A method and system for quantitatively evaluating user visual cognitive chaos in a digital twin scene

Abstract

The application belongs to the technical field of digital twin cities, and discloses a method and system for quantitatively evaluating the visual cognitive chaos of a user in a digital twin scene, which comprises the following steps: acquiring eye movement sequence data of a target subject in the digital twin scene and pre-processing the eye movement sequence data; determining a reference frame of the eye movement sequence data, reconstructing the phase space of the pre-processed eye movement sequence data based on the reference frame; calculating the nearest neighbor points of all observation points in the eye movement sequence data in the phase space, and calculating the average logarithmic separation curve through the observation points and the nearest neighbor points; obtaining the fitting slope of the average logarithmic separation curve based on a linear regression method, determining the chaos degree of the target subject based on the fitting slope; and outputting the chaos degree determination result of the target subject; the application can quantitatively detect the visual cognitive chaos of the target subject in the digital twin scene, and can be applied to the fields of traffic safety monitoring, pedestrian risk prediction and human-computer interaction optimization.

Description

Technical Field

[0001] This invention relates to the field of digital twin city technology, and in particular to a quantitative evaluation method and system for user visual cognitive chaos in digital twin scenarios. Background Technology

[0002] In the fields of intelligent transportation and pedestrian safety technology, pedestrians and drivers need to process visual information quickly in highly dynamic and complex traffic environments.

[0003] However, the application of eye-tracking technology in the transportation field currently faces significant bottlenecks. Traditional analysis methods struggle to capture the “disorder” or “chaotic” characteristics exhibited by an individual’s visual search patterns when faced with sudden traffic situations or complex road conditions. These characteristics are often closely related to an individual’s cognitive load imbalance or safety risks, but existing technologies lack effective means to quantify and evaluate these characteristics.

[0004] Existing technologies assess chaos using traditional statistical time-counting metrics, such as fixation duration, fixation frequency, fixation density maps, saccade length, and AOI (Area of ​​Interest) switching frequency. However, these metrics reflect average behavioral characteristics and cannot directly characterize the dynamic stability or unpredictability of time series. Some existing technologies also rely on spectral or linear time-series models for analysis, such as power spectrum analysis, segmental Fourier transform analysis, or autoregressive model analysis of eye-tracking sequences to find dominant frequencies or stationarity, but these are limited in revealing nonlinear chaotic characteristics. Therefore, existing technologies tend to rely on statistical or linear assumptions, failing to delve into the complex intrinsic patterns of eye-tracking data over time and neglecting to use deeper nonlinear dynamic methods to assess the degree of "chaos" or "disorder" in visual cognition.

[0005] Existing technologies still face the challenge of quantifying and assessing the degree of chaos in visual cognition; therefore, existing technologies require further improvement. Summary of the Invention

[0006] The technical problem to be solved by the present invention is that, in view of the defects of the prior art, the present invention provides a quantitative evaluation method and system for user visual cognition chaos in digital twin scenarios, so as to solve the problem that the prior art is difficult to quantitatively evaluate the degree of chaos of visual cognition.

[0007] The technical solution adopted by this invention to solve the technical problem is as follows: In a first aspect, the present invention provides a quantitative evaluation method for user visual cognitive chaos in digital twin scenarios, including: Obtain eye movement sequence data of the subject in the digital twin scenario, and preprocess the eye movement sequence data; A reference frame is determined for the eye-tracking sequence data, and the phase space of the preprocessed eye-tracking sequence data is reconstructed based on the reference frame. Calculate the nearest neighbor of all observation points in the eye-tracking sequence data in phase space, and calculate the average logarithmic separation curve using the observation points and the nearest neighbors; The fitting slope of the mean logarithmic separation curve is obtained based on the linear regression method, and the degree of chaos of the subject's target is determined based on the fitting slope. Output the chaos determination result of the test target.

[0008] In one implementation, the step of acquiring eye-tracking sequence data of the subject in the digital twin scenario and preprocessing the eye-tracking sequence data includes: Acquire eye movement sequence data of the subject target in a digital twin scenario; wherein, the eye movement sequence data is eye movement sequence data based on the chronological order of timestamps, and each eye movement data in the sequence includes a timestamp and gaze coordinates; Eye-tracking sequence data based on the chronological order of the timestamps are obtained; The eye-tracking sequence data is time-baseline aligned.

[0009] In one implementation, determining the reference frame for the eye-tracking sequence data and reconstructing the phase space of the preprocessed eye-tracking sequence data based on the reference frame includes: Based on the head pose of the subject target, the reference frame of the eye movement sequence data is determined according to the head pose; Calculate the attitude transformation matrix based on the gaze coordinates and the reference frame; The gaze coordinates are converted into reference coordinates using the posture transformation matrix to obtain eye movement sequence data in phase space.

[0010] In one implementation, the calculation of the nearest neighbors of all observation points in the eye-tracking sequence data in the phase space, and the calculation of the average logarithmic separation curve using the observation points and the nearest neighbors, includes: The temporal separation threshold of eye-tracking sequence data in phase space is calculated based on the average period estimation method. Calculate the nearest neighbor of all observation points based on the time separation threshold; Construct an initial point pair set; wherein each initial point pair in the initial point pair set contains one observation point and one nearest neighbor of the observation point; The average logarithmic separation curve is calculated based on the initial set of points.

