviation in PSI and a high overload in EOI, or a moderate deviation in PSI and a moderate overload in EOI, or a severe deviation in PSI and a low overload in EOI; Level 3 (moderate intervention state) is defined as a moderate deviation in PSI and a high overload in EOI, or a severe deviation in PSI and a moderate overload in EOI; and Level 4 (emergency intervention state) is defined as a severe deviation in PSI and a high overload in EOI. The comprehensive assessment results are output in real time to the graded dimensionality reduction execution module as the basis for initiating the corresponding anti-vertigo strategy. Meanwhile, to avoid frequent fluctuations in assessment results, a 0.5-second hysteresis window is set, meaning that the risk level difference between two adjacent assessment results must last for more than 0.5 seconds before the level switch is executed, ensuring the stability of the system response.

[0020] As an optional embodiment of the present invention, optionally, in step S3, based on the dynamic physiological risk level and environmental overload vector assessment, a graded dimensionality reduction execution strategy is generated using the cross-dimensional sensory breathing control algorithm module, including: S301. Based on the dynamic physiological risk level, a preliminary dimensionality reduction response level determination is generated using the hierarchical dimensionality reduction execution strategy framework in the cross-dimensional sensory breathing control algorithm module. In step S301, it is necessary to explain in detail that the hierarchical dimensionality reduction execution strategy framework corresponds one-to-one with the physiological risk levels output in step S204, setting a total of 5 response levels, each matching risk levels 0 to 4. When the risk level is 0 (safe state), the response level is determined to be level 0, maintaining the full 3D immersion mode, and the cross-dimensional sensory breathing control algorithm module only outputs regular breathing guidance signals (a natural breathing rhythm of 12-15 breaths / minute); when the risk level is 1 (warning state), the response level is determined to be level 1 (mild dimensionality reduction), and preventive adjustment is initiated; when the risk level is 2 (mild intervention state), the response level is determined to be level 2 (moderate dimensionality reduction), and basic dimensionality reduction measures are implemented; when the risk level is 3 (moderate intervention state), the response level is determined to be level 3 (deep dimensionality reduction), and an enhanced dimensionality reduction scheme is executed; when the risk level is 4 (emergency intervention state), the response level is determined to be level 4 (extreme dimensionality reduction), and the highest level of dimensionality reduction protection is initiated. The framework achieves rapid determination through a pre-set response level mapping table, which is stored in the local configuration file of the algorithm module. It can be fine-tuned through the parameter interface according to the actual application scenario (such as virtual training, remote control, entertainment experience). For example, in high-precision surgical scenarios, the response level trigger threshold corresponding to each risk level can be appropriately increased to avoid frequent dimensionality reduction affecting the operation accuracy.

[0021] S302. Based on the preliminary determination of the reduced-dimensional response level and the environmental overload vector, the specific parameters of the motion compensation mechanism are generated using the motion vector conservation algorithm. In step S302, the motion vector conservation algorithm needs to be explained in detail. This algorithm takes the real-time motion vector field, the dimensionality reduction response level, and the motion components in the environmental overload vector as input. Through multi-dimensional vector adjustment and conservation constraints, it generates a motion compensation parameter set adapted to the dimensionality reduction scenario. The specific process is as follows: First, spatial pyramid sampling is performed on the motion vector field of the current frame, retaining high-resolution vector information of the central region of the image (occupying 60% of the total image) and moving subjects (such as virtual surgical instruments or drones), while using low-resolution sampling for edge regions to reduce computational load. Second, the depth vector retention ratio R is determined according to the response level: Response level 1 (slight dimensionality reduction) R=0.9, level 2 (moderate dimensionality reduction) R=0.6, level 3 (deep dimensionality reduction) R=0.3, level 4 (extreme dimensionality reduction) R=0. The original 3D motion vector (Vx,Vy,Vz) is then... The vector field is transformed into a dimension-reduced vector (Vx', Vy', Vz'), where Vz' = Vz × R. Vx' and Vy' are rotated and corrected by angular acceleration αy (formula: Vx' = Vx × cos(αy × Δt) - Vy × sin(αy × Δt), Vy' = Vx × sin(αy × Δt) + Vy × cos(αy × Δt), where Δt is the frame interval 1 / 60s). Furthermore, the vector smoothing coefficient S is adjusted by combining the environmental overload index EOI: S = 1 - 0.3 × EOI (EOI ∈ [0,1]). Gaussian filtering is applied to the adjusted vector field. The higher the EOI, the stronger the filtering, reducing visual jitter caused by rapid motion. Finally, the dimension-reduced motion vector field, depth retention ratio, and smoothing coefficient are output as the core parameters of the synchronous rendering mechanism to ensure the consistency of motion continuity and vestibular perception during the dimension reduction process. The algorithm is implemented using GPU parallel computing, with a single-frame processing time of ≤1ms, meeting the requirements of real-time rendering. Furthermore, the parameter update frequency is synchronized with the hysteresis window in step S204, avoiding frequent fluctuations that could affect visual stability. Simultaneously, the rhythm of motion compensation parameter changes is linked to the breathing guidance signal from the cross-dimensional sensory breathing control module: when the response level increases, the vector adjustment speed is inversely proportional to the breathing rate (e.g., the adjustment speed decreases by 20% when the breathing rate is 6 breaths / minute), strengthening the coordinated regulation of physiological and visual functions.

