Panic pedestrian accelerated running behavior identification method based on dynamic centroid model

Abstract

The invention relates to a method for identifying accelerated running behaviors of panic pedestrians based on a dynamic centroid model. The method comprises the following steps: acquiring panic pedestrian images frame by frame from a video; establishing a coordinate system according to the panic pedestrian image, and constructing a dynamic centroid model of the panic pedestrian; and based on the dynamic centroid model, judging whether the panic pedestrian runs in an accelerated manner or not by utilizing a geometric judgment formula and a dynamic judgment formula. Compared with the prior art, the distinguishing method has the advantages of being convenient to calculate and high in robustness.

Description

Technical Field

[0001] The present invention relates to the technical field of crowd flow stability analysis, and in particular to a method for identifying panic pedestrian acceleration running behavior based on a dynamic centroid model. Background Art

[0002] Sudden safety accidents in public places can easily cause panic pedestrians to panic, and panic pedestrians are usually accompanied by accelerated running behavior. The identification of this behavior is of great significance to public safety, and has gradually become a hot topic for scholars in the field of public safety in recent years. At present, most studies related to panic pedestrians include panic pedestrians' panic behavior as abnormal behavior, and identify multiple abnormal behaviors such as panic behavior at the same time, while there are few studies specifically targeting the accelerated running behavior of panic pedestrians. In the prior art, there is a method that uses optical flow features to extract regions of interest (ROI) from video frames, and then uses convolutional neural networks as classifiers to detect runners, but this method is easily interfered by factors such as light and background. There is also a method that first extracts the runner's foot image, and identifies the running posture by analyzing the spatial, temporal, and frequency characteristics of the two feet when running, but this method does not involve the movement pattern of the upper limbs under the running posture.

[0003] At present, there are several deficiencies in the research on recognition of panic pedestrians’ accelerated running behavior:

[0004] 1. Lack of separate recognition of panicked pedestrians’ accelerated running behavior;

[0005] 2. There is a lack of modeling and analysis of the posture of panicked pedestrians when they accelerate. Summary of the invention

[0006] The purpose of the present invention is to provide a method for identifying the accelerated running behavior of panicked pedestrians based on a dynamic centroid model. The dynamic centroid is used to describe the characteristics of the accelerated running behavior, and the motion discriminant is used to judge the accelerated running behavior of panicked pedestrians, which has the advantages of easy calculation and strong robustness.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model, the method comprising:

[0009] Obtain panic pedestrian images frame by frame from the video;

[0010] Establish a coordinate system based on the panic pedestrian image and construct a dynamic centroid model of the panic pedestrian;

[0011] Based on the dynamic centroid model, the geometric discriminant formula and the dynamic discriminant formula are used to determine whether the panicking pedestrian is running at an accelerated speed.

[0012] Among them, the geometric discriminant formula is:

[0013]

[0014] β1 is the angle between the segment connecting the left hip and the left knee and the segment connecting the left knee and the left ankle, β2 is the angle between the segment connecting the right hip and the right knee and the segment connecting the right knee and the right ankle;

[0015] The dynamic discriminant formula is:

[0016]

[0017] x(DC) represents the horizontal coordinate value of the dynamic center of mass of the panicking pedestrian, x(SC) represents the horizontal coordinate value of the static center of mass of the panicking pedestrian, dir(DC) represents the direction of the dynamic center of mass of the panicking pedestrian, Indicates the acceleration direction of panicked pedestrians, Indicates the direction of panicked pedestrian speed.

[0018] Furthermore, the process of constructing the dynamic centroid model of the panicking pedestrian includes:

[0019] The pose estimation algorithm is used to extract the 2D skeleton key points of the panicking pedestrian from the panicking pedestrian image;

[0020] Through bipartite graph matching, the 2D skeleton key points of the panicking pedestrian are connected in segments according to the limb connection relationship;

[0021] The coordinates of the panicking pedestrian's 2D skeleton key points are used to calculate the dynamic center of mass coordinates of the panicking pedestrian.

