A data-driven crowd group motion simulation method and system

Abstract

This disclosure belongs to the field of crowd evacuation simulation technology, and provides a data-driven crowd group movement simulation method and system, including the following steps: acquiring crowd video data; dividing the acquired crowd video data into crowd groups; extracting crowd movement attributes based on the divided crowd groups; quantifying the extracted crowd movement attributes into group attributes to obtain group attribute quantification results; and obtaining a simulation animation of crowd group movement based on the obtained group attribute quantification results and a crowd simulation platform to realize the simulation of crowd group movement.

Description

Technical Field

[0001] This disclosure belongs to the field of crowd evacuation simulation technology, specifically relating to a data-driven crowd group movement simulation method and system. Background Technology

[0002] The statements in this section are merely background information relating to this disclosure and do not necessarily constitute prior art.

[0003] With the increase in the world's population and social activities, emergencies are increasingly likely to occur in crowded public places. In such emergencies, crowds can easily overcrowd and stampede. Realistic crowd evacuation drills not only consume significant resources but also fail to adequately handle changing scenarios. Therefore, crowd simulation technology has emerged to address this problem and has attracted widespread attention. This technology can scientifically simulate crowd movement, providing safe guidance for crowd evacuation. The development of this technology is of great significance for maintaining public safety.

[0004] Traditional crowd evacuation simulation methods focus on analyzing individual behavior within a crowd, neglecting the analysis of potential group attributes during crowd movement, thus reducing the visual realism of the simulation. During crowd movement, people unconsciously self-organize due to shared destinations and social relationships, forming groups. Considering the impact of group attributes on crowd movement is crucial for improving the visual realism of crowd movement simulations. The visual realism of the simulation results is paramount in the simulation process.

[0005] According to the inventors, data-driven methods are typically used in crowd simulations to analyze and study individual behavior, while neglecting inter-group attributes. Group attributes, as a major factor influencing group behavior, have received widespread attention and research in social psychology and biology. In research, group attributes are mainly divided into intra-group attributes and inter-group attributes. Intra-group attributes are characterized by stability, i.e., the stability of relationships within the same group, reflecting internal coordination among members. Inter-group attributes are characterized by conflict, reflecting the conflict characteristics between members of one group and other groups, representing inter-group interaction. Summary of the Invention

[0006] To address the aforementioned issues, this disclosure proposes a data-driven crowd movement simulation method and system. By analyzing crowd attributes, the method models crowd movement, accurately quantifies the group attributes within the crowd, analyzes the impact of these attributes on crowd movement, and realistically simulates crowd movement.

[0007] According to some embodiments, the first solution of this disclosure provides a data-driven crowd movement simulation method, which adopts the following technical solution:

[0008] A data-driven method for simulating crowd movement includes the following steps:

[0009] Acquire crowd video data;

[0010] The acquired video data of people is divided into groups.

[0011] Extracting crowd movement attributes based on the segmented crowd groups;

[0012] The extracted crowd movement attributes are quantified into group attributes to obtain the group attribute quantification results;

[0013] Based on the obtained group attribute quantification results and the crowd simulation platform, a simulation animation of crowd group movement is obtained, realizing the simulation of crowd group movement.

[0014] As a further technical limitation, in the process of extracting crowd motion attributes, a target tracking and detection learning framework is used to sample and track the trajectory of the crowd after group division, so as to obtain the position and speed of the crowd, and use the position and speed as the crowd motion attributes.

[0015] Furthermore, the target tracking and detection learning includes a tracking module, a detection module, and a learning module.

[0016] As a further technical limitation, the group attribute quantification includes at least intra-group stability and inter-group conflict.

[0017] Furthermore, the intra-group stability includes the stability of the magnitude of the velocity of the group members and the stability of the direction of the velocity of the group members.

[0018] Furthermore, the inter-group conflict includes conflicts in the position of members between groups, conflicts in the magnitude of the speed of members between groups, and conflicts in the direction of the speed of members between groups.

[0019] Furthermore, the effectiveness of the intra-group stability and inter-group conflict is verified in a synthetic scene. The stability and conflict during the movement of the crowd group are visualized graphically to obtain a simulation animation of the crowd group movement.

[0020] According to some embodiments, the second aspect of this disclosure provides a data-driven crowd movement simulation system, employing the following technical solution:

[0021] A data-driven crowd movement simulation system includes:

[0022] The group information extraction module is configured to acquire crowd video data, divide the acquired crowd video data into crowd groups, and extract crowd motion attributes based on the divided crowd groups.

