Intrinsic safety type intelligent AI network camera and image analysis system thereof

Abstract

The invention discloses an intrinsic safety type intelligent AI network camera and an image analysis system thereof, and relates to the technical field of image analysis, and the scheme is characterized by comprising an image feature analysis module, a rock stratum monitoring module, a risk assessment module and a behavior safety module. According to the method, the accuracy of crack identification and behavior analysis is improved through refined image data processing logic, the width, depth and distribution mean value of a crack pixel region are meticulously calculated, and a crack distribution table is constructed, so that the system can accurately analyze the crack expansion trend and determine the expansion direction and speed, the risk identification accuracy is improved, and the risk identification efficiency is improved. The prediction reliability of future fracture behaviors, the capability of evaluating the rock stratum stability in real time and the capability of predicting the collapse risk are enhanced, so that the system can warn potential risks in advance, provide a basis for safety management, analyze miner behaviors in real time, effectively identify potential abnormities, reduce potential safety hazards and enhance the safety monitoring and management capability of an under-mine environment.

Description

Technical Field

[0001] The present invention relates to the field of image analysis technology, and in particular to an intrinsically safe intelligent AI network camera and an image analysis system thereof. Background Art

[0002] Image analysis technology involves extracting useful information from images. This field combines technologies such as computer vision, image processing, machine learning, and pattern recognition, aiming to enable computers to understand and interpret visual data in a similar way to humans. Image analysis systems can be applied to a variety of scenarios, including but not limited to medical imaging diagnosis, security monitoring, autonomous vehicles, remote sensing, and multimedia management.

[0003] Among them, the intrinsically safe intelligent AI network camera and its image analysis system is a system that applies image analysis technology. The intrinsically safe intelligent AI network camera refers to a camera used in a high-security environment (such as explosive environments such as petrochemicals and mining). It is equipped with explosion-proof functions and can work stably under extreme conditions. Combining AI and image analysis technology, this camera can not only perform conventional video surveillance, but also identify and respond to specific events or anomalies in the environment in real time by analyzing captured images. Its main purpose is to improve the safety monitoring capabilities of specific dangerous areas, reduce the need for manual inspections through intelligent analysis, and thus improve the overall safety management efficiency.

[0004] In rock formation monitoring or mine safety applications, existing technologies lack accurate crack identification and risk prediction capabilities, leading to misjudgment of crack development trends. For example, the displacement rate and direction differences of rock formations cannot be accurately calculated, increasing the risk of safety accidents. In addition, existing technologies lack the ability to respond quickly to dynamic environments, making it difficult to respond effectively in an emergency. These limitations not only reduce monitoring efficiency in actual operations, but also increase management costs, especially in specific dangerous areas with high monitoring requirements. Summary of the invention

[0005] The present invention provides an intrinsically safe intelligent AI network camera and an image analysis system thereof.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme:

[0007] An image analysis system for an intrinsically safe intelligent AI network camera, the system comprising:

[0008] The image feature analysis module extracts the crack pixel area in the rock formation image based on the image data collected by the camera, calculates the width value, depth value and pixel distribution mean of the crack area, constructs a crack distribution table, analyzes the expansion trend of cracks in adjacent areas based on the crack distribution table, determines the expansion direction and expansion rate, and obtains crack trend data;

[0009] The rock formation monitoring module analyzes the displacement of the rock formation area boundary in the image frame sequence based on the crack trend data, calculates the boundary offset rate and direction difference, and extracts the displacement change characteristics in the continuous frames to determine the stability level of the rock formation and obtain the stability evaluation index;

[0010] The risk assessment module evaluates the stability level of the dangerous area based on the stability assessment index, analyzes the overlap coefficient between the rock formation fracture parameters and the dangerous area, calculates the potential collapse risk of the area based on the stability level and the overlap coefficient, and predicts the risk level change trend in the future time period to obtain collapse risk information;

[0011] Based on the collapse risk information, the behavioral safety module analyzes the movement trajectory of the miners in the monitoring video, calculates the movement amplitude and movement frequency, and evaluates the behavior pattern and movement deviation value to determine potential safety hazards and obtain behavioral hazard warning results.