[0022] Furthermore, the 2D skeleton key points include: top of the head, left ear, right ear, left eye, right eye, nose, left corner of the mouth, right corner of the mouth, neck, left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left ankle and right ankle.

[0023] Furthermore, the calculation process of the dynamic center of mass coordinates of the panicking pedestrian includes:

[0024] All 2D skeleton key points of the same panic pedestrian face in the panic pedestrian image are processed into one head node through calculation and marked as 0;

[0025] Calculate the dynamic centroid coordinates of different segments respectively;

[0026] The dynamic center of mass coordinates of panicking pedestrians are calculated using the dynamic center of mass coordinates of the segments.

[0027] Furthermore, the calculation formula of the head node is:

[0028]

[0029] Among them, i is the number of the facial 2D skeleton key point, (x i ,y i ) are the 2D bone key point coordinates.

[0030] Furthermore, the calculation formula of the dynamic center of mass of different segments is:

[0031]

[0032] Among them, k is the number of segments, p is the number of key points at the proximal end of the segment, d is the number of key points at the distal end of the segment, (x k ,y k ) is the segment dynamic centroid coordinate, (x p ,y p ) is the coordinate of the key point at the proximal end of the segment, (x d ,y d ) is the coordinate of the key point at the distal end of the segment, l p is the dynamic centroid coefficient of the initial segment, l d is the dynamic center of mass coefficient of the terminal segment.

[0033] Furthermore, the proximal key point of the segment is the 2D bone key point closer to the human torso among the two 2D bone key points connected by a segment, and the distal key point of the segment is the 2D bone key point farther away from the human torso among the two 2D bone key points connected by a segment. The dynamic center of mass coefficient of the initial segment and the dynamic center of mass coefficient of the terminal segment are calculated based on anatomical knowledge and statistical data, and are used to reflect the contribution of different body parts to the dynamic center of mass of panicked pedestrians.

[0034] Furthermore, the calculation formula of the dynamic center of mass of the panicking pedestrian is:

[0035]

[0036] Among them, (x k ,y k ) is the dynamic centroid coordinate of the segment, m k is the body mass of the panicking pedestrian in the kth segment, and M is the total body mass of the panicking pedestrian.

[0037] An electronic identification device, comprising:

[0038] one or more processors;

[0039] Memory;

[0040] One or more programs stored in a memory, wherein the one or more programs include instructions for executing the panic pedestrian acceleration running behavior recognition method based on a dynamic center of mass model as claimed in any one of claims 1 to 9.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The present invention extracts panic pedestrian images from video images and establishes a dynamic centroid model, thereby realizing the individual recognition of the panic pedestrian's accelerated running behavior. When judging whether the panic pedestrian is running, the present invention uses a geometric discriminant formula and a dynamic discriminant formula. The geometric discriminant formula judges the panic pedestrian's body posture according to the angle between the panic pedestrian's left thigh and left calf and the angle between the right thigh and right calf. The dynamic discriminant formula further analyzes whether the panic pedestrian is accelerating running from the dynamic centroid characteristics of the panic pedestrian by comparing the panic pedestrian's dynamic centroid direction with the pedestrian's speed direction, acceleration direction, and center of gravity direction. Finally, the panic pedestrian's accelerated running behavior is comprehensively judged. This method has the advantages of easy calculation and strong robustness.

[0043] 2. The present invention connects 2D skeleton key points in segments when establishing a dynamic center of mass model. The present invention models and analyzes the posture of panicked pedestrians when they accelerate. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Flow chart of the method of the present invention

[0045] Figure 2 This is an example diagram of key points of human skeleton extracted from an image of a panicked pedestrian in an embodiment of the present invention;

[0046] Figure 3 A comparison diagram of the static center of mass and the dynamic center of mass of a panic pedestrian's general walking posture, pocket-inserting walking posture, and running posture in an embodiment of the present invention;

[0047] Figure 4 The coordinate diagram of the dynamic center of mass and the static center of mass of a panic pedestrian running in an embodiment of the present invention;