[0023] The group attribute quantification module is configured to perform group attribute quantification on the extracted crowd motion attributes to obtain the group attribute quantification results.

[0024] The crowd simulation module is configured to generate a simulation animation of crowd movement based on the obtained group attribute quantification results and the crowd simulation platform, thereby simulating the movement of the crowd.

[0025] According to some embodiments, a third aspect of this disclosure provides a computer-readable storage medium, employing the following technical solution:

[0026] A computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the steps of the data-driven crowd movement simulation method as described in the first aspect of this disclosure.

[0027] According to some embodiments, the fourth solution of this disclosure provides an electronic device that adopts the following technical solution:

[0028] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the data-driven crowd movement simulation method as described in the first aspect of this disclosure.

[0029] Compared with the prior art, the beneficial effects of this disclosure are as follows:

[0030] (1) This disclosure uses a data-driven group attribute quantification (DGPQ) model to accurately describe the characteristics of group motion. Based on this model, the motion attributes of the crowd in the video are extracted, and the basic attributes of the crowd are quantified, namely, intra-group stability and inter-group conflict.

[0031] (2) This disclosure uses a stability and conflict-based crowd movement analysis (SC-CMA) model to analyze the impact of quantitative intragroup stability and intergroup conflict on crowd movement.

[0032] (3) This disclosure applies group stability and conflict to crowd simulation and proves the effectiveness of SC-CMA under different scenario settings;

[0033] (4) This invention builds a crowd simulation platform and compares it with different crowd simulation methods, which can more realistically simulate crowd movement. Attached Figure Description

[0034] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0035] Figure 1This is a flowchart of the data-driven crowd movement simulation method in Embodiment 1 of this disclosure;

[0036] Figure 2 This is a framework diagram of the data-driven crowd movement simulation method in Embodiment 1 of this disclosure;

[0037] Figure 3(a) is a group division result based on CT in Embodiment 1 of this disclosure;

[0038] Figure 3(b) is a diagram of the group trajectory extraction results in Embodiment 1 of this disclosure;

[0039] Figure 4(a) is a verification diagram of the effectiveness of group stability when S=0.35 and t=0s in Embodiment 1 of this disclosure;

[0040] Figure 4(b) is a verification diagram of the effectiveness of group stability when S=0.35 and t=10s in Embodiment 1 of this disclosure;

[0041] Figure 4(c) is a verification diagram of the effectiveness of group stability when S=0.35 and t=20s in Embodiment 1 of this disclosure;

[0042] Figure 5(a) is a verification diagram of the effectiveness of group stability when S=0.84 and t=0s in Embodiment 1 of this disclosure;

[0043] Figure 5(b) is a verification diagram of the effectiveness of group stability when S=0.84 and t=10s in Embodiment 1 of this disclosure;

[0044] Figure 5(c) is a verification diagram of the effectiveness of group stability when S=0.84 and t=20s in Embodiment 1 of this disclosure;

[0045] Figure 6(a) is a verification diagram of the effectiveness of group stability when S=1.76 and t=0s in Embodiment 1 of this disclosure;

[0046] Figure 6(b) is a verification diagram of the effectiveness of group stability when S=1.76 and t=10s in Embodiment 1 of this disclosure;

[0047] Figure 6(c) is a verification diagram of the effectiveness of group stability when S=1.76 and t=20s in Embodiment 1 of this disclosure;

[0048] Figure 7(a) is a verification diagram of the effectiveness of group conflict when C=1.53 in Embodiment 1 of this disclosure;

[0049] Figure 7(b) is a verification diagram of the effectiveness of group conflict when C=1.24 in Embodiment 1 of this disclosure;

[0050] Figure 8(a) is a simulation result of crowd movement without considering group attributes at t=5s in Embodiment 1 of this disclosure;

[0051] Figure 8(b) is a simulation result of crowd movement without considering group attributes at t=10s in Embodiment 1 of this disclosure;

[0052] Figure 8(c) is a simulation result of crowd movement without considering group attributes at t=15s in Embodiment 1 of this disclosure;

[0053] Figure 8(d) is a simulation result of crowd movement without considering group attributes at t=20s in Embodiment 1 of this disclosure;

[0054] Figure 9(a) is a simulation result of crowd movement considering group attributes at t=5s in Embodiment 1 of this disclosure;

