Method for identifying and early warning night voyage unsafe behaviors of ship console operator on duty

Abstract

The invention provides a method for identifying and early warning night voyage unsafe behaviors of a ship console operator on duty, which comprises the following steps of: acquiring an infrared video image of the ship console operator on duty during night voyage in real time, and preprocessing the image to highlight a human body feature target; carrying out key point extraction and tracking on the human body feature target by utilizing a human body key point detection algorithm, and generating a behavior judgment index quantity according to a relative position and an action state between key points; calculating a threshold value, a standard deviation, a kurtosis value and a skewness value of each index quantity, and performing preliminary identification on the unsafe behaviors; judging whether an unsafe behavior recognition result is correct or not by using a personnel unsafe behavior rechecking algorithm; and evaluating the risk level of the unsafe behavior by using an unsafe behavior risk evaluation algorithm, carrying out risk prompt, and carrying out visual and auditory early warning when the risk level reaches a high risk. The method can effectively identify whether the behavior of the operator on duty on the console is safe and compliant during the night voyage of the ship.

Description

Technical Field

[0001] The present invention relates to the field of marine transportation technology, and in particular to a method for identifying and warning unsafe behaviors of ship bridge duty personnel during night navigation. Background Art

[0002] The behavior of the Officer of the Watch (OOW) during night navigation directly affects the navigation safety of the ship. Traditional video surveillance systems cannot effectively monitor the behavior of the Officer of the Watch (OOW) in the low-light environment of the bridge, and cannot even identify the personnel on the bridge. Therefore, during night navigation, it is impossible to identify and warn of the degree of danger of the Officer of the Watch (OOW) smoking, looking at the phone, lying down, and other unsafe behaviors of the Officer of the Watch (OOW) on the bridge. The existing ship bridge navigation watch alarm system (BNWAS) can only ensure that there is always someone on duty on the bridge, but cannot detect the specific behavior of the Officer of the Watch during the duty process. Summary of the invention

[0003] According to the technical problems raised above, a method for identifying and warning unsafe behaviors of ship bridge duty personnel during night navigation is provided. The present invention mainly utilizes the infrared image of the human body provided by a thermal imaging camera, combined with a human key point detection algorithm, a behavior determination index, a personnel unsafe behavior review algorithm, and an unsafe behavior risk assessment algorithm, to achieve the identification of unsafe behaviors of duty personnel in a low-light environment of the bridge, and reduce the risk of accidents caused by unsafe behaviors of bridge personnel.

[0004] The technical means adopted by the present invention are as follows:

[0005] A method for identifying and warning unsafe behaviors of night navigation personnel on duty at a ship's bridge, comprising:

[0006] S1. Real-time acquisition of infrared video images of the on-duty personnel on the ship's bridge during night navigation, and image preprocessing to highlight human feature targets;

[0007] S2, using the human key point detection algorithm to extract and track the key points of the human feature targets in S1, and generate behavior judgment indicators based on the relative positions and action states between the key points;

[0008] S3, calculating the threshold, standard deviation, kurtosis and skewness of each indicator in S2, and preliminarily identifying the unsafe behavior of personnel in the infrared video image;

[0009] S4, using the personnel unsafe behavior review algorithm to determine whether the initial detection result in S3 is accurate according to the duration of the intermediate process of the personnel action change;

[0010] S5. Use the unsafe behavior risk assessment algorithm to assess the risk level of unsafe behavior and provide risk level prompts. When the risk level reaches a high risk, visual and auditory warnings are issued.

[0011] Furthermore, the infrared video image is acquired by using a thermal imaging camera, and the image preprocessing utilizes an infrared image enhancement algorithm to remove noise and enhance image edges.

[0012] Further, the key points include: nose Nose, neck Neck, right shoulder RShoulder, right elbow RElbow, right wrist RWrist, left shoulder LShoulder, left elbow LE1bow, left wrist LWrist, mid hip MidHip, right hip RHip, right knee RKnee, right ankle RAnkle, left hip LHip, left knee LKnee, left ankle LAnkle, right eye REye, left eye LEye, right ear REar, left ear LAr, left big toe LBigToe, left little toe LSmallToe, left heel LHeel, right big toe RBigToe, right little toe RSmallToe and right heel RHeel.