[0048] Figure 5 This is a center of mass trajectory diagram of the accelerated running behavior according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0050] This embodiment discloses a method for identifying panic pedestrians accelerating running behavior based on a dynamic centroid model. Figure 1 As shown, the main steps include:

[0051] S1. Obtain panic pedestrian images frame by frame from the video;

[0052] S2. Establish a coordinate system based on the panic pedestrian image and construct a dynamic centroid model of the panic pedestrian;

[0053] S3. Based on the dynamic center of mass model, use the geometric discriminant formula and the dynamic discriminant formula to determine whether the panicking pedestrian is running at an accelerated speed.

[0054] The steps of the identification method in this embodiment are described in detail below.

[0055] In this embodiment, the video of panicked pedestrians running faster is captured and observed in an open space as the verification scene video acquisition frequency f collection =29.97fps (Frames Per Second). For ease of calculation, f is used in the following text. collection =30fps, the single frame resolution of the video is 1920*1080 pixels, and MATLAB is used to capture video frame images with a capture frequency of FrameRate=30.

[0056] In this embodiment, the upper left corner of the video frame image is defined as the coordinate origin, the horizontal axis is the x-axis, the horizontal axis value represents 1920 pixels, and the vertical axis is the y-axis, the vertical axis value represents 1080 pixels.

[0057] Furthermore, this embodiment uses the posture estimation algorithm Openpose to extract the 2D skeleton key point coordinates of the panicking pedestrian from the video image. Openpose first estimates the feature map of each joint point of the human body through multiple convolutional layers, pooling layers and fully connected layers, and then connects the limbs of the same person in segments according to the limb connection relationship through bipartite graph matching to determine the connection relationship between the key points. Figure 2 As shown, the number of extracted 2D skeleton key points is 21, including: top of the head, left ear, right ear, left eye, right eye, nose, left corner of the mouth, right corner of the mouth, neck, left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left ankle, and right ankle.

[0058] In this embodiment, the process of constructing the dynamic centroid model of the panicking pedestrian includes:

[0059] The pose estimation algorithm is used to extract the 2D skeleton key points of the panicking pedestrian from the panicking pedestrian image;

[0060] Through bipartite graph matching, the 2D skeleton key points of the panicking pedestrian are connected in segments according to the limb connection relationship;

[0061] The coordinates of the panicking pedestrian's 2D skeleton key points are used to calculate the dynamic center of mass coordinates of the panicking pedestrian.

[0062] Furthermore, the coordinates of the dynamic center of mass of the panicking pedestrian are calculated as follows:

[0063] The eight 2D skeleton key points (serial numbers 14-21) of the face of the same panicking pedestrian in the panicking pedestrian image are processed into one head node through calculation and marked as 0.

[0064] The calculation formula of the head node is:

[0065]

[0066] Among them, i is the number of the facial 2D skeleton key point, (x i ,y i ) are the 2D bone key point coordinates.

[0067] The dynamic centroid coordinates of different segments are calculated separately, and the formula is as follows:

[0068]

[0069] Among them, k is the number of segments, p is the number of key points at the proximal end of the segment, d is the number of key points at the distal end of the segment, (x k ,y k ) is the segment dynamic centroid coordinate, (x p ,y p ) is the coordinate of the key point at the proximal end of the segment, (x d ,y d ) is the coordinate of the key point at the distal end of the segment, l p is the dynamic centroid coefficient of the initial segment, l d is the dynamic center of mass coefficient of the terminal segment.

[0070] The proximal key point of the segment is the 2D bone key point closer to the human torso among the two 2D bone key points connected by a segment, and the distal key point of the segment is the 2D bone key point farther away from the human torso among the two 2D bone key points connected by a segment. The dynamic center of mass coefficient of the initial segment and the dynamic center of mass coefficient of the terminal segment are calculated based on anatomical knowledge and statistical data, and are used to reflect the contribution of different body parts to the dynamic center of mass of panicked pedestrians.