[0011] In one implementation, calculating the temporal separation threshold of eye-tracking sequence data in phase space based on the average period estimation method includes: The average period of the eye-tracking sequence data was calculated using spectral analysis. The time separation threshold is calculated based on the average period and a preset window.

[0012] In one implementation, calculating the mean logarithmic separation curve based on the initial set of point pairs includes: Calculate the Euclidean distance of the initial point pair set over all time steps; wherein, each initial point pair in the initial point pair set corresponds to one time step; Calculate the logarithmic mean of the Euclidean distances between the initial set of point pairs based on the time step; Construct an average logarithmic separation curve based on the logarithmic mean.

[0013] In one implementation, obtaining the fitting slope of the mean logarithmic separation curve based on a linear regression method, and determining the degree of chaos of the subject's target based on the fitting slope, includes: Based on the linear fitting method, the linear fitting model corresponding to the linear growth interval of the mean logarithmic separation curve is obtained. The linear fitting model is fitted using the least squares method within a preset time interval to obtain the fitting slope; wherein, the fitting slope is the Lyapunov coefficient; Determine whether the fitted slope is greater than a preset chaos threshold; If the fitting slope is greater than the preset chaos threshold, the chaos determination result of the subject target is then obtained as chaotic. If the fitting slope is not greater than the preset chaos threshold, the chaos determination result of the subject target is non-chaotic.

[0014] Secondly, the present invention provides a quantitative evaluation system for user visual cognitive chaos in a digital twin scenario, comprising: The data acquisition module is used to acquire eye movement sequence data of the subject target in the digital twin scenario and to preprocess the eye movement sequence data. The phase space reconstruction module is used to determine the reference frame of the eye movement sequence data and to reconstruct the phase space of the preprocessed eye movement sequence data based on the reference frame. The curve calculation module is used to calculate the nearest neighbor of all observation points in the eye-tracking sequence data in phase space, and to calculate the average logarithmic separation curve using the observation points and the nearest neighbor. The chaos determination module is used to obtain the fitting slope of the mean logarithmic separation curve based on the linear regression method, and to determine the chaos of the subject target based on the fitting slope. The results output module is used to output the chaos determination results of the test target.

[0015] Thirdly, the present invention provides a terminal, comprising: a processor and a memory, wherein the memory stores a user visual cognition chaos quantification evaluation program in a digital twin scenario, and the user visual cognition chaos quantification evaluation program in a digital twin scenario is executed by the processor to implement the operation of the user visual cognition chaos quantification evaluation method in a digital twin scenario as described in the first aspect.

[0016] Fourthly, the present invention also provides a computer-readable storage medium storing a user visual cognition chaos quantification evaluation program in a digital twin scenario, wherein the user visual cognition chaos quantification evaluation program in a digital twin scenario, when executed by a processor, is used to implement the operation of the user visual cognition chaos quantification evaluation method in a digital twin scenario as described in the first aspect.

[0017] The present invention, by employing the above technical solution, has the following effects: This invention provides a method and system for quantitatively evaluating visual cognitive chaos in a digital twin scenario, comprising: acquiring eye-tracking sequence data of a subject target in a digital twin scenario; preprocessing the eye-tracking sequence data; determining a reference frame for the eye-tracking sequence data; reconstructing the phase space of the preprocessed eye-tracking sequence data based on the reference frame; calculating the nearest neighbor of all observation points in the phase space of the eye-tracking sequence data; calculating the average logarithmic separation curve using the observation points and the nearest neighbor; obtaining the fitting slope of the average logarithmic separation curve based on a linear regression method; determining the degree of chaos of the subject target based on the fitting slope; and outputting the degree of chaos determination result of the subject target. This invention can quantitatively detect visual cognitive chaos of a subject target in a digital twin scenario, and can therefore be applied to fields such as traffic safety monitoring, pedestrian risk prediction, and human-computer interaction optimization. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the quantitative evaluation method for user visual cognition chaos in the digital twin scenario of this invention.

[0020] Figure 2 This is a functional schematic diagram of the terminal in one implementation of the present invention.

[0021] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0023] Exemplary methods Current technologies for assessing the chaos of individual visual cognition focus on macroscopic statistics, failing to quantitatively describe the dynamic uncertainty or degree of chaos in time series data. The lack of systematic methodologies and engineering implementation processes leads to incomparable results or the inability to reproduce them across different experiments / devices. Furthermore, there is a lack of operable and reportable indicators for the potential nonlinear dynamic characteristics in eye-tracking trajectories, such as the degree of chaos in a single scalar. Continuous assessment of the complexity of subjects' visual cognition in real-time or online scenarios is also impossible. These limitations make it difficult to meet the practical needs of intelligent transportation for in-depth safety risk assessment of pedestrian or driver visual cognitive states.

[0024] Specifically, traditional statistical time-counting metrics are used to assess chaos, such as fixation duration, fixation frequency, fixation density map, saccade length, and AOI (Area of ​​Interest) switching frequency. However, these metrics reflect average behavioral characteristics and cannot directly characterize the dynamic stability or unpredictability of time series. Some existing techniques also rely on spectral or linear time-series models for analysis, such as power spectrum analysis, segmental Fourier transform analysis, or autoregressive model analysis of eye-tracking sequences to find dominant frequencies or stationarity, but these are limited in revealing nonlinear chaotic characteristics. Therefore, existing techniques tend to rely on statistical or linear assumptions, failing to delve into the complex intrinsic patterns of eye-tracking data over time and neglecting to use deeper nonlinear dynamic methods to assess the degree of "chaos" or "disorder" in visual cognition.