[0013] Furthermore, the generation of the behavior determination index includes:

[0014] Select the neck and the mid-hip from the key points to form the upper body vector V body , the left hip LHip and the left knee LKnee form the left thigh vector V lthigh , the right hip RHip and the right knee RKnee form the right thigh vector V rthigh ; The upper body vector V body With the left thigh vector V lthigh The angle is denoted as Angle1, and the upper body vector V body With the right thigh vector V rthigh The angle between the right hip RHip and the right ankle RAnkle is recorded as Angle2; the distance between the right hip RHip and the right ankle RAnkle is recorded as D1, and the distance between the left hip LHip and the left ankle LAnkle is recorded as D2; assuming that the horizon vector is V horizon , the upper body vector V bodv Vector V with horizon horizon The angle formed is recorded as Angle3;

[0015] The distance between the left ear Lear and the left big toe LBigToe is recorded as the length h1 of the human body's circumscribed rectangular frame, the distance between the right ear REar and the right big toe RBigToe is recorded as the length h2 of the human body's circumscribed rectangular frame, the distance between the right big toe RBigToe and the left big toe LBigToe is recorded as the width w1 of the human body's circumscribed rectangular frame, and the distance between the right little toe RSmallToe and the left little toe LSmallToe is recorded as the width w2 of the human body's circumscribed rectangular frame, and the aspect ratios of the human body's circumscribed rectangle are obtained, which are recorded as H1, H2, H3 and H4 respectively, where:

[0016]

[0017] Assuming that the non-standing state of the on-duty personnel includes two types of behaviors: Sit-down and Lie-down, Angle1, Angle2, D1 and D2 are used as indicators for determining the Sit-down behavior; Angle3, H1, H2, H3 and H4 are used as indicators for determining the Lie-down behavior.

[0018] Furthermore, the step S3 specifically includes:

[0019] Fit the data of different distributions to the corresponding probability distribution model, use the box plot to filter the abnormal data values, and then determine the threshold of each behavior judgment indicator through mathematical statistics;

[0020] Assume [α1, α2] as the threshold of Angle1 and Angle2; Assume [α3, α4] as the threshold of Angle3; Assume [β1, β2] as the threshold of D1 and D2; Assume [γ1, γ2], [γ3, γ4] as the threshold of H1, H2, H3 and H4; If, in the actual action state of the personnel, the behavior judgment index data of one of the two types of behaviors, Sit-down and Lie-down, is included in the corresponding threshold and lasts for a long time, then a judgment is made on the behavior category at this time.

[0021] Furthermore, the personnel unsafe behavior review algorithm includes:

[0022] The start time T1, end time T2 and duration T3 of the intermediate process of the personnel's motion change are obtained and recorded to obtain the threshold value [δ1, δ2] of the duration of the intermediate process; when it is detected that the duration of the personnel's behavior does not exceed the threshold value [δ1, δ2], it is determined that the personnel is in the intermediate process of the motion change at this time; if the duration of the personnel's behavior exceeds the threshold value [δ1, δ2], it means that the personnel is not in the intermediate process of the motion change, and the initial detection result in S3 is determined to be accurate.

[0023] Furthermore, the unsafe behavior risk assessment algorithm includes:

[0024] Based on the historical results of personnel unsafe behavior detection, three risk levels are designed: Low, Medium, and High;

[0025] For the three types of risks, Low, Medium, and High, the thresholds of the duration of each risk are set to [μ1, μ2], [ρ1, ρ2], and After using the personnel unsafe behavior review algorithm in S4 to judge the accuracy of the initial detection results, the unsafe behavior risk assessment algorithm is used to determine the risk level of the behavior. When the duration of the unsafe behavior reaches the time threshold corresponding to the different risk levels, a risk level prompt is given, and when a high risk is reached, a visual and auditory warning is given.

[0026] Compared with the prior art, the present invention has the following advantages:

[0027] The present invention provides a method for identifying and warning of unsafe behaviors of ship bridge duty personnel during night navigation. The method uses a thermal imaging camera to collect infrared video images of the bridge as the identification data source. With the help of the temperature difference between the human body surface and the environment, it can capture human activity images, solving the problem that traditional visible light cameras cannot detect human behavior in low-light environments. Based on the human key point detection algorithm, the spatial position of the human key points is obtained and the changes in the key point positions are continuously tracked, the unsafe behaviors of the bridge personnel are identified, and the risk analysis model is used to warn of the unsafe behaviors. It can detect various unsafe behaviors of OOW during night navigation (including: smoking, looking at mobile phones, non-standing state, etc.) and provide visual and auditory warnings according to the degree of danger of the behavior, standardize the behavior of OOW during night navigation, and improve the safety of ships during night navigation.