[0022] S303. Based on the specific parameters of the motion compensation mechanism and the preliminary determination of the dimensionality reduction response level, the breathing guidance subunit in the cross-dimensional sensory breathing control algorithm module is used to generate a breathing rhythm guidance signal that matches the dimensionality reduction level. In step S303, regarding the breathing guidance subunit in the cross-dimensional sensory breathing control algorithm module, it is necessary to explain in detail that this subunit adopts a three-layer architecture of "rhythm generation - multimodal output - feedback adjustment" to achieve dynamic coordination between breathing guidance and dimensionality reduction strategies. First, the rhythm generation layer determined the basic respiratory parameters based on the response level: Response level 0 corresponds to a natural breathing rhythm (12-15 breaths / min, inspiratory-expiratory time ratio 1:1.2); Level 1 (early warning) uses preventative slow abdominal breathing (8 breaths / min, 1:2); Level 2 (mild intervention) uses deep relaxation breathing (6 breaths / min, 1:2.5); Level 3 (moderate intervention) uses vagal nerve activation breathing (5 breaths / min, 1:3); and Level 4 (emergency intervention) uses extreme sedation breathing (4 breaths / min, 1:3.5). The parameter settings referenced the clinical research results published in the *Journal of Vestibular Medicine* in 2022, which showed that slow abdominal breathing can reduce vestibular sensitivity by 32% ± 5%.

[0023] Secondly, the multimodal output layer provides simultaneous guidance through three channels: visual, auditory, and tactile. Visually, a dynamic breathing waveform is displayed in the center of the HUD (green corresponds to a safe state, yellow to a warning, and red to intervention). Auditorily, stereo prompts are output (high-frequency short tones during inhalation and low-frequency long tones during exhalation, with volume positively correlated with the Environmental Overload Index (EOI). Tactilely, vibration feedback is provided through a wrist-worn device (weak vibration intensity during inhalation and strong vibration intensity during exhalation, with frequency matching the breathing rhythm). This three-channel synergy can significantly improve user breathing compliance.

[0024] Finally, the feedback regulation layer optimizes guidance parameters in real time based on physiological signals: if the dynamic heart rate variability (HRV real-time) is detected to decrease by more than 5%, the respiratory rate is automatically reduced by 1 breath / minute, and the inspiratory-expiratory ratio is increased by 0.2; if the peak value of ground skin activity (GSR real-time) increases by more than 10%, the intensity of tactile feedback is increased by 20%. At the same time, the sub-unit has a built-in adaptive learning module that records the user's breathing matching degree at different response levels (such as the synchronization rate between breathing rhythm and guidance signal), and updates the rhythm parameters through the Q-learning algorithm. For example, for users with a matching degree of less than 70%, the prompt duration of the exhalation phase is extended by 15%.

[0025] The computational delay of this sub-unit is ≤0.2 seconds, and the synchronization error with the dimensionality reduction execution module is controlled within ±0.1 seconds, ensuring the synergy between breathing guidance and visual dimensionality reduction, and further reducing the probability of dizziness.

[0026] S304. Based on the breathing rhythm guidance signal, the specific parameters of the motion compensation mechanism, and the preliminary dimensionality reduction response level determination, a graded dimensionality reduction execution strategy is generated, which includes the image dimension reduction ratio, the degree of field of view compression, and the color saturation adjustment value.