[0025] Existing technologies still have the problem that pedestrian interaction models cannot support real-time simulation of scenarios with large-scale concurrent demands. Therefore, existing technologies need to be improved.

[0026] To address the above-mentioned technical problems, this invention provides a method for quantitatively evaluating user visual cognitive chaos in a digital twin scenario, comprising: acquiring eye-tracking sequence data of a subject target in a digital twin scenario; preprocessing the eye-tracking sequence data; determining a reference frame for the eye-tracking sequence data; reconstructing the phase space of the preprocessed eye-tracking sequence data based on the reference frame; calculating the nearest neighbor of all observation points in the phase space of the eye-tracking sequence data; calculating the average logarithmic separation curve using the observation points and the nearest neighbor; obtaining the fitting slope of the average logarithmic separation curve based on a linear regression method; determining the degree of chaos of the subject target based on the fitting slope; and outputting the degree of chaos determination result of the subject target. This invention can quantitatively detect visual cognitive chaos of a subject target in a digital twin scenario, and can therefore be applied to fields such as traffic safety monitoring, pedestrian risk prediction, and human-computer interaction optimization.

[0027] like Figure 1 As shown, this embodiment of the invention provides a quantitative evaluation method for user visual cognitive chaos in a digital twin scenario, including the following steps: Step S100: Obtain eye movement sequence data of the subject in the digital twin scenario, and preprocess the eye movement sequence data.

[0028] Specifically, in one implementation of this embodiment, step S100 includes the following steps: Step S101: Obtain eye movement sequence data of the subject target in the digital twin scenario; wherein, the eye movement sequence data is eye movement sequence data based on the chronological order of timestamps, and each eye movement data in the sequence includes a timestamp and gaze coordinates.

[0029] In this embodiment, an eye tracker captures real-time eye movement sequence data of the subject in a digital twin scenario. The eye movement sequence data is based on the chronological order of timestamps, with each eye movement data point representing a sampling point, and each sampling point including a timestamp. With gaze coordinates The gaze coordinates can be two-dimensional screen coordinates or three-dimensional coordinates in the head / world coordinate system.

[0030] In this embodiment, an eye tracker with a sampling rate of 100Hz was used to collect eye movement data of subject target number 001 in a digital twin traffic scene in real time. The scene is a simulated urban intersection environment, which includes dynamic elements such as traffic lights, vehicle flow and pedestrian flow. The collection time is 60 seconds, and a total of 6,000 sampling points are obtained.

[0031] Each sampling point contains: Timestamp (Unit: ms); 3D gaze coordinates: ; Head position coordinates of the subject: ; Head yaw angle .

[0032] Step S102: Eye-tracking sequence data based on the chronological order of the timestamps is obtained.

[0033] In this embodiment, EyeTrace, an eye movement sequence data based on the timestamps of the eye movement data, is obtained.

[0034] Step S103: Perform time baseline alignment on the eye-tracking sequence data.

[0035] In this embodiment, the eye-tracking sequence data is preprocessed. The preprocessing method mainly involves aligning the acquired eye-tracking sequence data with a time baseline to ensure that the data is sampled at a uniform interval. This indicates that the uniform sampling interval in this embodiment... .

[0036] In addition, if the original time is not uniform, interpolation can be used in the implementation to resample to a uniform time grid; if there are gaps or missing records, obvious gaps or missing records are marked.

[0037] like Figure 1 As shown, this embodiment of the invention provides a quantitative evaluation method for user visual cognitive chaos in a digital twin scenario, including the following steps: Step S200: Determine the reference frame of the eye-tracking sequence data, and reconstruct the phase space of the preprocessed eye-tracking sequence data based on the reference frame.

[0038] In this embodiment, the relative coordinate transformation method is used to obtain eye-tracking sequence data including gaze coordinates. It is necessary to transform the eye-tracking fixation point from the world / screen coordinates to a local coordinate system centered on the subject's head or torso to remove the exogenous influence caused by head displacement or posture changes. This will allow the visual search strategy and cognitive process itself to be better reflected in the subsequent fixation trajectory dynamics analysis, rather than reflecting changes in the subject's spatial position.

[0039] Specifically, in one implementation of this embodiment, step S200 includes the following steps: Step S201: Based on the head pose of the subject, determine the reference frame of the eye movement sequence data according to the head pose.

[0040] In this embodiment, based on the head position and posture of the test subject, a local reference frame L is selected with the head of the test subject as the center and the head posture as the coordinate axis.

[0041] In one implementation of this embodiment, a local reference system L centered on the subject's eyeball and with head posture as the coordinate axis can also be selected. To distinguish it from this local reference coordinate system, the gaze coordinate is represented as... .

[0042] Step S202: Calculate the attitude transformation matrix based on the gaze coordinates and the reference frame.

[0043] In this embodiment, the attitude transformation matrix is ​​calculated based on the gaze coordinates of the subject target and the reference frame. . is the column orthogonal form of the rotation matrix from world coordinates to head coordinates, or expressed by equivalent transformation.

[0044] In this embodiment, the attitude transformation matrix is ​​a rotation matrix about the vertical axis (Z-axis).

