Intelligent detection algorithm for abnormal state of passenger in toilet of high-speed train

Abstract

The application discloses an intelligent detection algorithm for abnormal state of passengers in a high-speed train toilet, and comprises the following steps: S1, installing a three-dimensional laser radar and an infrared body temperature sensor in the high-speed train toilet, wherein the three-dimensional laser radar is arranged on the top of the train compartment; S2, scanning the train compartment of the high-speed train toilet by the three-dimensional laser radar when the toilet is empty, and obtaining an initial background point cloud diagram at this time; S3, obtaining a to-be-identified point cloud diagram by scanning the state of passengers in the high-speed train toilet by the three-dimensional laser radar; S4, comparing the to-be-identified point cloud diagram with the initial background point cloud, screening out points with the same point cloud number position distance in the to-be-identified point cloud diagram and the initial background point cloud diagram, and deleting the points, and the point cloud diagram composed of the remaining points is a body state point cloud diagram of passengers in the toilet compartment; and S5, judging whether the passengers are in a normal state or an abnormal state according to the body state point cloud diagram of the passengers, and if the passengers are in the abnormal state, the three-dimensional laser radar automatically alarms.

Description

Technical Field

[0001] This invention relates to the field of rail transit, and in particular to an intelligent detection algorithm for abnormal status of people in the toilets of high-speed trains. Background Technology

[0002] High-speed trains are a major mode of transportation for travelers, and safety is a primary concern during the journey. Due to the high speed of the train, any harmful gases produced inside will spread throughout the carriage. The harmful gases produced by smoking are often more toxic than the smoke itself. For the safety of other passengers, smoking is prohibited on high-speed trains. Currently, there are no monitoring devices in the toilets of high-speed trains, so smoking in these areas would go undetected. Furthermore, people with health conditions, such as fever or mobility issues, are more prone to accidents in the toilets. Even healthy passengers could accidentally fall, faint against a wall, or fall onto the toilet, and without timely assistance, there is a risk to passenger safety. Currently, 3D LiDAR has broad application prospects in digital cities, autonomous driving, virtual reality, human target recognition in complex scenes, and obstacle recognition. It segments the environment and people by processing LiDAR point clouds. Common LiDAR point cloud processing methods include grid-based methods and methods that directly process in 3D space based on existing basic segmentation algorithms. Common methods include edge detection-based segmentation, region growing-based segmentation, feature clustering-based segmentation, and deep learning-based segmentation. Research on 3D LiDAR for human target recognition is currently limited to the "recognition" stage, and the aforementioned methods are relatively complex and unsuitable for simple, specific scenarios. There are no research reports on the detection of abnormal personnel in high-speed train restrooms based on 3D LiDAR. Summary of the Invention

[0003] The technical problem to be solved by this invention is to provide an intelligent detection algorithm for abnormal status of people in the toilets of high-speed trains. By installing three-dimensional lidar and infrared body temperature sensors in the toilets of high-speed trains, the algorithm can automatically identify abnormal status of people through the background algorithm, so as to ensure that passengers are safer during their journey on high-speed trains.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an intelligent detection algorithm for abnormal status of personnel in the toilets of high-speed trains, comprising the following steps:

[0005] Step S1: Install a three-dimensional lidar and an infrared body temperature sensor qualitatively connected to the three-dimensional lidar in the toilet of the EMU. The three-dimensional lidar is set on the top of the carriage.

[0006] Step S2: When the toilet in the EMU is unoccupied, use a 3D LiDAR to scan the toilet carriage of the EMU to obtain the initial background point cloud map.

[0007] Step S3: Use 3D LiDAR to scan the status of people in the train's restroom to obtain a point cloud map to be identified;

[0008] Step S4: Compare the point cloud image to be identified with the initial background point cloud, filter out the points whose point cloud numbers are equidistant from the points in the point cloud image to be identified and delete them, and the remaining points form the point cloud image of the body state of the people in the toilet carriage.

[0009] Step S5: Determine whether the person is in a normal or abnormal state based on the point cloud map of their body condition. If the person is determined to be in an abnormal state, the 3D LiDAR will automatically sound an alarm.