[0027] In step S304, it is necessary to explain in detail the specific parameter configuration of the hierarchical dimensionality reduction execution strategy as follows: the screen dimension reduction ratio is directly related to the response level. At response level 1 (slight dimensionality reduction), 90% of the three-dimensional information is retained, and only the 10% area at the edge of the screen is subjected to depth blurring. At response level 2 (moderate dimensionality reduction), the three-dimensional information is reduced to 60%, the central 40% area maintains the original depth, and the outer area is converted to pseudo-3D (only 50% of the depth value in the Z-axis direction is retained). At response level 3 (depth dimensionality reduction), 30% of the three-dimensional information is retained, only the true depth of the 20% core operation area in the center of the screen is retained, and the remaining area is converted to 2D planar display. At response level 4 (extreme dimensionality reduction), the entire system is switched to 2D mode, and all depth information is cleared to zero. The field of view (FOV) compression is dynamically adjusted based on the Environmental Overload Index (EOI). The base FOV is set to 90°. Compression begins when EOI ≥ 0.4, using the formula FOV = 90° × (1 - 0.3 × EOI). For example, when EOI = 0.8, the FOV is compressed to 90° × (1 - 0.3 × 0.8) = 68.4°, with a minimum compression of 50° (corresponding to EOI = 1.0). This reduces visual stimulation by narrowing the field of view. Color saturation adjustment uses a graded attenuation mechanism. At response level 0, 100% original saturation is maintained; level 1 reduces it by 10% and slightly increases the proportion of the green channel (enhancing the sense of calm); level 2 reduces it by 20% and adjusts the color temperature from 6500K (cool tone) to 5500K (neutral tone); level 3 reduces it by 35%, lowers the color temperature to 4500K (warm tone), and reduces the intensity of the blue channel (blue easily causes visual fatigue); level 4 reduces saturation by 50% and switches to a black-and-white display mode to minimize visual sensory load. All parameter adjustments employ a smooth transition algorithm, limiting the rate of change for dimensionality reduction ratio, field of view, and color parameters to ≤15% per second to prevent secondary dizziness caused by sudden visual changes. Furthermore, the adjustment process is synchronized with the exhalation phase output by the breathing guidance subunit (the parameter change rate decreases by 30% during exhalation), utilizing the physiological relaxation window of the respiratory cycle to enhance user adaptability. After the strategy is generated, parameters are pushed in real-time through the VR device's rendering engine interface, enabling dynamic adjustment of the rendering effect. The single-frame parameter update time is ≤2ms, ensuring synchronized execution with the motion compensation mechanism.

[0028] As an optional embodiment of the present invention, the cross-dimensional sensory breathing control algorithm module may include a breathing rhythm generation unit, a breathing guidance subunit, a multimodal feedback fusion unit, and a dimensionality reduction parameter mapping unit.

[0029] Regarding the cross-dimensional sensory breathing control algorithm module, it's important to note that this module employs a modular design, with each unit working collaboratively to dynamically generate cross-dimensional sensory adjustment and dimensionality reduction strategies. The breathing rhythm generation unit is the core control center. Based on the response level determination result in step S301, it calls upon the built-in physiological rhythm database (containing respiratory baseline parameters for different age and gender groups) and generates an initial breathing rhythm signal using an existing PID control algorithm. The signal sampling frequency is 100Hz to ensure rhythm smoothness. The breathing guidance subunit is responsible for converting the abstract rhythm signal into multimodal guidance commands, as detailed in step S303. It interacts with the breathing rhythm generation unit via an internal bus, with an interaction delay ≤0.1 seconds. The multimodal feedback fusion unit receives real-time data (such as HRV, GSR, and respiratory rate) from the physiological state monitoring module and user response data to breathing guidance (determining visual guidance fixation duration through eye tracking and collecting respiratory voiceprints through a microphone). It uses a weighted average method to fuse the multi-source feedback information, with weighting coefficients dynamically adjusted based on the signal-to-noise ratio (e.g., 0.6 weight for HRV signals with a signal-to-noise ratio ≥30dB, 0.3 weight for GSR signals, and 0.1 weight for user response data). The fusion result is fed back to the respiratory rhythm generation unit in real time, forming a closed-loop regulation. The dimensionality reduction parameter mapping unit establishes a mapping relationship between response levels, motion compensation parameters, and image dimensionality reduction parameters. It has a built-in nonlinear mapping function library that can call different mapping curves according to different application scenarios (such as military simulation and medical training). For example, in scenarios requiring high spatial awareness, the decay curve of the depth retention ratio R is adjusted to an exponential type (R=e(-k×level), where k is the scene coefficient), while in entertainment scenarios, linear decay (R=1-0.25×level) is used. This module also integrates a fault diagnosis subunit, which monitors the operating status of each unit in real time. When an abnormal breathing guidance signal is detected (such as the frequency exceeding the preset range of 4-20 breaths / minute) or a parameter calculation error is detected, it automatically switches to a backup algorithm (such as an interpolation algorithm based on historical optimal parameters) and triggers an audible and visual alarm to ensure the reliability of the system.