[0045] Step S203: The gaze coordinates are converted into reference coordinates using the posture transformation matrix to obtain eye movement sequence data in phase space.

[0046] In this embodiment, the gaze coordinates are transformed using a pose transformation matrix. The coordinates are converted to reference frame coordinates through translation and rotation transformations, and the corresponding formulas are as follows: ; in, The head position of the subject.

[0047] In this embodiment, taking the first sampling point as an example, the target head position of the subject is (12.35, 0.82, 5.67), the yaw angle is 135.2°, and the gaze coordinates are (15.21, 1.05, 8.43); after the rotation matrix transformation, the local reference frame coordinates are (-1.926, 0.230, -3.924).

[0048] In this embodiment, all coordinates in the eye-tracking sequence data are converted to reference frame coordinates to obtain the eye-tracking sequence data in the reference frame, and the phase space is further constructed: If the observation is in three-dimensional local coordinates, then the observation vector is directly used as the state vector. ,Right now: ; If the observation is one-dimensional or if time-delay embedding reconstruction is necessary, then time-delay embedding can be used to calculate the state vector. : ; in, Indicates the embedding dimension. Indicates a delay.

[0049] Specifically, when the observation is one-dimensional or noise-sensitive, time-delay embedding can be used, and the delay can be selected using the average mutual information method. Selecting the embedding dimension using the False Nearest Neighbors method .

[0050] In one implementation of this embodiment, if the observation is a multidimensional eigenvector, principal component analysis (PCA) can be performed first to reduce the dimensionality before constructing the phase space to reduce the impact of noise.

[0051] In this embodiment, all state vectors are finally obtained, each state vector corresponds to an observation point in the phase space, and finally eye-tracking sequence data in the phase space based on the state vectors are obtained.

[0052] In this embodiment, the observation is a three-dimensional local coordinate system, and the observation vector is directly used as the state vector. No time-delay embedding reconstruction is required. After coordinate transformation, 6000 three-dimensional state vectors are obtained, which constitute eye-tracking sequence data in phase space.

[0053] like Figure 1 As shown, this embodiment of the invention provides a quantitative evaluation method for user visual cognitive chaos in a digital twin scenario, including the following steps: Step S300: Calculate the nearest neighbor of all observation points in the eye-tracking sequence data in the phase space, and calculate the average logarithmic separation curve using the observation points and the nearest neighbor.

[0054] In this embodiment, the Average Period Estimation method is used to determine the nearest neighbor by estimating the typical time scale in the sequence, i.e., the average period. The average logarithmic separation curve is determined based on all observation points and their corresponding nearest neighbors in the eye-tracking sequence data in phase space.

[0055] Specifically, in one implementation of this embodiment, step S300 includes the following steps: Step S301: Calculate the time separation threshold of eye-tracking sequence data in phase space based on the average period estimation method.

[0056] In this embodiment, the average period of the eye-tracking sequence in phase space is estimated based on the average period estimation method, the Theiler window is further determined, and the average period is converted into the Theiler window in the form of sample number. .

[0057] Specifically, step S301 includes the following steps: Step S301a: Calculate the average period of the eye-tracking sequence data using spectral analysis.

[0058] In this embodiment, the typical timescale of the eye-tracking sequence in phase space is estimated based on the average period estimation method and used as the average period. Spectral analysis is then performed on the local coordinate sequence by dimension or by its magnitude sequence using methods such as Discrete Fourier Transform (DFT) and Fast Fourier Transform (FFT) to identify the dominant frequency. And from this, the average period is obtained. The corresponding formula is as follows: .

[0059] In this embodiment, a Fast Fourier Transform is performed on the X component of the local coordinate sequence to identify the dominant frequency. Analysis results show that the main frequency Corresponding average period .

[0060] In one implementation of this embodiment, in addition to the FFT main frequency method, the local or time-varying main period can also be estimated by methods such as the first main peak or the first zero of the autocorrelation function, the short-time Fourier transform (STFT), and the wavelet transform, and the Theiler window can be set accordingly.

[0061] Step S301b: Calculate the time separation threshold based on the preset window according to the average period.

[0062] In this embodiment, the average period is converted into a preset window in the form of sample number; the preset window is a Theiler window. The Theiler window Using sampling points as units, this is used as a time separation threshold in subsequent neighbor search to exclude candidate neighbors that are less than a certain time apart, so as to avoid points that are close in time but belong to the same trajectory segment being misclassified as nearest neighbors.

[0063] In this embodiment, the average period is converted into a Theiler window in the form of sample number: ; 1000 sampling points were calculated; considering computational efficiency, the Theiler window in this embodiment uses 10 sampling points as the time separation threshold.

[0064] Step S302: Calculate the nearest neighbor of all observation points based on the time separation threshold.

[0065] In this embodiment, the constrained nearest-neighbor search method is used to find candidate points in the phase space that satisfy the time separation constraint for each observation point (reference point) as the nearest neighbor points, which serve as the initial pairs for subsequent divergence calculations.

[0066] In this embodiment, the time separation constraint is: ; in, and These represent the time indices of the observation point and the candidate point, respectively.

[0067] Based on the above time separation constraints, and using Euclidean distance as a metric, for each observation point... Find the nearest neighbor among all candidate points that satisfies the time separation constraint. ,have: ; In this embodiment, if there are more than one nearest neighbor point that satisfies the time separation constraint, then the unique candidate point is selected as the nearest neighbor point according to the principle of minimum distance, that is, the candidate point with the smallest Euclidean distance to the observation point.