[0055] Figure 9(b) is a simulation result of crowd movement considering group attributes at t=10s in Embodiment 1 of this disclosure;

[0056] Figure 9(c) is a simulation result of crowd movement considering group attributes at t=15s in Embodiment 1 of this disclosure;

[0057] Figure 9(d) is a simulation result of crowd movement considering group attributes at t=20s in Embodiment 1 of this disclosure;

[0058] Figure 10 This is a structural block diagram of the data-driven crowd movement simulation system in Embodiment 2 of this disclosure. Detailed Implementation

[0059] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0060] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0061] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0062] Where there is no conflict, the embodiments and features described herein can be combined with each other.

[0063] Example 1

[0064] Embodiment 1 of this disclosure introduces a data-driven method for simulating the movement of crowd groups.

[0065] like Figure 1 and Figure 2 The data-driven crowd movement simulation method shown includes the following steps:

[0066] Acquire crowd video data;

[0067] The acquired video data of people is divided into groups.

[0068] Extracting crowd movement attributes based on the segmented crowd groups;

[0069] The extracted crowd movement attributes are quantified into group attributes to obtain the group attribute quantification results;

[0070] Based on the obtained group attribute quantification results and the crowd simulation platform, a simulation animation of crowd group movement is obtained, realizing the simulation of crowd group movement.

[0071] Extracting group information from a crowd

[0072] Groups are an important component of crowds. Pedestrians typically form groups when engaging in a series of social activities. To more realistically simulate crowd dispersal, it's convenient to extract the motion characteristics of crowds and better capture their potential dynamics. Crowd videos are processed primarily from the following two aspects:

[0073] (1) Grouping

[0074] The CT method, a robust approach for grouping people in video footage, is employed for segmentation. This method proposes a CT prior to capture collective behavior prior to group formation. By learning the CT prior, accurate grouping can be achieved. Given a short video clip, the crowd can be divided into several groups; Figure 3(a) shows the results of the group segmentation, where feature points of the same color represent pedestrians in the same group, and the coordinates of the feature points for each pedestrian are recorded.

[0075] (2) Motion Attribute Extraction

[0076] The Tracking-Detection-Learning (TLD) framework is used to sample and track the trajectories of crowds in a video. In video moving object tracking, TLD is an architecture for long-term tracking of objects in a video. Simply put, the TLD algorithm consists of three parts: a tracking module, a detection module, and a learning module. The tracking module observes the movement of the target between frames. The detection module treats each image as independent and then locates the target. The learning module evaluates the errors of the detection module based on the results of the tracking module, generates training samples to update the target model of the detection module, and avoids similar errors in the future. As shown in Figure 3(b), the required position and velocity of the crowd are obtained through the TLD object tracking algorithm, and position and velocity are used as motion attributes to quantify the group attributes.

[0077] The position and velocity of each individual in the video are represented by a series of two-dimensional position coordinates; based on the tracking results, the grouping of the crowd in the video is recorded and a triple I = (P i V i O i ) represents the position, speed, and orientation information of each individual i in the video. This represents the position of individual i at time t. This represents the velocity magnitude of individual i at time t. This represents the direction of individual i at time t, where Here, n represents all the frames in the video.

[0078] Group attribute quantification

[0079] Groups possess a variety of group attributes. Different group systems share similar principles and can be represented by a set of basic behavioral attributes. These group behavioral attributes play a crucial role in improving the realism of crowd simulations. Among these attributes, intragroup stability and intergroup conflict are the most important.

[0080] On the one hand, intragroup stability is primarily manifested in the attributes within the group. The characteristic of stability attributes is that individuals within the group move in small groups, and members maintain a stable connection over a period of time. For example, in mass disasters, the stability of the group's collective movement becomes particularly important when there are different social relationships within the group. On the other hand, intergroup conflict is primarily manifested in the attributes between groups. The characteristic of conflict attributes is the interaction and influence between groups when they approach each other. For example, two groups of pedestrians with different goals may exhibit conflicting behavior when crossing a road from different directions.

[0081] (1) Quantification of intragroup stability

[0082] Stability is an important group property in group motion, referring to the characteristics of group motion formed by the special relationships among members within the group. It is a general property for describing group motion. Stable groups have the following characteristics: (1) the speed magnitude of members within the group remains stable; (2) the speed direction of members within the group remains stable. Therefore, based on the extracted real data, this embodiment quantifies the stability of the group from the above two aspects.