[0010] A further improvement to the technical solution of the present invention is that: in step S1, if the temperature inside the infrared body temperature sensor is set to be higher than 50°C, it is determined that there is open flame smoking; and a smoke alarm is also installed in the toilet of the EMU.

[0011] A further improvement to the technical solution of the present invention is that: the initial background point cloud in step S2 is denoted as A. i (x i ,y i ,z i ), where i represents the initial background point cloud number, which is 1, 2, 3, ..., M, and is stored in the set S1(A1, A2, ..., A...). M ).

[0012] A further improvement to the technical solution of the present invention is that: the point cloud to be identified in step S3 is B. nj (x j ,y j ,z j ), where j represents the number of the point cloud to be identified, namely 1, 2, 3, ..., N, N = M, and the point cloud image of a certain ring to be identified is stored in set S. n2 (B1,B2…B N The point clouds of all rings are stored in set S(S). 12 ,S 22 …S n2 ), n = 1, 2, 3...n p n represents the number of scan cycles, n p This indicates the total number of scan cycles.

[0013] A further improvement to the technical solution of the present invention is that: in step S4, the points whose positional distance between the point cloud number in the point cloud image to be identified and the point cloud image in the initial background point cloud image are Δx=|x j -x i |=α1、Δy=|yj -y i |=β1、Δz=|z j -z i |=γ1 and its α1, β1, γ1 are all constants tending to 0, where i=j, the point cloud of the body state of the person is stored in the set B(B1,B2…B n ).

[0014] A further improvement to the technical solution of the present invention is that the specific process in step S5 is as follows:

[0015] Step S51: Screen the point cloud set B(B1,B2…B) of the physical status of the personnel. n minz for each ring in ) j Calculate the height h of a person with their head upright. r =h cd -min{minz 1j ,minz 2j ,…,minz nj}, where h cd The height of the 3D LiDAR above the vehicle floor;

[0016] Step S52: Determine if the person is in a normal state: If the nth ring in the person's physical state point cloud satisfies h r -η≤h cd -minz nj ≤h cd η is half the length of the head, at which point the person is considered to be in a normal state;

[0017] Step S53: Determine if the person is in an abnormal lying state: If the n1st ring of the person's body state point cloud satisfies n1∈n, ξ is the thickness of the side of the human body, and in this ring Small fluctuation value, i.e., variance If μ is a constant approaching 0, then the person is determined to be lying flat on the ground. Data points for the n1+hσ ring are selected, where σ represents the ring number and h represents a positive integer. Within a continuous time range ε1, the n1+hσ ring simultaneously satisfies... If all of these conditions are met, then the person is considered to be experiencing an abnormality while lying flat.

[0018] Step S54: Determine if the person is in an abnormal upright state: Filter the minimum height value of the n2th ring in the person's body state point cloud. and maximum height value n2∈n, the point where the minimum height value is located is denoted as . The point where the maximum height value is located is denoted as like If both α² and β² are constants approaching 0, then the person is determined to be in an upright position. Based on finding point P, select a point O within the ring; this point has the largest coordinates along the x, y, and z axes, denoted as . like If δ is a constant approaching -1, then we assume the person is sitting on the floor of the carriage with their back against the carriage wall. Selecting data points in the n²+hσ ring, the n²+hσ ring within a continuous time interval ε² satisfies: If this person is sitting on the floor of the carriage with their back against the carriage wall, it is determined that this person is in an abnormal upright state.

[0019] Step S55: Determine if the person is in an abnormal backward-falling state: Filter the minimum height value of the n3rd ring in the person's body state point cloud. If n3∈n, if If the distance between γ1 and γ2 is the distance above the buttocks, then the person is squatting on the toilet. Points within the range, h dmt This represents the height of the laser radar reaching the scanning point when it scans a human body, where γ3 is the length of the person's head. The slope is obtained by fitting a straight line formed by points within this range using the least squares method. like If λ is 1, then it is assumed that the person is lying face down on the toilet, and the n3+hσ-th cycle within the continuous time range of ε3 satisfies: slope of points within the range This indicates that the person is exhibiting abnormal backward tilting.