[0068] In this embodiment, each observation point in phase space is represented by a specific point. , , The number of sampling points corresponding to the time separation threshold in the Theiler window. The number of sampling points, i.e. Under the condition of satisfying the time separation constraint Among all candidate points, search for the nearest neighbor using Euclidean distance as the metric. .

[0069] Taking the 500th sampling point as an example: Reference point ; In the candidate point set (time index) Search among all points to find the nearest neighbor. Corresponding Euclidean distance .

[0070] In one implementation of this embodiment, weighted Euclidean distance, Mahalanobis distance, or other metrics based on local manifold learning can be used instead of Euclidean distance to adapt to the distribution of various types of data.

[0071] Step S303: Construct an initial point pair set; wherein each initial point pair in the initial point pair set contains one observation point and one nearest neighbor of the observation point.

[0072] In this embodiment, based on eye-tracking sequence data in phase space, the nearest neighbor point is calculated for each observation point in the eye-tracking sequence data. Initial point pairs are constructed based on the observation point and its corresponding nearest neighbor point, and an initial point pair set containing all initial point pairs is further obtained.

[0073] In this embodiment, each initial point pair in the initial point pair set contains an observation point and the nearest neighbor point corresponding to that observation point.

[0074] Step S304: Calculate the average logarithmic separation curve based on the initial set of point pairs.

[0075] In this embodiment, the Separation Evolution Calculation method is used to calculate the average logarithmic separation curve by tracking the Euclidean distance change of each initial point pair over time.

[0076] Specifically, step S304 includes the following steps: Step S304a: Calculate the Euclidean distance of the initial point pair set across all time steps; wherein each initial point pair in the initial point pair set corresponds to one time step.

[0077] In this embodiment, the Euclidean distance of the initial point pair set is calculated over all time steps. For any initial point pair... In the subsequent The corresponding point pairs of each time step are Under the premise that the index does not exceed the bounds and records exist for both the observation point and its nearest neighbor, the first is defined as... The Euclidean distance of the step is: .

[0078] Step S304b: Calculate the logarithmic mean of the Euclidean distances between the initial point pairs based on the time step.

[0079] In this embodiment, the logarithm of the Euclidean distance between the initial set of point pairs is calculated based on the time step, i.e. It is important to avoid numerical problems caused by zero distance. For example, a very small constant can be added to small distances to avoid this. .

[0080] Step S304c: Construct an average logarithmic separation curve based on the logarithmic mean.

[0081] In this embodiment, the logarithmic mean of the Euclidean distances between the initial point pairs is calculated based on the time step. Specifically, the logarithm is taken and applied to all initial point pairs in the same way. Taking the average, we obtain the logarithmic separation curve: ; in In the first The set of valid point-pair indices that can still be computed in one step. , This represents the number of valid corresponding point pairs that can still be found when advancing the time to the k-th step. The corresponding time for each step is , This indicates the time interval for eye-tracking sampling.

[0082] In this embodiment, a total of 5489 valid initial point pairs were used, and the maximum number of evolution steps was 4500. The average logarithmic separation values ​​of some key time steps are shown in Table 1 below: Table 1. Schematic diagram of the average logarithmic separation value of the key time step.

[0083]

[0084] like Figure 1 As shown, this embodiment of the invention provides a quantitative evaluation method for user visual cognitive chaos in a digital twin scenario, including the following steps: Step S400: Obtain the fitting slope of the mean logarithmic separation curve based on the linear regression method, and determine the degree of chaos of the subject target based on the fitting slope.

[0085] In this embodiment, the chaos degree of the subject's target is determined by the linear fitting and LLE estimation method. Specifically, the Rosenstein small sample method is used to perform linear regression on the approximately linear segment of the mean logarithmic separation curve, and the fitting slope is the estimated value of the maximum Lyapunov exponent.

[0086] In one implementation of this embodiment, in addition to the Rosenstein small sample method, the Kantz algorithm, the Wolf algorithm, or the LLE estimation method based on maximum likelihood can also be used to obtain robust estimates under different conditions.

[0087] It should be noted that Lyapunov coefficients are quantitative indicators used to describe the sensitivity of orbits to initial conditions in deterministic dynamical systems. The largest Lyapunov exponent (LLE): For multidimensional systems, there exists a set of Lyapunov exponents (equal to the phase space dimension) that describe the local linear growth rate of the system in different characteristic directions; the largest value is called the largest Lyapunov exponent.

[0088] Specifically, in one implementation of this embodiment, step S400 includes the following steps: Step S401: Based on the linear fitting method, obtain the linear fitting model corresponding to the linear growth interval of the average logarithmic separation curve.

[0089] In this embodiment, the mean logarithmic separation curve In its approximately linear growth range ( to Within this range, based on the linear fitting method, the mean logarithmic separation curve... For the linear growth interval, a linear fitting method is applied, and the corresponding fitting model is: ; in, The maximum Lyapunov index, The intercept is usually a constant.

[0090] In this embodiment, the parameters corresponding to the fitting model are as follows: Fitting slope ; intercept ; Goodness of fit .

[0091] Step S402: Use the least squares method to fit the linear fitting model within a preset time interval to obtain the fitting slope; wherein, the fitting slope is the Lyapunov coefficient.