[0030] As an optional embodiment of the present invention, optionally, in step S4, based on the hierarchical dimensionality reduction execution strategy, a motion vector conservation algorithm is used to generate a synchronous rendering mechanism and vector continuation effect at the moment of dimension switching. By extracting the motion vector of the previous frame and superimposing radial blur, the visual impact caused by the disappearance of the sense of depth is compensated, and the final visual impact compensation is generated, including: S401. Based on the hierarchical dimensionality reduction execution strategy, the starting point of the vertical synchronization signal is used to trigger dimensional switching and generate a synchronous rendering mechanism. In step S401, it is necessary to explain in detail that the synchronous rendering mechanism adopts a three-level synchronous architecture of "vertical synchronization trigger - dual buffer alternation - parameter preloading" to ensure that the dimension switching process is precisely aligned with the refresh cycle of the display device. First, the system monitors the vertical synchronization signal VSync of the VR device in real time and strictly binds the execution start point of the dimension switching command to the rising edge of the VSync signal to avoid screen tearing. For devices with a refresh rate of 90Hz, the VSync signal period is approximately 11.1ms. The system completes the final confirmation of the dimensionality reduction parameters 2ms before triggering, ensuring that the time deviation between parameter loading and the synchronization signal is ≤0.5ms. Second, a dual buffer rendering mechanism is adopted. The main buffer is responsible for the normal rendering of the current frame, while the secondary buffer preloads the first frame of the image data after dimensionality reduction. When a switching command is received, a seamless transition of the image is achieved through hardware-level buffer switching (switching latency ≤0.1ms), avoiding rendering interruption in the traditional single buffer mode. Finally, after generating the hierarchical dimensionality reduction strategy in step S304, the parameter preloading module immediately writes core parameters such as the image dimension reduction ratio and field of view compression value into the GPU's dedicated registers. These registers employ a dual-channel parallel read / write design with a data transfer bandwidth ≥20GB / s, ensuring parameter configuration is completed before VSync triggers and preventing frame loss due to parameter loading delays. The triggering of the synchronous rendering mechanism is also deeply tied to the exhalation phase of the breathing guidance subunit: dimensionality switching is only allowed when the breathing guidance signal detects that the user is at the end of exhalation (expiration phase percentage ≥80%). This leverages the physiological characteristic that the human vestibular system has the lowest sensitivity during exhalation to further reduce the perception of dizziness during the switching process.