[0020] The technological advancements achieved by this invention due to the adoption of the above technical solutions are as follows:

[0021] 1. This invention installs a three-dimensional lidar and an infrared body temperature sensor in the toilet of a high-speed train. Through a background algorithm, it realizes the automatic identification of abnormal status of personnel, ensuring greater safety for passengers during their journey on high-speed trains.

[0022] 2. This invention uses three-dimensional lidar to collect key feature points and detect the posture of personnel to determine whether there are abnormal situations such as falling to the ground, fainting, or lying on the toilet, thereby facilitating timely rescue of personnel;

[0023] 3. This invention uses an infrared body temperature sensor in the toilet of a high-speed train to monitor the body temperature of passengers and determine whether they are smoking. Attached Figure Description

[0024] Figure 1 This is a flowchart of steps S51-S53 of the present invention;

[0025] Figure 2 This is a flowchart of step S54 of the present invention;

[0026] Figure 3 This is a point cloud image of a person sitting on the floor of the carriage with their back against the carriage wall in step S54 of the present invention.

[0027] Figure 4 This is a flowchart of step S55 of the present invention;

[0028] Figure 5 This is a point cloud diagram of a person squatting on a toilet in step S55 of the present invention. Detailed Implementation

[0029] The present invention will be further described in detail below with reference to embodiments:

[0030] An intelligent detection algorithm for abnormal status of passengers in the restrooms of high-speed trains includes the following steps:

[0031] Step S1: Install a three-dimensional lidar and an infrared body temperature sensor qualitatively connected to the three-dimensional lidar in the train's restroom. The three-dimensional lidar is installed on the ceiling of the carriage. The infrared body temperature sensor is set to detect smoking with an open flame if the temperature exceeds 50°C. Since the infrared body temperature sensor can only detect smoking with an open flame to a certain extent, a smoke alarm can be added to the restroom to assist in distinguishing between smoking with an open flame and e-cigarettes.

[0032] Step S2: When the toilet in the high-speed train is unoccupied, use a 3D LiDAR to scan the toilet carriage of the high-speed train. At this time, the initial background point cloud map is obtained; the initial background point cloud is denoted as A. i (x i ,y i ,z i ), where i represents the initial background point cloud number, which is 1, 2, 3, ..., M, and is stored in the set S1(A1, A2, ..., A...). M ).

[0033] Step S3: Use a 3D LiDAR scanner to scan the status of people in the train's restroom to obtain a point cloud image to be identified. The point cloud to be identified is B. nj (x j ,y j ,z j ), where j represents the number of the point cloud to be identified, namely 1, 2, 3, ..., N. Since the number of points scanned by the lidar is the same in any situation, the number of point clouds in the point cloud to be identified is the same as the number of point clouds in the initial background point cloud. N = M. The point cloud to be identified for a certain ring is stored in set S. n2 (B1,B2…B N The point clouds of all rings are stored in set S(S). 12 ,S 22 …S n2 ), n = 1, 2, 3...n p n represents the number of scan cycles, np This indicates the total number of scan cycles.

[0034] Step S4: Compare the point cloud image to be identified with the initial background point cloud. Filter out points whose point cloud numbers are equidistant from each other and delete them. The points whose point cloud numbers are equidistant from each other are Δx=|x j -x i |=α1、Δy=|y j -y i |=β1、Δz=|z j -z i |=γ1 and its α1, β1, and γ1 are all constants approaching 0, where i=j, the remaining points form the point cloud map of the body state of the people in the toilet carriage. The point cloud of the people's body state is stored in the set B(B1,B2…B n ).

[0035] Step S5: Determine whether the person is in a normal or abnormal state based on the point cloud map of their body condition. If the person is determined to be in an abnormal state, the 3D LiDAR will automatically sound an alarm.