[0092] In this embodiment, the least squares method is used to fit the above linear model to the selected time interval, and the fitting slope is the estimate of LLE. .

[0093] Step S403: Determine whether the fitting slope is greater than a preset chaos threshold.

[0094] In this embodiment, the method for determining the chaos degree of the test target is to determine whether the fitting slope of the average logarithmic separation curve corresponding to the eye movement sequence data of the test target is greater than a preset chaos degree threshold.

[0095] Step S404: If the fitting slope is greater than the preset chaos threshold, then the chaos determination result of the subject target is obtained as chaotic.

[0096] In this embodiment, the preset chaos threshold is 0, that is... When the slope of the average logarithmic separation curve corresponding to the eye movement sequence data of the test target is greater than the preset degree threshold, the chaos degree judgment result of the test target is obtained as chaotic.

[0097] In this embodiment, Significantly positive indicates that the visual cognitive state of the subject's target number 001 in this digital twin traffic scene exhibits chaotic characteristics, suggesting that its visual search pattern is sensitively dependent on initial conditions, and the gaze trajectory shows an exponential divergence trend in phase space.

[0098] Step S405: If the fitting slope is not greater than the preset chaos threshold, the chaos determination result of the subject target is non-chaotic.

[0099] In this embodiment, the preset chaos threshold is 0, that is... When the slope of the average logarithmic separation curve corresponding to the eye-tracking sequence data of the test target is less than the preset degree threshold, the chaos degree judgment result of the test target is non-chaotic.

[0100] In this embodiment, the slope of the average logarithmic separation curve fitted to the eye-tracking sequence data of the test subject may also fail to support the determination. For example, the eye-tracking sequence data may be too short to fully unfold the evolutionary trajectory in the phase space; or the average logarithmic separation curve may not show a clear linear growth interval, making it impossible to perform effective least squares fitting; or the maximum Lyapunov exponent obtained by fitting may be extremely close to 0, and its confidence interval may span 0 (i.e., include both positive and negative values), making it impossible to statistically distinguish whether it is positive or negative; or the data missing rate may be too high, leading to the failure of phase space reconstruction. When the above problems occur, the output result is undeterminable.

[0101] like Figure 1 As shown, this embodiment of the invention provides a quantitative evaluation method for user visual cognitive chaos in a digital twin scenario, including the following steps: Step S500: Output the chaos determination result of the test target.

[0102] In this embodiment, the chaos determination result of the test target is used as the final output, that is, the Lyapunov coefficient of the test target is output. The result of the chaos determination; where the chaos determination result is chaotic, non-chaotic, or undeterminable.

[0103] In this embodiment, when the quantitative evaluation method for user visual cognition chaos in a digital twin scenario is applied to fields such as traffic safety monitoring, pedestrian risk prediction, and human-computer interaction optimization, in addition to the chaos degree judgment result of the subject's target, the output content may also include, but is not limited to: the confidence interval of LLE, the start and end time of the fitting interval, and the fitting quality index (such as...). ), the average period value used for the Theiler window, and the phase space construction parameters (if time-delay embedding is used). and The initial number of point pairs and preprocessed information are used to combine the above objective indicators with the visual cognitive characteristics of the subjects, such as attention stability and cognitive load status, in order to conduct in-depth reproducibility verification and statistical comparison.

[0104] In this embodiment, the output result of the chaos determination for subject target number 001 can be referred to Table 2.

[0105] Table 2 Output results of the chaos determination for subject target number 001.

[0106]

[0107] In this embodiment, the output results can be output in the form of offline files, such as CSV or JSON formats, or a visual report with a mean logarithmic separation curve and a fitted straight line can be output; the output results can also be pushed to the upper-layer application in real time through a network interface.

[0108] In one implementation of this embodiment, eye-tracking data is collected from twelve subjects in different scenarios. Each subject completes a 60-second pedestrian crossing task in each of the two scenarios, and the maximum Lyapunov index for each subject in different scenarios is calculated. .

[0109] Scenario A is a low-complexity scenario, a one-way two-lane road with a traffic flow of 6 vehicles per minute, no alternating traffic lights, and few visual interference elements; Scenario B is a high-complexity scenario, a two-way six-lane intersection with a traffic flow of 30 vehicles per minute, including periodic changes in traffic lights, pedestrian flow, non-motorized vehicle flow, and dynamic billboards, among other visual interference elements.

[0110] The final results are shown in Table 3 below: Table 3 Output results of chaos determination for the test subjects in different scenarios.

[0111]

[0112] Table 3 shows that in low-complexity scenarios, the visual search patterns of most participants tend to be ordered or weakly chaotic; in high-complexity traffic scenarios, the visual cognitive state of the participants' targets generally exhibits stronger chaotic characteristics, and the divergence of the gaze trajectory in phase space is significantly increased; based on the maximum Lyapunov index of the same participant's target in different scenarios... It can be seen that the increase in scene complexity significantly increases the degree of chaos in visual cognition; and according to the goodness of fit The values ​​show that the linear growth characteristic of the mean logarithmic separation curve is more significant in complex scenarios, and the LLE estimation is more reliable; the maximum Lyapunov exponent of the chaos determination output results of the above 12 subjects in different scenarios. The results are statistically significant, demonstrating the effective ability of the proposed method for quantifying and evaluating user visual cognitive chaos in digital twin scenarios to distinguish differences in visual cognitive chaos under different complexity scenarios.