[0028] Based on the above reasons, the present invention can be widely promoted in the fields of marine transportation and the like. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 The present invention is a flow chart of the method for identifying and warning unsafe behaviors of ship bridge duty personnel during night navigation. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] like Figure 1 As shown, the present invention provides a method for identifying and warning unsafe behaviors of ship bridge duty personnel at night, comprising:

[0034] S1. Real-time acquisition of infrared video images of the on-duty personnel on the ship's bridge during night navigation, and image preprocessing to highlight human feature targets;

[0035] In specific implementation, as a preferred embodiment of the present invention, the infrared video image is acquired by using a thermal imaging camera, and the image preprocessing uses an infrared image enhancement algorithm to remove noise and enhance the image edge.

[0036] During implementation, infrared image data enhancement processing is performed. The contrast-limited adaptive histogram equalization, median filtering processing and image sharpening filtering processing based on the Sobel operator are combined to form an infrared image data enhancement module.

[0037] In the first step, contrast-limited adaptive histogram equalization is used to enhance the contrast of infrared image data, while avoiding noise and artifacts. The contrast limit is calculated as follows:

[0038]

[0039] Wherein, H′(i) represents the histogram value after contrast limitation; H(i) represents the number of pixels with intensity value i in the original histogram; and C represents the set contrast limitation threshold.

[0040] The second step is to use median filtering to denoise the infrared image data, which can effectively remove the noise in the data while well preserving the edge information of the data. The median filter calculation is as follows:

[0041]

[0042] Among them, p′(i, j) represents the pixel value of the original image at position (i, j); median represents the calculation of the median of the specified pixel value; p represents the pixel value matrix of the original image; i represents the row index in the image, pointing to the vertical position of the current pixel; j represents the column index in the image, pointing to the horizontal position of the current pixel; m represents the row offset relative to the current pixel; n represents the column offset relative to the current pixel; k represents the size of the filter.

[0043] The third step is to use the image sharpening filter based on the Sobel operator to effectively enhance the edge and detail information of the data and reduce the impact of noise on the edge. The Sobel operator is divided into two templates, horizontal and vertical, when processing data, and processes data from different directions. The horizontal direction (X axis) operator template is as follows:

[0044]

[0045] The vertical direction (Y axis) operator template is as follows:

[0046]

[0047] High-quality infrared image data can be obtained through the infrared image data enhancement module.

[0048] S2, using the human key point detection algorithm to extract and track the key points of the human feature targets in S1, and generate behavior judgment indicators based on the relative positions and action states between the key points;

[0049] In specific implementation, as a preferred embodiment of the present invention, the key points include: nose Nose, neck Neck, right shoulder RShoulder, right elbow RElbow, right wrist RWrist, left shoulder LShoulder, left elbow LElbow, left wrist LWrist, mid hip MidHip, right hip RHip, right knee RKnee, right ankle RAnkle, left hip LHip, left knee LKnee, left ankle LAnkle, right eye REye, left eye LEye, right ear REar, left ear LAr, left big toe LBigToe, left little toe LSmallToe, left heel LHeel, right big toe RBigToe, right little toe RSmallToe and right heel RHeel. The corresponding relationship between the 25 key points and their serial numbers is shown in Table 1.

[0050] Table 1 25 key points and their serial numbers

[0051] Serial number name Serial number name 0 Nose 13 LKnee 1 Neck 14 LAnkle 2 RShoulder 15 R 3 RElbow 16 LEye 4 RWrist 17 R 5 L Shoulder 18 LEAR 6 LElbow 19 LBigToe 7 LWrist 20 LSmallToe 8 MidHip 21 LHeel 9 RHip 22 RBigToe 10 RKnee 23 RSmallToe 11 RAnkle 24 RHeel 12 LHip

[0052] In specific implementation, as a preferred embodiment of the present invention, the generation of the behavior determination index includes:

[0053] Select the neck and mid-hip from the key points to form the upper torso vector V body =(p1,p8), the left hip LHip and the left knee LKnee form the left thigh vector V lthigh =(p12, p13), the right hip RHip and the right knee RKnee form the right thigh vector V rthigh =(p9, p10); the upper body vector V body With left thigh vector V lthigh The angle is recorded as Angle1, and the upper body vector V body With right thigh vector V rthigh The angle is recorded as Angle2; the distance from the right hip RHip to the right ankle RAnkle is recorded as D1 = (p9, p11), and the distance from the left hip LHip to the left ankle LAnkle is recorded as D2 = (p12, p14); assuming that the horizon vector is V horizon =(1,0), the upper body vector V body Vector V with horizon horizon The angle formed is recorded as Angle3;

[0054]

[0055]

[0056]

[0057]

[0058]

[0059] Among them, arccos represents the inverse cosine function; pIx represents the x-axis coordinate value of the I-th key point; pIy represents the y-axis coordinate value of the I-th key point.

[0060] The distance between the left ear Lear and the left big toe LBigToe is recorded as the length h1 of the human body's circumscribed rectangular frame, the distance between the right ear REar and the right big toe RBigToe is recorded as the length h2 of the human body's circumscribed rectangular frame, the distance between the right big toe RBigToe and the left big toe LBigToe is recorded as the width w1 of the human body's circumscribed rectangular frame, and the distance between the right little toe RSmallToe and the left little toe LSmallToe is recorded as the width w2 of the human body's circumscribed rectangular frame, and the aspect ratios of the human body's circumscribed rectangle are obtained, which are recorded as H1, H2, H3 and H4 respectively, where:

[0061]

[0062]

[0063]

[0064]

[0065]

[0066] Assuming that the non-standing state of the on-duty personnel includes two types of behaviors: Sit-down and Lie-down, Angle1, Angle2, D1 and D2 are used as indicators for determining the Sit-down behavior; Angle3, H1, H2, H3 and H4 are used as indicators for determining the Lie-down behavior.

[0067] S3, calculating the threshold, standard deviation, kurtosis and skewness of each indicator in S2, and preliminarily identifying the unsafe behavior of personnel in the infrared video image;

[0068] In specific implementation, as a preferred embodiment of the present invention, step S3 specifically includes:

[0069] Fit the data of different distributions to the corresponding probability distribution model, use the box plot to filter the abnormal data values, and then determine the threshold of each behavior judgment indicator through mathematical statistics;

[0070] Assume [α1, α2] as the threshold of Angle1 and Angle2; Assume [α3, α4] as the threshold of Angle3; Assume [β1, β2] as the threshold of D1 and D2; Assume [γ1, γ2], [γ3, γ4] as the threshold of H1, H2, H3 and H4; If, in the actual action state of the personnel, the behavior judgment index data of one of the two types of behaviors, Sit-down and Lie-down, is included in the corresponding threshold and lasts for a long time, then a judgment is made on the behavior category at this time.

[0071] In the implementation, D1 and D2 are both spatial distance data for judging Sit-down behavior, but they do not belong to the same distribution. Therefore, the kernel density (KDE) method is used to solve the thresholds of the two. The Gaussian kernel function in KDE is selected for calculation. The Gaussian kernel function formula is as follows:

[0072] K(x, x') = exp(-||x-x'|| 2 / (2×sigma 2 ))

[0073] Among them, K(x, x') represents the similarity or kernel value between point x and point x'; x represents a data point or feature vector, which is the actual value of the input sample; x' represents a reference point, an estimated value or a model predicted value; sigma represents the standard deviation of the Gaussian kernel, which is used to control the width of the kernel.

[0074] The mapping change of the sample data x after Gaussian kernel function processing is shown as follows:

[0075] x→(exp(-γ||x-q1|| 2 ), exp(-γ||x-q2|| 2 ))

[0076] Among them, γ represents the parameter used to control the width of the Gaussian kernel; q1 and q2 represent the reference point or center point in the model.

[0077] The bandwidth parameter Sigma in the Gaussian kernel function determines the shape and characteristics of the data distribution, data accuracy and other aspects. The optimal bandwidth of D1 and D2 is obtained by using the five-fold cross-validation method. It is brought into the Gaussian kernel function to obtain the thresholds of D1 and D2. When calculating the thresholds of Angle1, Angle2, Angle3, H1, H2, H3 and H4, it is found that there are extreme outliers with obvious distribution in the data. First, in order to solve this problem, the extreme outliers of the data are filtered using a box plot, and then the threshold of each indicator is obtained using mathematical statistics. Set [α1, α2] as the threshold of Angle1 and Angle2; set [α3, α4] as the threshold of Angle3; set [β1, β2] as the threshold of D1 and D2; set [γ1, γ2], [γ3, γ4] as the threshold of H1, H2, H3 and H4. The judgment conditions for Sit-down behavior and Lie-down behavior of the bridge duty personnel in the non-standing state are shown in Table 2.