[0012] The present invention is improved in that the steps of constructing the crack distribution table are specifically as follows:

[0013] Based on the image data collected by the camera, the rock formation image is processed to extract the crack pixel area and determine the crack width and depth of each area using the formula:

[0014]

[0015] Calculate the average characteristic size M of the crack area and obtain the preliminary characteristic data of the crack, where w i represents the width of the i-th crack, d i represents the depth value of the i-th crack, and n represents the total number of cracks;

[0016] Based on the preliminary crack feature data, the pixel distribution mean of the cracks is integrated to construct a crack distribution table.

[0017] The present invention is improved in that the steps of obtaining the crack trend data are specifically as follows:

[0018] Based on the crack distribution table, the relative position of each crack and its adjacent cracks is analyzed, and the distance and angle difference between the cracks are calculated to obtain geometric relationship data;

[0019] Based on the geometric relationship data, the vector field analysis method is used to determine the crack extension direction, and the formula is used:

[0020]

[0021] The crack expansion rate v is calculated to obtain the crack trend data, where Δd represents the change in crack length and Δt is the time interval between two monitorings.

[0022] The present invention is improved in that the calculation steps of the boundary offset rate and the direction difference are specifically as follows:

[0023] Based on the fracture trend data, the rock formation boundary positions of the continuous image frames are extracted, and the position change between each frame and the previous frame is analyzed to obtain the boundary position change data;

[0024] Applying a differential algorithm to process the boundary position change data, determining the boundary offset rate between consecutive frames, and obtaining boundary offset rate information for each time interval;

[0025] Based on the boundary offset rate information, the formula is adopted:

[0026]

[0027] Calculate the offset direction difference of the boundary, where θ represents the angle between the two vectors. represents the boundary offset vector at the initial time point, represents the boundary offset vector at the subsequent time point, and Represent the magnitude of the two vectors, cos -1 Represents the arccosine function, which is used to calculate the angle between two vectors from the result of the dot product.

[0028] The present invention is improved in that the step of obtaining the stability evaluation index is specifically as follows:

[0029] Based on the boundary shift rate and direction difference, key trends are extracted, the consistency and volatility of boundary shift are analyzed, and preliminary analysis results of boundary stability are obtained;

[0030] Based on the preliminary boundary stability analysis results, compared with known geological data, combined with the rock material and activity frequency, the stability level of the rock formation is judged to obtain a stability assessment index.

[0031] The present invention is improved in that the analysis steps of the overlap coefficient are specifically as follows:

[0032] Based on the stability evaluation index, the stability level of the dangerous area is evaluated, and at the same time, the fracture data of the associated rock formations, including the size, shape and direction of the fractures, are collected to obtain the fracture parameters of the dangerous area;

[0033] Based on the crack parameters of the hazardous area, the spatial overlap between the crack and the hazardous area is calculated using the formula:

[0034]

[0035] Get the overlap coefficient CY, where A overlap is the area where the crack overlaps with the danger zone, Atotal is the total area of ​​the danger zone.

[0036] The present invention is improved in that the step of obtaining the collapse risk information is specifically as follows:

[0037] The stability levels and overlap coefficients are weighted, and the potential collapse risks of multiple regions are calculated to obtain regional risk scores;

[0038] Based on the regional risk score, the risk level change trend in the future time period is predicted using the formula:

[0039] RH t =RH t-1 ×e k·(1-CY) ;

[0040] Get collapse risk information, where RH t is the risk level at the prediction time, RH t-1 is the risk level at the previous moment, k is the adjustment coefficient, which indicates the sensitivity of risk to time changes, CY is the overlap coefficient, and t is the time.