[0113] In one implementation of this embodiment, to further verify the reliability of the chaos determination result, a statistical significance test based on surrogate data is introduced as follows: Step a: Obtain the raw eye movement sequence data of the subject target and calculate its maximum Lyapunov index (LLE) as the baseline value; Step b involves using a phase randomization algorithm (such as the IAAFT algorithm) to preserve the power spectrum and amplitude distribution characteristics of the original data, destroy its nonlinear deterministic structure, and generate multiple sets (such as 39 sets) of proxy data. Step c: Calculate the LLE value for each group of proxy data and construct the LLE distribution of the proxy data; In step d, if the LLE baseline value of the original data is significantly smaller than the mean of the LLE distribution of the surrogate data (e.g., exceeding the 95% confidence interval), then the null hypothesis is rejected, and the original eye-tracking sequence is determined to have significant nonlinear chaotic characteristics.

[0114] This embodiment achieves the following technical effects through the above technical solution: Unlike traditional statistical indicators of visual cognitive chaos, this embodiment uses the maximum Lyapunov index as a quantitative indicator to directly measure the "unpredictability / chaos" of visual cognition, which facilitates comparison and statistical testing. By organically linking key steps such as relative coordinate transformation, period estimation, restricted neighbor search, divergence calculation and linear fitting, a reproducible and implementable systematic engineering process is formed, which is convenient for cross-laboratory or cross-equipment applications. It supports direct phase space construction of multidimensional observation vectors; it is also compatible with time-delay embedding methods for one-dimensional observations, making it highly adaptable; it also supports real-time estimation based on sliding windows, making it suitable for both online monitoring and offline in-depth analysis scenarios. In summary, the quantitative evaluation method for user visual cognition chaos in the digital twin scenario presented in this embodiment has significant application value and promotion potential in fields such as intelligent traffic safety monitoring, pedestrian behavior risk prediction, and human-computer interaction optimization. It can effectively solve the practical problem of the lack of deep visual dynamics evaluation methods in this field.

[0115] Exemplary device Based on the above embodiments, the present invention also provides a quantitative evaluation system for user visual cognitive chaos in a digital twin scenario, comprising: The data acquisition module is used to acquire eye movement sequence data of the subject target in the digital twin scenario and to preprocess the eye movement sequence data. The phase space reconstruction module is used to determine the reference frame of the eye movement sequence data and to reconstruct the phase space of the preprocessed eye movement sequence data based on the reference frame. The curve calculation module is used to calculate the nearest neighbor of all observation points in the eye-tracking sequence data in phase space, and to calculate the average logarithmic separation curve using the observation points and the nearest neighbor. The chaos determination module is used to obtain the fitting slope of the mean logarithmic separation curve based on the linear regression method, and to determine the chaos of the subject target based on the fitting slope. The results output module is used to output the chaos determination results of the test target.

[0116] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 2 As shown.

[0117] The terminal includes: a processor, a memory, an interface, a display screen, and a communication module connected via a system bus; wherein, the processor of the terminal provides computing and control capabilities; the memory of the terminal includes a computer-readable storage medium and internal memory; the computer-readable storage medium stores an operating system and computer programs; the internal memory provides an environment for the operation of the operating system and computer programs in the computer-readable storage medium; the interface is used to connect to external devices; the display screen is used to display relevant information; and the communication module is used to communicate with a cloud server or other devices.

[0118] When executed by a processor, this computer program is used to implement a quantitative evaluation method for user visual cognitive chaos in digital twin scenarios.

[0119] It will be understood by those skilled in the art that Figure 2The schematic diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal to which the present invention is applied. A specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0120] In one embodiment, a terminal is provided, comprising: a processor and a memory, the memory storing a user visual cognition chaos quantification evaluation program in a digital twin scenario, the user visual cognition chaos quantification evaluation program in a digital twin scenario being executed by the processor to implement the operation of the user visual cognition chaos quantification evaluation method in a digital twin scenario as described above.

[0121] In one embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a user visual cognition chaos quantification evaluation program in a digital twin scenario, which, when executed by a processor, is used to implement the operation of the above-described user visual cognition chaos quantification evaluation method in a digital twin scenario.

[0122] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, database, or other media used in the embodiments provided by this invention can include both non-volatile and volatile memory.

[0123] In summary, this invention provides a method and system for quantitatively evaluating user visual cognitive chaos in a digital twin scenario, comprising: acquiring eye-tracking sequence data of a subject target in a digital twin scenario, and preprocessing the eye-tracking sequence data; determining a reference frame for the eye-tracking sequence data, and reconstructing the phase space of the preprocessed eye-tracking sequence data based on the reference frame; calculating the nearest neighbor of all observation points in the phase space of the eye-tracking sequence data, and calculating the average logarithmic separation curve using the observation points and the nearest neighbor; obtaining the fitting slope of the average logarithmic separation curve based on a linear regression method, and determining the degree of chaos of the subject target based on the fitting slope; and outputting the degree of chaos determination result of the subject target. This invention can quantitatively detect visual cognitive chaos of a subject target in a digital twin scenario, and can therefore be applied to fields such as traffic safety monitoring, pedestrian risk prediction, and human-computer interaction optimization.