[0078] Table 2 Judgment conditions for sit-down behavior and lie-down behavior in non-standing state

[0079]

[0080] S4. Using the personnel unsafe behavior review algorithm to determine whether the unsafe behavior identification result is correct;

[0081] In specific implementation, as a preferred embodiment of the present invention, the personnel unsafe behavior review algorithm includes:

[0082] The start time T1, end time T2 and duration T3 of the intermediate process of the personnel's motion change are obtained and recorded to obtain the threshold value [δ1, δ2] of the duration of the intermediate process; when it is detected that the duration of the personnel's behavior does not exceed the threshold value [δ1, δ2], it is determined that the personnel is in the intermediate process of the motion change at this time; if the duration of the personnel's behavior exceeds the threshold value [δ1, δ2], it means that the personnel is not in the intermediate process of the motion change, and the initial detection result in S3 is determined to be accurate.

[0083] During implementation, after an accurate judgment is made on the current result, the values ​​of T1, T2, and T3 will be automatically reset to zero, so that the judgment results of subsequent personnel behaviors can be cyclically checked.

[0084] S5. Use the unsafe behavior risk assessment algorithm to assess the risk level of unsafe behavior and provide risk level prompts. When the risk level reaches a high risk, visual and auditory warnings are issued.

[0085] In specific implementation, as a preferred embodiment of the present invention, the unsafe behavior risk assessment algorithm includes:

[0086] Based on the historical results of personnel unsafe behavior detection, three risk levels are designed: Low, Medium, and High;

[0087] For the three types of risks, Low, Medium, and High, the thresholds of the duration of each risk are set to [μ1, μ2], [ρ1, ρ2], and After using the personnel unsafe behavior review algorithm in S4 to judge the accuracy of the initial detection results, the unsafe behavior risk assessment algorithm is used to determine the risk level of the behavior. When the duration of the unsafe behavior reaches the time threshold corresponding to the different risk levels, a risk level prompt is given, and when a high risk is reached, a visual and auditory warning is given.

[0088] During implementation, assuming that the duration of unsafe behavior is T4, when T4∈[μ1,μ2], the level of unsafe behavior of personnel is low risk; when T4∈[ρ1,ρ2], the level of unsafe behavior of personnel is medium risk; , the level of personnel unsafe behavior is high risk.

[0089] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0090] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0091] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0092] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0093] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0094] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying and warning unsafe behaviors of ship bridge duty personnel at night, characterized in that: include: S1. Real-time acquisition of infrared video images of the on-duty personnel on the ship's bridge during night navigation, and image preprocessing to highlight human feature targets; S2, using the human key point detection algorithm to extract and track the key points of the human feature targets in S1, and generate behavior judgment indicators based on the relative positions and action states between the key points; S3, calculating the threshold, standard deviation, kurtosis and skewness of each indicator in S2, and preliminarily identifying the unsafe behavior of personnel in the infrared video image; S4. Using the personnel unsafe behavior review algorithm to determine whether the unsafe behavior identification result is correct; S5. Use the unsafe behavior risk assessment algorithm to assess the risk level of unsafe behavior and provide risk warnings. When the risk level reaches a high risk, visual and auditory warnings are issued.

2. The method for identifying and warning unsafe behaviors of ship bridge duty personnel at night according to claim 1 is characterized in that: The infrared video image is acquired by using a thermal imaging camera, and the image preprocessing uses an infrared image enhancement algorithm to remove noise and enhance image edge features.

3. The method for identifying and warning unsafe behaviors of ship bridge duty personnel at night according to claim 1 is characterized in that: The key points include: nose, neck, right shoulder RShoulder, right elbow RElbow, right wrist RWrist, left shoulder LShoulder, left elbow LElbow, left wrist LWrist, mid hip MidHip, right hip RHip, right knee RKnee, right ankle RAnkle, left hip LHip, left knee LKnee, left ankle LAnkle, right eye REye, left eye LEye, right ear REar, left ear LAr, left big toe LBigToe, left little toe LSmallToe, left heel LHeel, right big toe RBigToe, right little toe RSmallToe and right heel RHeel.