[0083] Regarding the stability of velocity magnitude, first calculate the average velocity magnitude of all members in the group at time t:

[0084]

[0085] In the formula, n represents the number of members in the group. Let represent the velocity magnitude of i at time t. Then, define a velocity difference amplitude function spe(i,n,Δt) to represent the stability of the group's velocity magnitude over time Δt.

[0086]

[0087] In the formula, This represents the difference between the velocity of member i at time t and the average velocity. To make the difference more significant, the velocity fluctuation is calculated using variance in this embodiment. The smaller spe(i,n,Δt) is, the more stable the group.

[0088] Regarding stability in the velocity direction, firstly, the vector sum of the velocities of all members in the group at time t is calculated as follows:

[0089]

[0090] In the formula, Let represent the velocity direction of member i at time t. Then, define a direction difference amplitude function dir(i,n,Δt) to represent the stability of the velocity direction of the group over time Δt.

[0091]

[0092] In the formula, This represents the angular difference between the velocity direction of member i at time t and the average direction. The larger dir(i,n,Δt) is, the more stable the group.

[0093] Regarding group stability, based on the previously mentioned definition of stability, the stability of a group is defined as S(i,n,Δt), which is calculated as follows:

[0094]

[0095] In the above formula, W s +Wd =1, W s W d These represent the weights for stability in velocity magnitude and stability in velocity direction, respectively. The larger S(i,n,Δt) is, the more stable the group is.

[0096] (2) Quantification of intergroup conflict

[0097] In group movement, the movements of different groups influence each other. To better simulate crowd movement and improve the realism of the simulation, this embodiment proposes conflict to represent the influence of movement between groups. Generally, the greater the influence between groups, the greater the conflict. Therefore, based on extracted real data, this embodiment quantifies the conflict between groups from three aspects: positional conflict, velocity magnitude conflict, and velocity direction conflict.

[0098] Regarding positional conflict, firstly, the average position of all members in each group at time t is calculated as the group center. Then, the Euclidean distance formula is used to calculate the positional conflict experienced by group a at time t.

[0099]

[0100]

[0101] Where n represents the number of members in the current group, Let represent the position coordinates of member i in the group at time t, and N represent the set of all groups in the scene. This represents the group center of group a at time t. Let d(a,b,t) represent the group center of group b at time t. The smaller d(a,b,t) is, the greater the positional conflict of group a.

[0102] Regarding the conflict of velocity directions, first calculate the vector sum of the velocity directions of all members in the group at time t as the velocity direction of the group:

[0103]

[0104] In the formula, Let the velocity direction of member i within the group at time t be represented. Then, the cosine function of the angle between the velocity directions of the group is used to quantify the conflict of all velocity directions experienced by group a:

[0105]

[0106] In the formula, and Let represent the direction vectors of groups a and b at time t, respectively. The larger or(a,b,t) is, the greater the velocity conflict experienced by group a.

[0107] Regarding the conflict of velocity magnitudes, firstly, the average velocity of all members within a group at time t is calculated as the velocity of that group. Then, a velocity difference function v(a,b,t) is defined to represent the conflict of velocity magnitudes of group a and group b at time t.

[0108]

[0109]

[0110] In the formula, n represents the number of members in the current group. This represents the speed of member i in the current group at time t. and Let vc(a,b,t) represent the velocity magnitudes of groups a and b at time t, respectively. The larger vc(a,b,t) is, the greater the velocity conflict experienced by group a.

[0111] Regarding group conflict, based on the definition of conflict mentioned above, the conflict of group a with respect to group b is defined as C(a,b,t), which is calculated as follows:

[0112]

[0113] In the above formula, W d +W o +W v =1, W d W o W v These represent the weights of positional conflict, velocity direction conflict, and velocity magnitude conflict, respectively. The larger C(a,b,t) is, the greater the conflict experienced by group a.

[0114] Crowd movement influenced by group attributes

[0115] This study analyzes the impact of stability and conflict on individuals within a group. During crowd movement, the motion state of individuals within a group is directly affected by intra-group stability and indirectly by inter-group conflict. This embodiment uses real data to calculate the velocity direction vectors of individuals affected by stability and conflict.

[0116] (1) Velocity direction vector of individual i under the influence of stability

[0117] In order to maintain stability within the group over a period of time, individual i will gradually keep its velocity direction close to the average velocity direction of the group.