[0041] The present invention is improved in that the steps of obtaining the behavior hidden danger warning result are specifically as follows:

[0042] Based on the collapse risk information, the movement trajectory of the miners in the monitoring video is analyzed, and the movement trajectory of each miner is extracted, including the moving path and timestamp, to obtain the movement trajectory information;

[0043] Based on the motion trajectory information, the movement amplitude and movement frequency of each miner are calculated, the deviation from the standard safety behavior is evaluated, and the behavior pattern data is obtained;

[0044] Based on the behavior pattern data, the formula is used:

[0045]

[0046] Identify potential safety hazards and obtain behavioral hazard warning results, where DF is the behavioral deviation percentage, DB is the current behavior value, and DS is the safety behavior standard value.

[0047] An intrinsically safe intelligent AI network camera comprises a processor, a memory, a lens and a power supply. The memory stores a computer program, and the processor implements the above-mentioned image analysis system when executing the computer program.

[0048] Compared with the prior art, the advantages and positive effects of the present invention are:

[0049] In the present invention, the accuracy of crack identification and behavior analysis is improved through refined image data processing logic. By carefully calculating the width, depth and distribution mean of the crack pixel area and constructing a crack distribution table, the system can accurately analyze the crack expansion trend, establish the expansion direction and rate, improve the accuracy of risk identification, enhance the reliability of predicting future crack behavior, and evaluate the stability of rock formations and predict collapse risks in real time. The system can warn of potential risks in advance and provide a basis for safety management. At the same time, real-time analysis of miners' behavior can effectively identify potential anomalies, reduce safety hazards, and enhance the safety monitoring and management capabilities of the underground mine environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. The drawings are only used to illustrate the implementation methods and are not to be considered as limitations of the present invention.

[0051] Figure 1 is a system flow chart in an embodiment of the present invention;

[0052] Figure 2 A flow chart of constructing a crack distribution table in an embodiment of the present invention;

[0053] Figure 3 A flowchart of obtaining crack trend data in an embodiment of the present invention;

[0054] Figure 4 Flow chart of calculation of boundary offset rate and direction difference in an embodiment of the present invention;

[0055] Figure 5 is a flow chart for obtaining stability evaluation indicators in an embodiment of the present invention;

[0056] Figure 6 is a flow chart of analysis of overlap coefficients in an embodiment of the present invention;

[0057] Figure 7 This is a flowchart of obtaining collapse risk information in an embodiment of the present invention;

[0058] Figure 8 The figure is a flow chart of obtaining behavioral hidden danger warning results in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by technicians in the field of the present invention; the terms used in the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings, and are intended to cover non-exclusive inclusions.

[0061] In the description of the embodiments of the present invention, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present invention, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0062] In the description of the embodiments of the present invention, the term "and / or" is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0063] In the description of the embodiments of the present invention, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0064] In the description of the embodiments of the present invention, the technical terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the embodiments of the present invention.

[0065] In the description of the embodiments of the present invention, unless otherwise clearly specified and limited, technical terms such as "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present invention can be understood according to the specific circumstances.

[0066] Example

[0067] The embodiment of the present invention provides an image analysis system for an intrinsically safe intelligent AI network camera, such as Figure 1 As shown, including:

[0068] The image feature analysis module extracts the crack pixel area in the rock formation image based on the image data collected by the camera, calculates the width value, depth value and pixel distribution mean of the crack area, constructs a crack distribution table, analyzes the expansion trend of cracks in adjacent areas based on the crack distribution table, determines the expansion direction and expansion rate, and obtains crack trend data;

[0069] The rock formation monitoring module analyzes the displacement of the rock formation area boundary in the image frame sequence based on the crack trend data, calculates the boundary offset rate and direction difference, and extracts the displacement change characteristics in the continuous frames to determine the stability level of the rock formation and obtain the stability evaluation index;

[0070] The risk assessment module evaluates the stability level of the dangerous area based on the stability assessment index, analyzes the overlap coefficient between the rock formation fracture parameters and the dangerous area, calculates the potential collapse risk of the area based on the stability level and overlap coefficient, and predicts the trend of risk level changes in the future time period to obtain collapse risk information;

[0071] Based on the collapse risk information, the behavioral safety module analyzes the movement trajectory of miners in the monitoring video, calculates the movement amplitude and frequency, and evaluates the behavioral pattern and movement deviation value to determine potential safety hazards and obtain behavioral hazard warning results.