[0071] The dynamic centroid coordinates of the panicking pedestrian are calculated using the segment dynamic centroid coordinates. The formula is as follows:

[0072]

[0073] Among them, (x k ,y k) is the dynamic centroid coordinate of the segment, m k is the body mass of the panicking pedestrian in the kth segment, and M is the total body mass of the panicking pedestrian.

[0074] Furthermore, the accelerated running behavior is identified. The static center of mass of the panic pedestrian is the center of mass of the panic pedestrian in a static state. The continuous frames of the static center of mass and the dynamic center of mass of the running posture are compared with the general walking posture and the walking posture with the hands in pockets, such as Figure 3 As shown in the figure, P D.C is the dynamic center of mass, P S.C It is the static center of mass. It can be found that the positions of the static center of mass and the dynamic center of mass of the general walking posture and the walking posture with hands in pockets basically coincide, while the position of the dynamic center of mass in the running posture is biased towards the direction of running.

[0075] Therefore, by combining the geometric characteristics of running posture and walking posture, the geometric discriminant formula and dynamic discriminant formula of running posture, general walking posture and walking posture with hands in pockets are obtained. The obtained 2D skeleton key points, segment angles and dynamic center of mass data are substituted into the following geometric discriminant formula and dynamic discriminant formula:

[0076]

[0077] Among them, Figure 4 As shown, β1 is the angle between the segment connecting the left hip and the left knee and the segment connecting the left knee and the left ankle (segments 8-9 and 9-10), β2 is the angle between the segment connecting the right hip and the right knee and the segment connecting the right knee and the right ankle (segments 11-12 and 12-13), x(DC) represents the horizontal coordinate value of the dynamic center of mass of the panic pedestrian, x(SC) represents the horizontal coordinate value of the static center of mass of the panic pedestrian, dir(DC) represents the direction of the dynamic center of mass of the panic pedestrian, Indicates the acceleration direction of panicked pedestrians, Indicates the direction of panicked pedestrian speed.

[0078] If the 2D skeleton key points, segment angles and dynamic center of mass data satisfy the geometric discriminant formula and the dynamic discriminant formula, it means that the panicked pedestrian is running faster.

[0079] The center of mass trajectory of a panicked pedestrian when he accelerates is as follows Figure 5 As shown, the upper curve represents the dynamic center of mass of the panic pedestrian running faster, and the lower curve represents the static center of mass of the panic pedestrian running faster. The acceleration process lasts for two seconds, and a total of 62 frames of images are extracted.

[0080] According to the comparison between the dynamic center of mass and the static center of mass of the panicked pedestrian who is running at an accelerated speed, it can be seen that the dynamic center of mass of the panicked pedestrian has a gentle platform at the peak, while the static center of mass of the panicked pedestrian has no platform and is sharper at the peak. Secondly, in terms of the time meaning of the dynamic center of mass and the static center of mass of the panicked pedestrian, the dynamic center of mass of the panicked pedestrian has an advance amount compared with the static center of mass. Taking the 20th frame as an example, the advance amount of the dynamic center of mass compared with the static center of mass is one frame time difference, that is, 0.033 seconds (1 / 30). Then, the peak time of the dynamic center of mass of the panicked pedestrian appears earlier than the static center of mass. Taking the third peak platform in the figure as an example (from left to right), the dynamic center of mass of the panicked pedestrian reaches the quasi-peak value at the 20th frame, and the static center of mass of the panicked pedestrian reaches the peak value at the 22nd frame. The dynamic center of mass is two sampling cycles ahead of the static center of mass, that is, 0.067 seconds (2 / 30). Therefore, the dynamic center of mass curve can lead the static center of mass curve in identifying the accelerated running behavior and show the characteristics of accelerated running.