[0036] The specific process of determining whether a person is in a normal or abnormal state based on the point cloud map of their physical condition is as follows:

[0037] Step S51: Screen the point cloud set B(B1,B2…B) of the physical status of the personnel. n minz for each ring in ) j Calculate the height h of a person with their head upright. r =h cd -min{minz 1j ,minz 2j ,…,minz nj}, where h cd The height of the 3D LiDAR above the vehicle floor;

[0038] Step S52: Determine if the person is in a normal state: If the nth ring in the person's physical state point cloud satisfies h r -η≤h cd -minz nj ≤h cd Since a person in the restroom may be walking, washing their hands, or excreting, their head may be in an upright or bowed position. Therefore, η is half the length of the head, and the person is considered to be in a normal state.

[0039] Step S53: Determine if the person is in an abnormal lying state: If the n1st ring of the person's body state point cloud satisfies n1∈n, ξ is the thickness of the side of the human body. Since it is only necessary to determine if the person's entire body has fainted, the following applies to this ring. Small fluctuation value, i.e., variance If μ is a constant approaching 0, then the person is determined to be lying flat on the ground. Data points for the n1+hσ ring are selected, where σ represents the ring number and h represents a positive integer. Within a continuous time range ε1, the n1+hσ ring simultaneously satisfies... If all of these conditions are met, then the person is considered to be experiencing an abnormality while lying flat.

[0040] Step S54: Determine if the person is in an abnormal upright state: such as Figure 2 As shown, the minimum height value of the n2th ring in the point cloud of the body status of the screening personnel. and maximum height value n2∈n, the point where the minimum height value is located is denoted as . The point where the maximum height value is located is denoted as like If both α² and β² are constants approaching 0, then the person is determined to be in an upright position. Based on finding point P, select a point O within the ring; this point has the largest coordinates along the x, y, and z axes, denoted as . like If δ is a constant approaching -1, then we assume the person is sitting on the floor of the carriage with their back against the carriage wall. Selecting data points in the n²+hσ ring, the n²+hσ ring within a continuous time interval ε² satisfies: like Figure 3 As shown, this person is sitting on the floor of the carriage with their back against the carriage wall, indicating that this person is in an abnormal upright state.

[0041] Step S55: Determine if the person is in an abnormal backward-falling state: such as Figure 4 As shown: The minimum height value of the n3rd ring in the point cloud of the body status of the personnel being screened. If n3∈n, if The distance between γ1 and γ2 represents the distance above the buttocks of the person. Therefore, if the person is squatting on a toilet, the body can be considered a straight line. Figure 5 As shown, in the plane formed by the X and Z axes, LR is used as the head of the person, LF as the upper body, and EF as the legs. (Filtering) Points within the range, h dmt This represents the height of the laser radar reaching the scanning point when it scans a human body. γ3 is the length of the person's head, a constant; generally, the head is 1 / 8 of the height, which can be used as a reference for selection. The slope is obtained by fitting a straight line formed by points within this range using the least squares method. like If a person lies supine on the toilet, their head and upper body cannot be in a straight line, so λ is 1. Therefore, it is assumed that the person lies supine on the toilet, and the n3+hσ-th cycle within the continuous time range of ε3 satisfies: slope of points within the range This indicates that the person is exhibiting abnormal backward tilting.