[0072] Crack trend data include crack directionality, expansion speed, and crack distribution density. Stability assessment indicators include boundary displacement index, offset rate, and displacement direction difference. Collapse risk information includes risk level, potential collapse area, and risk change trend. Behavioral hazard warning results include abnormal motion trajectory, behavior frequency change information, and safety hazard level.

[0073] like Figure 2 As shown in Figure 2, the construction steps of the crack distribution table are as follows:

[0074] Based on the image data collected by the camera, the rock formation image is processed to extract the crack pixel area and determine the crack width and depth of each area using the formula:

[0075]

[0076] Calculate the average characteristic size M of the crack area and obtain the preliminary characteristic data of the crack, where w i represents the width of the i-th crack, which is the crack width directly measured from the rock formation image, d irepresents the depth value of the i-th fracture, which is the fracture depth directly measured from the rock formation image, and n represents the total number of fractures, which indicates the number of fractures identified and analyzed in the image;

[0077] A camera is used to obtain high-resolution rock formation images and convert them into grayscale images for subsequent processing. The pixel threshold segmentation method is used to distinguish the crack area from the background area. The crack boundary data is extracted by pixel-by-pixel scanning. The contour detection algorithm is used to track the boundary and identify the closed crack area. The crack width is measured by the pixel distance between the boundaries on both sides of the crack area. The crack depth is calculated by analyzing the change of pixel grayscale value combined with distribution fitting. The width and depth data of the crack are calculated independently in each crack area and stored as a crack feature data file. The average feature size of the crack area is calculated by the statistical average of the width and depth of each crack, and finally the preliminary feature data of the crack is formed. There are three cracks with widths of 2cm, 3cm and 4cm, and depths of 3cm, 4cm and 5cm, respectively. First, the width of each crack is calculated. Value, that is:

[0078] 2 2 +3 2 =13;

[0079] 3 2 +4 2 =25;

[0080] 4 2 +5 2 =41;

[0081] Then summing and averaging the values ​​gives:

[0082]

[0083] Finally, taking the square root of the mean, we get:

[0084]

[0085] That is, the average characteristic size M ≈ 5.13 cm. This result shows that the average crack size is 5.13 cm, which reflects the general distribution of cracks in the rock formation and provides a benchmark for subsequent crack analysis.

[0086] Based on the preliminary characteristic data of cracks, the pixel distribution mean of cracks is integrated to construct a crack distribution table;

[0087] First, the grayscale value of each pixel in the crack area is normalized to reduce noise interference. The grayscale value distribution of all pixels in the crack area is extracted, and the average grayscale value of each crack area is calculated and stored as the pixel distribution mean. At the same time, the width, depth and pixel distribution mean of each crack area are integrated, and the data are arranged in matrix form according to the crack area number sequence, including the geometric characteristics and pixel distribution characteristics of the crack. Finally, a crack distribution table is constructed based on the matrix data, which includes detailed data such as the crack location, size and pixel mean.

[0088] like Figure 3 As shown in Figure 2, the steps for obtaining crack trend data are as follows:

[0089] Based on the crack distribution table, the relative position of each crack and its adjacent cracks is analyzed, the distance and angle difference between the cracks are calculated, and the geometric relationship data is obtained;

[0090] The coordinate data of each crack and its adjacent cracks are extracted. The straight-line distance between adjacent cracks is calculated using the starting and ending positions of the cracks. The directional data of the cracks are analyzed to determine the directional change of each pair of cracks. The angle difference between the cracks is determined by the angle between the directions of the two cracks. The directional data is determined by the geometric orientation of the cracks in the image. In the specific execution process, the starting and ending positions of the cracks are extracted in turn to generate the directional data of each crack. Subsequently, the angular relationship between the two cracks is calculated according to the directional data. Combined with the changes in the straight-line distance and directional angle, the geometric relationship data between the cracks is generated.