[0081] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model, characterized in that: The method comprises: Obtain panic pedestrian images frame by frame from the video; Establish a coordinate system based on the panic pedestrian image and construct a dynamic centroid model of the panic pedestrian; Based on the dynamic centroid model, the geometric discriminant formula and the dynamic discriminant formula are used to determine whether the panicking pedestrian is running at an accelerated speed. Among them, the geometric discriminant formula is: β1 is the angle between the segment connecting the left hip and the left knee and the segment connecting the left knee and the left ankle, β2 is the angle between the segment connecting the right hip and the right knee and the segment connecting the right knee and the right ankle; The dynamic discriminant formula is: x(DC) represents the horizontal coordinate value of the dynamic center of mass of the panicking pedestrian, x(SC) represents the horizontal coordinate value of the static center of mass of the panicking pedestrian, dir(DC) represents the direction of the dynamic center of mass of the panicking pedestrian, Indicates the acceleration direction of panicked pedestrians, Indicates the direction of panicked pedestrian speed.

2. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 1, characterized in that: The construction process of the dynamic center of mass model of the panic pedestrian includes: The pose estimation algorithm is used to extract the 2D skeleton key points of the panicking pedestrian from the panicking pedestrian image; Through bipartite graph matching, the 2D skeleton key points of the panicking pedestrian are connected in segments according to the limb connection relationship; The coordinates of the panicking pedestrian's 2D skeleton key points are used to calculate the dynamic center of mass coordinates of the panicking pedestrian.

3. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 2 is characterized in that: The posture estimation algorithm extracts the key points of the 2D skeleton of the panicking pedestrian by: Input panic pedestrian images and coordinate systems, and identify the specific positions of each panic pedestrian in the image; The key points of the panicking pedestrian’s body joints are calculated through convolutional layers, pooling layers, and fully connected layers; The key points are numbered and output as the 2D skeleton key points of the panicking pedestrian.

4. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 2 is characterized in that: The 2D bone key points include: top of the head, left ear, right ear, left eye, right eye, nose, left corner of the mouth, right corner of the mouth, neck, left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left ankle and right ankle.

5. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 2, characterized in that: The calculation process of the dynamic center of mass coordinates of the panicking pedestrian includes: All 2D skeleton key points of the same panic pedestrian face in the panic pedestrian image are processed into one head node through calculation and marked as 0; Calculate the dynamic centroid coordinates of different segments respectively; The dynamic center of mass coordinates of panicking pedestrians are calculated using the dynamic center of mass coordinates of the segments.

6. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 5, characterized in that: The calculation formula of the head node is: Among them, i is the number of the facial 2D skeleton key point, (x i ,y i ) are the 2D bone key point coordinates.

7. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 5, characterized in that: The calculation formula of the dynamic center of mass of the different segments is: Among them, k is the number of segments, p is the number of key points at the proximal end of the segment, d is the number of key points at the distal end of the segment, (x k ,y k ) is the segment dynamic centroid coordinate, (x p ,y p ) is the coordinate of the key point at the proximal end of the segment, (x d ,y d ) is the coordinate of the key point at the distal end of the segment, l p is the dynamic centroid coefficient of the initial segment, l d is the dynamic center of mass coefficient of the terminal segment.

8. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 7, characterized in that: The proximal key point of the segment is the 2D bone key point closer to the human torso among the two 2D bone key points connected by a segment, and the distal key point of the segment is the 2D bone key point farther away from the human torso among the two 2D bone key points connected by a segment. The dynamic center of mass coefficient of the initial segment and the dynamic center of mass coefficient of the terminal segment are calculated based on anatomical knowledge and statistical data, and are used to reflect the contribution of different body parts to the dynamic center of mass of panicked pedestrians.

9. The method for identifying panicked pedestrians' accelerated running behavior based on a dynamic centroid model according to claim 5, characterized in that: The calculation formula of the dynamic center of mass of the panic pedestrian is: Among them, (x k ,y k ) is the dynamic centroid coordinate of the segment, m k is the body mass of the panicking pedestrian in the kth segment, and M is the total body mass of the panicking pedestrian.

10. An electronic identification device, characterized in that: include: one or more processors; Memory; One or more programs stored in a memory, wherein the one or more programs include instructions for executing the panic pedestrian acceleration running behavior recognition method based on a dynamic center of mass model as claimed in any one of claims 1 to 9.

Images