[0118]

[0119] In the formula, Let α represent the velocity direction vector of individual i at time t. α is a stability control parameter used to represent the stability of the group to which the individual belongs during motion. Its value is determined by the function S in Definition 3; the larger α is, the higher the stability of the group.

[0120] (2) Velocity direction vector of individual i under the influence of conflict

[0121] To reduce the impact of conflict, individual i within the group will shift in the velocity direction.

[0122]

[0123] In the formula, Let represent the velocity direction vector of group a at time t. N represents the set of all groups in the scene. β is a conflict control parameter used to represent the degree of conflict experienced by an individual during movement. Its value is determined by function C in Definition 7. The larger the value of β, the higher the degree of conflict experienced by the individual.

[0124] (3) Velocity direction vector of individual i under the influence of group attributes

[0125] E i =w1×E1(i,t)+w2×E2(i,a,t) (15)

[0126] In the formula, E i Let w1 represent the velocity direction vector of individual i under the influence of group attributes, and w2 represent the weights of stability influence and conflict influence, respectively.

[0127] To further verify that the method in this embodiment can enhance the visual realism of the simulation results, the effectiveness of intra-group stability and inter-group conflict is verified in a synthetic scene, and the stability and conflict during the group movement process are visualized graphically.

[0128] (1) The effectiveness of stability

[0129] The proposed stability is evaluated using a self-driven particle model (SDP) because SDPs have been widely used to study the characteristics of particle motion and exhibit high similarity to various human population systems in nature. Synthetic scene SDPs can well reproduce the main observed characteristics of collective and self-ordered motion in a population. To evaluate the effectiveness of the stability of this embodiment, all particles are assumed to have randomized velocity magnitudes, and the direction of motion of each particle is equal to the average direction of motion of its neighboring particles under some random perturbations. The stability of the population is then measured within this hybrid human population system.

[0130] The effectiveness of the intra-group stability is evaluated in the SDP model, where particles are randomly distributed in a 25×25 region. The initial positions of the particles are initialized to make them appear as a group, and the average motion direction of the group is changed by setting the neighborhood radius. The effectiveness of the intra-group stability is evaluated from three cases, thus demonstrating the feasibility of the quantization method.

[0131] Case 1: Particles move randomly for a period of time, maintaining an irregular motion state. As shown in Figures 4(a), 4(b), and 4(c), initially the particles move randomly, but as time goes on, the particles continue to maintain an irregular motion state, and the group stability is minimized during this period.

[0132] Case 2: The particle moves randomly for a period of time, and then maintains an ordered motion state. As shown in Figures 5(a), 5(b), and 5(c), the initial position of the particle remains unchanged, the radius of the particle's neighborhood is changed, and the particle moves in an ordered manner over time, and the stability is calculated.

[0133] Case 3: The particle maintains an ordered motion state for a period of time. As shown in Figures 6(a), 6(b), and 6(c), the initial position of the particle remains unchanged, and the particle's neighborhood radius is changed to keep the particle in an ordered motion. Then, the stability is calculated.

[0134] In verifying intragroup stability using the SDP model, the highest stability was observed when particles maintained ordered motion, followed by stability when particles gradually became ordered from disordered motion, with the lowest stability occurring when particles remained in disordered motion. The results indicate that stability effectively measures the motion state of particles, and the proposed definition effectively quantifies group stability during population movement.

[0135] (2) Conflicting effectiveness

[0136] To improve the realism of the simulation and further verify the effectiveness of the inter-group conflict resolution, when using the SDP model to quantify the inter-group conflict resolution, the particles were divided into two groups to verify the inter-group conflict resolution in order to balance the computational complexity and reliability of the model. Based on the characteristics of real-world group conflict, the initial positions and orientations of the two groups were initialized, the magnitude of the inter-group conflict resolution was calculated, and the patterns of inter-group conflict resolution were analyzed. Thirty moving particles were set up, with each group divided into 15 particles. The SDP model was used to simulate the movement of a crowd, and the inter-group conflict resolution was analyzed. As shown in Figures 7(a) and 7(b), the results show that this definition effectively quantifies the inter-group conflict resolution.