[0031] S402. Based on the synchronous rendering mechanism, the reference parameters for vector continuation are generated using the motion vector data of the first 3 frames. In step S402, it is necessary to explain in detail that the generation of vector continuation reference parameters adopts a three-order processing flow of "historical data weighted fusion - motion trend prediction - outlier filtering". First, the system extracts the dimensionality-reduced motion vector field data of the first 3 frames (denoted as t-3, t-2, t-1) from the output buffer of the motion vector conservation algorithm, including the two-dimensional motion vector (Vx', Vy') of each pixel and the depth preservation ratio R. The three frames of data are weighted and fused over time, with the weight coefficients decreasing over time: frame t-1 is assigned a weight of 0.5, frame t-2 a weight of 0.3, and frame t-3 a weight of 0.2. The average motion vector is calculated using the formulas Vxavg=0.5×Vx'(t-1)+0.3×Vx'(t-2)+0.2×Vx'(t-3) and Vyavg=0.5×Vy'(t-1)+0.3×Vy'(t-2)+0.2×Vy'(t-3) to enhance data stability. Secondly, based on the fused average vector, a linear regression algorithm is used to predict the motion trend of the current frame (frame t): Let the timestamps of the previous three frames be t3, t2, and t1 (in ms). The linear equations Vx(t) = a1×t + b1 and Vy(t) = a2×t + b2 are fitted using the least squares method, where a1 and a2 are motion acceleration coefficients, and b1 and b2 are initial vector values, thus obtaining the predicted motion vector (Vxpred, Vypred). Finally, an outlier filtering mechanism is introduced, calculating the deviation rate between the predicted vector and the average vector of the previous three frames: Dev = |Vxpred - Vxavg| / |Vxavg| (if Vxavg is 0, the deviation in the Vy direction is used). When Dev > 30%, it is determined to be a sudden motion change, and the predicted vector is automatically replaced with the vector of the previous frame (Vx'(t-1), Vy'(t-1)) to prevent distortion of the continuity parameters caused by violent motion. The baseline parameters also include a depth retention factor Dr, which is equal to the sliding average of the depth retention ratios of the first three frames (Dr=(R(t-1)+R(t-2)+R(t-3)) / 3), used for dynamic adjustment of the radial blur intensity in subsequent frames. All baseline parameter calculations are performed in the GPU's vector processor, with a single-frame processing time of ≤0.3ms, ensuring synchronization with the timing of the synchronous rendering mechanism.

[0032] S403. Based on the reference parameters of the vector continuation, the GPU is used to superimpose radial blur on the edge region of the first frame image after dimensionality reduction to obtain a radial blur effect. The blur intensity is proportional to the amplitude of the motion vector. In step S403, it is necessary to explain in detail that the radial blur effect is generated using a progressive processing scheme of "edge region localization - motion vector amplitude mapping - dynamic generation of blur kernel". First, the system locates the edge region of the first frame image after dimensionality reduction using existing image segmentation algorithms: taking the center of the image as the origin, the edge region range is determined according to the core operation region ratio in the hierarchical dimensionality reduction strategy (e.g., 20% of the center is the core region in 3-level dimensionality reduction). No blur processing is applied to the core region, and the edge region is divided into an inner transition zone (10% width outside the core region) and an outer blur region (the remaining edge part). The blur intensity of the inner transition zone increases linearly from the boundary of the core region outward to avoid harsh boundaries of the blur effect. Secondly, establish the mapping relationship between motion vector amplitude and blur intensity: Let the motion vector amplitude of a certain pixel be Vm=sqrt(Vxpred2+Vypred2), the baseline blur intensity Im=Vm×Dr×K (where K is the scene coefficient, K=0.8 for military simulation scene, K=1.2 for entertainment scene), the actual blur intensity of the inner transition area Iinner=Im×(d / D), where d is the distance from the pixel to the boundary of the core area, and D is the width of the inner transition area; the actual blur intensity of the outer blur area Iouter=Im, to ensure the correlation between the blur effect and the motion trend. Finally, a Gaussian blur kernel is dynamically generated based on the blur intensity: the kernel size N (odd number) is positively correlated with I, satisfying N=2×round(I×5)+1 (e.g., N=3 when I=0.3, N=9 when I=0.8). The kernel function adopts a two-dimensional Gaussian distribution G(x,y)=(1 / (2πσ2))e(-(x²+y²) / (2σ²)), where σ=N / 6, and the direction of the blur kernel is consistent with the average direction of the motion vector in the region (determined by averaging the motion vectors in the 3×3 neighborhood), achieving a directional blur effect. Radial blur processing is executed in parallel in the fragment shader of the GPU, using a block calculation strategy (each block is 128×128 pixels). The blur processing time per frame is ≤1.5ms, and MIPMAP texture optimization technology is used to avoid noise amplification in the edge region, ensuring a natural blur effect without sacrificing the image clarity of the core region.

[0033] S404. Based on the radial blurring effect, generate the final visual impact compensation.