Claims

1. An intelligent detection algorithm for abnormal status of passengers in the restrooms of high-speed trains, characterized in that: Includes the following steps: Step S1: Install a three-dimensional lidar and an infrared body temperature sensor in the toilet of the EMU train. The three-dimensional lidar is installed on the top of the carriage. Step S2: When the toilet in the EMU is unoccupied, use a 3D LiDAR to scan the toilet carriage of the EMU to obtain the initial background point cloud map. Step S3: Use 3D LiDAR to scan the status of people in the train's restroom to obtain a point cloud map to be identified; Step S4: Compare the point cloud image to be identified with the initial background point cloud, filter out the points whose point cloud numbers are equidistant from the points in the point cloud image to be identified and delete them, and the remaining points form the point cloud image of the body state of the people in the toilet carriage. Step S5: Determine whether the person is in a normal or abnormal state based on the point cloud map of the person's body status. If the person is determined to be in an abnormal state, the 3D LiDAR will automatically alarm. The specific process is as follows: Step S51: Screen the point cloud set B(B1,B2…B) of the physical status of the personnel. n minz for each ring in ) j Calculate the height h of a person with their head upright. r =h cd -min{minz 1j ,minz 2j ,…,minz nj }, where h cd The height of the 3D LiDAR above the vehicle floor; Step S52: Determine if the person is in a normal state: If the nth ring in the person's physical state point cloud satisfies h r -η≤h cd -minz nj ≤h cd η is half the length of the head, at which point the person is considered to be in a normal state; Step S53: Determine if the person is in an abnormal lying state: If the n1st ring of the person's body state point cloud satisfies ξ represents the thickness of the side of the human body, and within this ring... Small fluctuation value, i.e., variance If μ is a constant approaching 0, then the person is determined to be lying flat on the ground. Data points for the n1+hσ ring are selected, where σ represents the ring number and h represents a positive integer. Within a continuous time range ε1, the n1+hσ ring simultaneously satisfies... If all of these conditions are met, then the person is considered to be experiencing an abnormality while lying flat. Step S54: Determine if the person is in an abnormal upright state: Filter the minimum height value of the n2th ring in the person's body state point cloud. and maximum height value The point where the minimum height value is located is denoted as The point where the maximum height value is located is denoted as like If both α² and β² are constants approaching 0, then the person is determined to be in an upright position. Based on finding point P, select a point O within the ring; this point has the largest coordinates along the x, y, and z axes, denoted as . like If δ is a constant approaching -1, then we assume the person is sitting on the floor of the carriage with their back against the carriage wall. Selecting data points in the n²+hσ ring, the n²+hσ ring within a continuous time interval ε² satisfies: If this person is sitting on the floor of the carriage with their back against the carriage wall, it is determined that this person is in an abnormal upright state. Step S55: Determine if the person is in an abnormal backward-falling state: Filter the minimum height value of the n3rd ring in the person's body state point cloud. like If the distance between γ1 and γ2 is the distance above the buttocks, then the person is squatting on the toilet. Points within the range, h dmt This represents the height of the laser radar reaching the scanning point when it scans a human body, where γ3 is the length of the person's head. The slope is obtained by fitting a straight line formed by points within this range using the least squares method. like If λ is 1, then it is assumed that the person is lying face down on the toilet, and the n3+hσ-th cycle within the continuous time range of ε3 satisfies: slope of points within the range This indicates that the person is exhibiting abnormal backward tilting.

2. The intelligent detection algorithm for abnormal status of personnel in the toilet of a high-speed train as described in claim 1, characterized in that: In step S1, if the temperature inside the infrared body temperature sensor is set to be higher than 50°C, it is determined that there is an open flame smoking; the toilet of the EMU is also equipped with a smoke alarm.

3. The intelligent detection algorithm for abnormal status of personnel in the toilet of a high-speed train as described in claim 1, characterized in that: In step S2, the initial background point cloud is denoted as A. i (x i ,y i ,z i ), where i represents the initial background point cloud number, which is 1, 2, 3, ..., M, and is stored in the set S1(A1, A2, ..., A...). M ).

4. The intelligent detection algorithm for abnormal status of personnel in the toilet of a high-speed train as described in claim 3, characterized in that: In step S3, the point cloud to be identified is B. nj (x j ,y j ,z j ), where j represents the number of the point cloud to be identified, namely 1, 2, 3, ..., N, N = M, and the point cloud image of a certain ring to be identified is stored in set S. n2 (B1,B2…B N The point clouds of all rings are stored in set S(S). 12 ,S 22 …S n2 ), n = 1, 2, 3...n p n represents the number of scan cycles, n p This indicates the total number of scan cycles.

5. The intelligent detection algorithm for abnormal status of personnel in the toilet of a high-speed train as described in claim 4, characterized in that: In step S4, the points whose positional distance between the point cloud image to be identified and the point cloud image in the initial background point cloud image is Δx=|x j -x i |=α1、Δy=|y j -y i |=β1、Δz=|z j -z i |=γ1 and its α1, β1, γ1 are all constants tending to 0, where i=j, the point cloud of the body state of the person is stored in the set B(B1,B2…B n ).

Images