[0137] In this crowd movement simulation system, a group of 30 people was simulated, divided into two groups of 10 and 20. Both groups maintained stability for a period during the movement, while individual movement directions changed due to inter-group conflicts. To demonstrate that the method in this embodiment simulates crowd movement more realistically, a comparison was made with existing methods. Figures 8(a), 8(b), 8(c), and 8(d) show the crowd simulation results of existing methods. The pedestrians in the red boxes did not change direction when affected by stability and conflict. Figures 9(a), 9(b), 9(c), and 9(d) show the crowd movement results simulated by the method in this embodiment. It can be observed that the pedestrians in the red boxes changed their movement direction when affected by stability and conflict, consistent with the movement patterns of pedestrians in real-world situations. The results indicate that the simulation method proposed in this embodiment is effective and can simulate crowd movement more realistically.

[0138] Example 2

[0139] Embodiment 2 of this disclosure introduces a data-driven crowd movement simulation system.

[0140] like Figure 10 The data-driven crowd movement simulation system shown includes:

[0141] The group information extraction module is configured to acquire crowd video data, divide the acquired crowd video data into crowd groups, and extract crowd motion attributes based on the divided crowd groups.

[0142] The group attribute quantification module is configured to perform group attribute quantification on the extracted crowd motion attributes to obtain the group attribute quantification results.

[0143] The crowd simulation module is configured to generate a simulation animation of crowd movement based on the obtained group attribute quantification results and the crowd simulation platform, thereby simulating the movement of the crowd.

[0144] The detailed steps are the same as those of the data-driven crowd movement simulation method provided in Example 1, and will not be repeated here.

[0145] Example 3

[0146] Embodiment 3 of this disclosure provides a computer-readable storage medium.

[0147] A computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the steps of the data-driven crowd movement simulation method as described in Embodiment 1 of this disclosure.

[0148] The detailed steps are the same as those of the data-driven crowd movement simulation method provided in Example 1, and will not be repeated here.

[0149] Example 4

[0150] Embodiment 4 of this disclosure provides an electronic device.

[0151] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the data-driven crowd movement simulation method as described in Embodiment 1 of this disclosure.

[0152] The detailed steps are the same as those of the data-driven crowd movement simulation method provided in Example 1, and will not be repeated here.

[0153] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A data-driven method for simulating crowd movement, characterized in that, Includes the following steps: Acquire crowd video data; The acquired video data of people is divided into groups. Extracting crowd movement attributes based on the segmented crowd groups; The extracted crowd motion attributes are quantified into group attributes to obtain the group attribute quantification results; the group attribute quantification includes at least intra-group stability and inter-group conflict. The inter-group conflict includes conflicts in the position of members between groups, conflicts in the magnitude of the speed of members between groups, and conflicts in the direction of the speed of members between groups; Regarding location conflict, first calculate in The average position of all members in each group in the time-space scenario is used as the group center, and then the Euclidean distance formula is used to calculate the group. exist Constant positional conflict: in, This indicates the number of members in the current group. Indicates group members exist Position coordinates at that moment This represents the set of all groups in the scene. Indicates in Time Groups The group center, Indicates in Time Groups The group center, The smaller the group The greater the positional conflict; Regarding the conflict of velocity directions, first calculate the conflict of velocity directions for all members within the group. The vector sum of the velocity directions at each instant is taken as the velocity direction of this group: In the formula, express Time group members The velocity direction is determined, and then the cosine function of the angle between the velocity directions of the group is used to quantize the group. All velocity directions subject to conflict: In the formula, and Representing groups and exist The direction vector at time, The larger the group The greater the speed conflict; Regarding the conflict of speed magnitudes, first calculate the speed of all members within the group. The average velocity at each moment is taken as the velocity of the group, and then a velocity difference function is defined. To represent a group Subject to group exist Conflict in the magnitude of velocity at any given moment: In the formula, This indicates the number of members in the current group. express Current group members speed magnitude, and Representing groups and exist The magnitude of the velocity at any given moment The larger the group The greater the speed conflict; Regarding group conflict, combining the previously mentioned definition of conflict, we can consider groups... The group Conflict is defined as The calculation is as follows: In the above formula, , , , These represent the weights for positional conflict, velocity direction conflict, and velocity magnitude conflict, respectively. The larger the group The greater the conflict, the stronger the potential for conflict. individual Velocity direction vector under the influence of group attributes In the formula, Represents an individual Velocity direction vector under the influence of group attributes , These represent the weights of the stability impact and the conflict impact, respectively. in For individuals The velocity direction vector under the influence of stability; For individuals The velocity direction vector under the influence of conflict; In the formula, express individual moment The velocity direction vector, It is a stability control parameter used to represent the stability of an individual's group during movement; In the formula, Indicates group exist The velocity direction vector at time t, This represents the set of all groups in the scene. It is a conflict control parameter used to represent the degree of conflict an individual experiences during movement; Based on the obtained group attribute quantification results and the crowd simulation platform, a simulation animation of crowd group movement is obtained, realizing the simulation of crowd group movement.