[0034] In step S404, it is necessary to explain in detail that the final visual impact compensation is achieved through a comprehensive mechanism of "blurring effect and physiological feedback closed-loop calibration - multimodal sensory collaboration - user adaptive learning". First, the radial blurring effect generated in step S403 is fed back to the multimodal feedback fusion unit in real time, and a correlation analysis is performed with the synchronously collected user physiological data (HRV change rate ΔHRV, GSR fluctuation value ΔGSR): when ΔHRV>5% or ΔGSR>10%, it is determined that the blur intensity is insufficient, and the blur kernel coefficient K is automatically dynamically increased (increased by 0.1 each time, with an upper limit of 1.5); conversely, if ΔHRV<-3% and ΔGSR<-5%, it is determined that the blur is excessive, causing image distortion, and the K value is decreased (decreased by 0.05 each time, with a lower limit of 0.5), forming a closed-loop calibration of "rendering effect - physiological response". Secondly, the visual impact compensation and breathing guidance subunit work in deep collaboration: Simultaneously with the generation of the radial blur effect, the VR device's bone conduction headphones output low-frequency soothing sound waves matching the blur intensity (frequency varies with blur intensity Iinner / Iouter, ranging from 40-80Hz). The headset's built-in vibration module generates a slight 0.3G vibration during exhalation (vibration duration equal to the number of frames the blur effect lasts × frame period). This multimodal input of hearing and touch distracts the visual focus, further reducing the dizziness caused by sudden depth changes. Furthermore, the system integrates a user adaptive learning module, recording different users' visual impact compensation parameters (such as K-value, blur kernel size N, and multimodal feedback intensity) and corresponding dizziness scores (automatically generated through changes in scan rate based on real-time button feedback or eye-tracking) at various dimensionality reduction levels. Using a K-means clustering algorithm, users are divided into three groups: "highly sensitive," "moderately sensitive," and "lowly sensitive." New users are loaded with the optimal compensation parameters for their group by default, and the personalized parameter library is automatically updated every 10 dimensionality reduction switches during use, achieving continuous optimization of the compensation effect. The final output visual impact compensation screen must meet three indicators: edge transition smoothness of the dynamic blur area ≥95% (evaluated by gradient descent algorithm), clarity retention rate of the core operation area ≥98% (using SSIM image similarity index), and user dizziness perception reduction ≥40% (weighted evaluation based on subjective rating and physiological indicators), to ensure that while effectively compensating for visual impact, it does not affect the user's cognition and operation of core content.

[0035] Example 2 A three-dimensional immersive anti-vertigo system, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to implement a three-dimensional immersive anti-vertigo method when executing executable instructions.

[0036] It should be noted that the computer device includes a processor, a memory, and may also include one or more of a multimedia component, an input / output (I / O) interface, and a communication component.

[0037] The processor controls the overall operation of the computer device to complete all or part of the steps in the above-mentioned three-dimensional immersive anti-vertigo method.

[0038] Memory is used to store various types of data to support the operation of the computer device. This data may include, for example, instructions for any application or method used to operate on the computer device, as well as application-related data. Memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0039] The multimedia component may include a screen and an audio component, wherein the screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals; for example, the audio component may include a microphone for receiving external audio signals, the received audio signals may be further stored in memory or transmitted via a communication component; the audio component may also include at least one speaker for outputting audio signals.

[0040] I / O interfaces provide interfaces between the processor and other interface modules, such as keyboards, mice, buttons, etc.; these buttons can be virtual buttons or physical buttons.

[0041] The communication component is used for wired or wireless communication between the computer device and other devices; wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G or 5G, or one or more combinations thereof, and the corresponding communication component may include: Wi-Fi module, Bluetooth module, NFC module, mobile communication module.

[0042] As a preferred embodiment, the computer device may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described three-dimensional immersive anti-vertigo method.

[0043] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A three-dimensional immersive anti-vertigo method, characterized in that, The method includes: A biofeedback-based individualized baseline calibration procedure uses a standardized vestibular-visual coupling test to generate an individual's vertigo sensitivity coefficient and resting heart rate variability baseline. Based on the individual's vertigo sensitivity coefficient and resting heart rate variability baseline, a dynamic physiological risk level and environmental overload vector assessment are generated using a nonlinear regression model and real-time monitored physiological signals and environmental physical vectors. Based on the dynamic physiological risk level and environmental overload vector assessment, a graded dimensionality reduction execution strategy is generated using the cross-dimensional sensory breathing control algorithm module. Based on the aforementioned hierarchical dimensionality reduction execution strategy, the motion vector conservation algorithm is used to generate a synchronous rendering mechanism and vector continuation effect at the moment of dimension switching. By extracting the motion vector of the previous frame and superimposing radial blur, the visual impact caused by the disappearance of the sense of depth is compensated, and the final visual impact compensation is generated. Based on the aforementioned hierarchical dimensionality reduction execution strategy and the final visual impact compensation, the centrifugal dynamic recovery model is used to gradually restore the image size and parallax according to the exponential recovery function after the physiological indicators are restored to a safe baseline.