4. The method for identifying and warning unsafe behaviors of ship bridge duty personnel at night according to claim 1 or 3, characterized in that: The generation of the behavior determination index comprises: Select the neck and the mid-hip from the key points to form the upper body vector V body , the left hip LHip and the left knee LKnee form the left thigh vector V lthigh , the right hip RHip and the right knee RKnee form the right thigh vector V rthigh ; The upper body vector V body With the left thigh vector V lthigh The angle is denoted as Angle1, and the upper body vector V body With the right thigh vector V rthigh The angle between the right hip RHip and the right ankle RAnkle is recorded as Angle2; the distance between the right hip RHip and the right ankle RAnkle is recorded as D1, and the distance between the left hip LHip and the left ankle LAnkle is recorded as D2; assuming that the horizon vector is V horizon , the upper body vector V body With horizon vector v horizon The angle formed is recorded as Angle3; The distance between the left ear Lear and the left big toe LBigToe is recorded as the length h1 of the human body's circumscribed rectangular frame, the distance between the right ear REar and the right big toe RBigToe is recorded as the length h2 of the human body's circumscribed rectangular frame, the distance between the right big toe RBigToe and the left big toe LBigToe is recorded as the width w1 of the human body's circumscribed rectangular frame, and the distance between the right little toe RSmallToe and the left little toe LSmallToe is recorded as the width w2 of the human body's circumscribed rectangular frame, and the aspect ratios of the human body's circumscribed rectangle are obtained, which are recorded as H1, H2, H3 and H4 respectively, where: Assuming that the non-standing state of the on-duty personnel includes two types of behaviors: Sit-down and Lie-down, Angle1, Angle2, D1 and D2 are used as indicators for determining the Sit-down behavior; Angle3, H1, H2, H3 and H4 are used as indicators for determining the Lie-down behavior.

5. The method for identifying and warning unsafe behaviors of ship bridge duty personnel at night according to claim 1 is characterized in that: The step S3 specifically includes: Fit the data of different distributions to the corresponding probability distribution model, use the box plot to filter the abnormal data values, and then determine the threshold of each behavior judgment indicator through mathematical statistics; Assume [α1, α2] as the thresholds of Angle1 and Angle2; Assume [α3, α4] as the thresholds of Angle3; Assume [β1, β2] as the thresholds of D1 and D2; Assume [γ1, γ2], [γ3, γ4] as the thresholds of H1, H2, H3 and H4; If, under the actual behavior state of the personnel, the behavior judgment index data of one of the two types of behaviors, Sit-down and Lie-down, is included in the corresponding threshold and lasts for a long time, then a judgment is made on the behavior category at this time.

6. The method for identifying and warning unsafe behaviors of ship bridge duty personnel at night according to claim 1 is characterized in that: The personnel unsafe behavior review algorithm includes: The start time T1, end time T2 and duration T3 of the intermediate process of the personnel's motion change are obtained and recorded to obtain the threshold value [δ1, δ2] of the duration of the intermediate process; when it is detected that the duration of the personnel's behavior does not exceed the threshold value [δ1, δ2], it is determined that the personnel is in the intermediate process of the motion change at this time; if the duration of the personnel's behavior exceeds the threshold value [δ1, δ2], it means that the personnel is not in the intermediate process of the motion change, and the initial detection result in S3 is determined to be accurate.

7. The method for identifying and warning unsafe behaviors of ship bridge duty personnel at night according to claim 1 is characterized in that: The unsafe behavior risk assessment algorithm includes: Based on the historical results of personnel unsafe behavior detection, three risk levels are designed: Low, Medium, and High; For the three types of risks, Low, Medium, and High, the thresholds of the duration of each risk are set to [μ1,μ2], [ρ1,ρ2], and After using the personnel unsafe behavior review algorithm in S4 to judge the accuracy of the initial detection results, the unsafe behavior risk assessment algorithm is used to determine the risk level of the behavior. When the duration of the unsafe behavior reaches the time threshold corresponding to the different risk levels, a risk level prompt is given, and when a high risk is reached, a visual and auditory warning is given.

Images