2. The data-driven crowd movement simulation method as described in claim 1, characterized in that, In the process of extracting crowd motion attributes, a target tracking and detection learning framework is used to sample and track the trajectory of the crowd after group division, so as to obtain the position and speed of the crowd, and use the position and speed as the crowd motion attributes.

3. The data-driven crowd movement simulation method as described in claim 2, characterized in that, The target tracking and detection learning includes a tracking module, a detection module, and a learning module.

4. The data-driven crowd movement simulation method as described in claim 1, characterized in that, The stability within a group includes the stability of the magnitude of the velocity of the group members and the stability of the direction of the velocity of the group members.

5. The data-driven crowd movement simulation method as described in claim 1, characterized in that, The effectiveness of the intra-group stability and inter-group conflict is verified in a synthetic scene. The stability and conflict during the movement of the crowd group are visualized graphically to obtain a simulation animation of the crowd group movement.

6. A data-driven crowd movement simulation system, characterized in that, include: The group information extraction module is configured to acquire crowd video data, divide the acquired crowd video data into crowd groups, and extract crowd motion attributes based on the divided crowd groups. The group attribute quantification module is configured to perform group attribute quantification on the extracted crowd motion attributes to obtain the group attribute quantification result; the group attribute quantification includes at least intra-group stability and inter-group conflict. The inter-group conflict includes conflicts in the position of members between groups, conflicts in the magnitude of the speed of members between groups, and conflicts in the direction of the speed of members between groups; Regarding location conflict, first calculate in The average position of all members in each group in the time-space scenario is used as the group center, and then the Euclidean distance formula is used to calculate the group. exist Constant positional conflict: in, This indicates the number of members in the current group. Indicates group members exist Position coordinates at that moment This represents the set of all groups in the scene. Indicates in Time Groups The group center, Indicates in Time Groups The group center, The smaller the group The greater the positional conflict; Regarding the conflict of velocity directions, first calculate the conflict of velocity directions for all members within the group. The vector sum of the velocity directions at each instant is taken as the velocity direction of this group: In the formula, express Time group members The velocity direction is determined, and then the cosine function of the angle between the velocity directions of the group is used to quantize the group. All velocity directions subject to conflict: In the formula, and Representing groups and exist The direction vector at time, The larger the group The greater the speed conflict; Regarding the conflict of speed magnitudes, first calculate the speed of all members within the group. The average velocity at each moment is taken as the velocity of the group, and then a velocity difference function is defined. To represent a group Subject to group exist Conflict in the magnitude of velocity at any given moment: In the formula, This indicates the number of members in the current group. express Current group members speed magnitude, and Representing groups and exist The magnitude of the velocity at any given moment The larger the group The greater the speed conflict; Regarding group conflict, combining the previously mentioned definition of conflict, we can consider groups... The group Conflict is defined as The calculation is as follows: In the above formula, , , , These represent the weights for positional conflict, velocity direction conflict, and velocity magnitude conflict, respectively. The larger the group The greater the conflict, the stronger the potential for conflict. individual Velocity direction vector under the influence of group attributes In the formula, Represents an individual Velocity direction vector under the influence of group attributes , These represent the weights of the stability impact and the conflict impact, respectively. in For individuals The velocity direction vector under the influence of stability; For individuals The velocity direction vector under the influence of conflict; In the formula, express individual moment The velocity direction vector, It is a stability control parameter used to represent the stability of an individual's group during movement; In the formula, Indicates group exist The velocity direction vector at time t, This represents the set of all groups in the scene. It is a conflict control parameter used to represent the degree of conflict an individual experiences during movement; The crowd simulation module is configured to generate a simulation animation of crowd movement based on the obtained group attribute quantification results and the crowd simulation platform, thereby simulating the movement of the crowd.

7. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the data-driven crowd movement simulation method as described in any one of claims 1-5.

8. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the data-driven crowd movement simulation method as described in any one of claims 1-5.

Images