2. The three-dimensional immersive anti-vertigo method as described in claim 1, characterized in that, The standardized vestibular-visual coupling test employs a rotating chair-eye-tracking joint paradigm, using sinusoidal stimulation within an angular acceleration range of 0.1-1.5 Hz. Simultaneously, the subject's slow-phase nystagmus velocity, subjective vertigo score, and skin conductance are recorded. The vertigo sensitivity coefficient is determined by calculating the Pearson correlation coefficient between eye movement gain and vertigo score. At the same time, electrocardiogram data are collected for n minutes at rest, and the high-frequency power spectral density is extracted using wavelet transform as the baseline for resting heart rate variability.

3. The three-dimensional immersive anti-vertigo method as described in claim 1, characterized in that, The generation of dynamic physiological risk levels and environmental overload vector assessments includes: Based on the individual's dizziness sensitivity coefficient and resting heart rate variability baseline, a nonlinear regression model is used to generate individualized basic parameters for physiological risk assessment. Based on the individualized physiological risk assessment parameters, a real-time individual physiological status assessment is generated using real-time monitored physiological signals. Based on the real-time individual physiological state assessment, an environmental overload vector assessment is generated using environmental physical vectors monitored by airborne IMU data. Based on the real-time assessment of individual physiological state and the real-time assessment of environmental overload vector, a dynamic comprehensive assessment of physiological risk level and environmental overload vector is generated using a preset hierarchical decision-making logic.

4. The three-dimensional immersive anti-vertigo method as described in claim 1, characterized in that, The generation of hierarchical dimensionality reduction execution strategies includes: Based on the dynamic physiological risk level, a preliminary dimensionality reduction response level determination is generated using the hierarchical dimensionality reduction execution strategy framework in the cross-dimensional sensory breathing control algorithm module. Based on the preliminary determination of the reduced-dimensional response level and the environmental overload vector, the specific parameters of the motion compensation mechanism are generated using the motion vector conservation algorithm. Based on the specific parameters of the motion compensation mechanism and the preliminary determination of the dimensionality reduction response level, the breathing guidance subunit in the cross-dimensional sensory breathing control algorithm module is used to generate a breathing rhythm guidance signal that matches the dimensionality reduction level. Based on the respiratory rhythm guidance signal, the specific parameters of the motion compensation mechanism, and the preliminary dimensionality reduction response level determination, a graded dimensionality reduction execution strategy is generated, which includes the image dimension reduction ratio, the degree of field of view compression, and the color saturation adjustment value.

5. A three-dimensional immersive anti-vertigo method as described in claim 4, characterized in that, The cross-dimensional sensory breathing control algorithm module includes a breathing rhythm generation unit, a breathing guidance subunit, a multimodal feedback fusion unit, and a dimensionality reduction parameter mapping unit.

6. The three-dimensional immersive anti-vertigo method as described in claim 1, characterized in that, The final visual impact compensation includes: Based on the aforementioned hierarchical dimensionality reduction execution strategy, a synchronous rendering mechanism is generated by triggering dimensionality switching using the starting point of the vertical synchronization signal. Based on the aforementioned synchronous rendering mechanism, the baseline parameters for vector continuation are generated using the motion vector data from the first three frames. Based on the baseline parameters of the vector continuation, radial blur is superimposed on the edge region of the first frame image after dimensionality reduction using the GPU to obtain a radial blur effect, and the blur intensity is proportional to the amplitude of the motion vector. Based on the radial blurring effect, the final visual impact compensation is generated.

7. The three-dimensional immersive anti-vertigo method as described in claim 1, characterized in that, The expression for the exponential recovery function is: in, Indicates time The degree of recovery at that time Indicates the maximum degree of recovery. This indicates the recovery slope.

8. A three-dimensional immersive anti-vertigo system, characterized in that, The system includes: processor; Memory used to store processor-executable instructions; The processor is configured to implement the three-dimensional immersive anti-vertigo method according to any one of claims 1 to 7 when executing the executable instructions.

Images