[0124] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A quantitative evaluation method for user visual cognitive chaos in a digital twin scenario, characterized in that, include: Obtain eye movement sequence data of the subject in the digital twin scenario, and preprocess the eye movement sequence data; A reference frame is determined for the eye-tracking sequence data, and the phase space of the preprocessed eye-tracking sequence data is reconstructed based on the reference frame. Calculate the nearest neighbor of all observation points in the eye-tracking sequence data in phase space, and calculate the average logarithmic separation curve using the observation points and the nearest neighbors; The fitting slope of the mean logarithmic separation curve is obtained based on the linear regression method, and the degree of chaos of the subject's target is determined based on the fitting slope. Output the chaos determination result of the test target.

2. The method for quantitative evaluation of user visual cognitive chaos in a digital twin scenario according to claim 1, characterized in that, The acquisition of eye-tracking sequence data of the subject in the digital twin scenario, and the preprocessing of the eye-tracking sequence data, include: Acquire eye movement sequence data of the subject target in a digital twin scenario; wherein, the eye movement sequence data is eye movement sequence data based on the chronological order of timestamps, and each eye movement data in the sequence includes a timestamp and gaze coordinates; Eye-tracking sequence data based on the chronological order of the timestamps are obtained; The eye-tracking sequence data is time-baseline aligned.

3. The method for quantitative evaluation of user visual cognitive chaos in a digital twin scenario according to claim 2, characterized in that, The step of determining the reference frame for the eye-tracking sequence data and reconstructing the phase space of the preprocessed eye-tracking sequence data based on the reference frame includes: Based on the head pose of the subject target, the reference frame of the eye movement sequence data is determined according to the head pose; Calculate the attitude transformation matrix based on the gaze coordinates and the reference frame; The gaze coordinates are converted into reference coordinates using the posture transformation matrix to obtain eye movement sequence data in phase space.

4. The method for quantitative evaluation of user visual cognitive chaos in a digital twin scenario according to claim 1, characterized in that, The calculation of the nearest neighbor of all observation points in the eye-tracking sequence data in the phase space, and the calculation of the mean logarithmic separation curve using the observation points and the nearest neighbors, includes: The temporal separation threshold of eye-tracking sequence data in phase space is calculated based on the average period estimation method. Calculate the nearest neighbor of all observation points based on the time separation threshold; Construct an initial point pair set; wherein each initial point pair in the initial point pair set contains one observation point and one nearest neighbor of the observation point; The average logarithmic separation curve is calculated based on the initial set of points.

5. The quantitative evaluation method for user visual cognitive chaos in a digital twin scenario according to claim 4, characterized in that, The calculation of the temporal separation threshold of eye-tracking sequence data in phase space based on the average period estimation method includes: The average period of the eye-tracking sequence data was calculated using spectral analysis. The time separation threshold is calculated based on the average period and a preset window.

6. The method for quantitative evaluation of user visual cognitive chaos in a digital twin scenario according to claim 4, characterized in that, The calculation of the average logarithmic separation curve based on the initial set of point pairs includes: Calculate the Euclidean distance of the initial point pair set over all time steps; wherein, each initial point pair in the initial point pair set corresponds to one time step; Calculate the logarithmic mean of the Euclidean distances between the initial set of point pairs based on the time step; Construct an average logarithmic separation curve based on the logarithmic mean.

7. The method for quantitative evaluation of user visual cognitive chaos in a digital twin scenario according to claim 1, characterized in that, The step of obtaining the fitting slope of the mean logarithmic separation curve based on the linear regression method, and determining the degree of chaos of the subject's target based on the fitting slope, includes: Based on the linear fitting method, the linear fitting model corresponding to the linear growth interval of the mean logarithmic separation curve is obtained. The linear fitting model is fitted using the least squares method within a preset time interval to obtain the fitting slope; wherein, the fitting slope is the Lyapunov coefficient; Determine whether the fitted slope is greater than a preset chaos threshold; If the fitting slope is greater than the preset chaos threshold, the chaos determination result of the subject target is then obtained as chaotic. If the fitting slope is not greater than the preset chaos threshold, the chaos determination result of the subject target is non-chaotic.

8. A quantitative evaluation system for user visual cognitive chaos in a digital twin scenario, characterized in that, include: The data acquisition module is used to acquire eye movement sequence data of the subject target in the digital twin scenario and to preprocess the eye movement sequence data. The phase space reconstruction module is used to determine the reference frame of the eye movement sequence data and to reconstruct the phase space of the preprocessed eye movement sequence data based on the reference frame. The curve calculation module is used to calculate the nearest neighbor of all observation points in the eye-tracking sequence data in phase space, and to calculate the average logarithmic separation curve using the observation points and the nearest neighbor. The chaos determination module is used to obtain the fitting slope of the mean logarithmic separation curve based on the linear regression method, and to determine the chaos of the subject target based on the fitting slope. The results output module is used to output the chaos determination results of the test target.

9. A terminal, characterized in that, include: The processor and memory, wherein the memory stores a quantitative evaluation program for user visual cognition chaos in a digital twin scenario, and the quantitative evaluation program for user visual cognition chaos in a digital twin scenario, when executed by the processor, is used to implement the operation of the quantitative evaluation method for user visual cognition chaos in a digital twin scenario as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a quantitative evaluation program for user visual cognition chaos in a digital twin scenario. When executed by a processor, the quantitative evaluation program for user visual cognition chaos in a digital twin scenario is used to implement the operation of the quantitative evaluation method for user visual cognition chaos in a digital twin scenario as described in any one of claims 1-7.

Images