[0091] Based on the geometric relationship data, the vector field analysis method is used to determine the propagation direction of the crack (the propagation direction is determined by the relative angle of the crack), and the formula is used:

[0092]

[0093] Calculate the crack expansion rate v and obtain the crack trend data, where Δd represents the change in crack length, which is the difference in crack length measured during two consecutive monitoring periods, and Δt is the time interval between two monitoring periods, which is used to represent the length of time for measuring crack changes, which can be days, months, or years;

[0094] The angle difference and distance change data between cracks are analyzed. The crack expansion direction is determined by integrating the directional data of adjacent crack areas. The crack expansion direction is determined by the relative angle between adjacent cracks. The dynamic changes of the cracks are measured in combination with the time interval. The length change of the crack is obtained from the two monitoring data of the crack length. Then the crack expansion rate is calculated. The expansion direction and expansion rate are combined to form the crack trend data. In the first observation, the crack length was 2 meters. In the second observation two weeks later, the crack length increased to 2.1 meters. Here, Δd = 2.1-2 = 0.1 meters, and the time interval Δt = 2 weeks, i.e. 14 days. Substitute the values ​​into the formula to calculate the crack expansion rate:

[0095]

[0096] The results indicate that the crack expands by about 0.0071 meters per day, a rate that can help assess the activity of the crack and its potential environmental impact.

[0097] like Figure 4 As shown in the figure, the calculation steps of boundary offset rate and direction difference are as follows:

[0098] Based on the fracture trend data, the rock formation boundary positions of the continuous image frames are extracted, and the position changes between each frame and the previous frame are analyzed to obtain the boundary position change data;

[0099] The pixel coordinate information of the rock layer boundary is obtained by reading the image data frame by frame, and the pixel position difference of the same boundary feature point in adjacent frames is calculated. The boundary displacement vectors between frames are integrated with the frame number as the index, and converted into actual distance units according to the image coordinates. By comparing the displacement of the boundary feature points of each frame, the integrity and continuity of all calculated differences are ensured. The time interval between frames is recorded in combination with the timestamp information and corresponds to the displacement data, and the boundary position change data marked with the time interval and frame number are generated.

[0100] Applying a differential algorithm to process the boundary position change data, determine the boundary offset rate between consecutive frames, and obtain the boundary offset rate information for each time interval;

[0101] The displacement of each frame is obtained by calculating the modulus of the inter-frame displacement vector, and the displacement of each frame is normalized using the corresponding time interval. The displacement rate is obtained by dividing the modulus by the time interval, and the rate data is matched with the frame number. The rate sequence is reordered according to the frame time sequence to generate a rate sequence. The overall data change trend is analyzed through the rate sequence, and finally the boundary offset rate information of each time interval is sorted out.

[0102] Based on the boundary offset rate information, the formula is used:

[0103]

[0104] Calculate the offset direction difference of the boundary, where θ represents the angle between the two vectors and is the direction change of the crack boundary. Represents the boundary offset vector at the initial time point, representing the displacement vector from the first frame to the second frame, Represents the boundary offset vector of the subsequent time point, representing the displacement vector from the second frame to the third frame, and Represents the modulus of two vectors, used to calculate the size of the vector, cos -1 Represents the arccosine function, which is used to calculate the angle between two vectors from the result of the dot product;

[0105] Consider the offset vector from frame 1 to frame 2 in the boundary offset rate data Offset vector from frame 2 to frame 3 First, calculate the dot product of the vectors:

[0106]

[0107] Then calculate the magnitude of the vectors separately:

[0108]

[0109] Substitute the dot product result and the modulus value into the formula:

[0110]

[0111] θ=cos -1 (0.7407)≈0.88;

[0112] The result shows that the direction difference of the inter-frame offset is about 50.4 degrees, which quantifies the change of the boundary offset direction in continuous time and provides a clear basis for further evaluating the dynamic stability of the rock formation.

[0113] like Figure 5 As shown in Figure 2, the steps for obtaining the stability evaluation index are as follows:

[0114] Based on the boundary shift rate and direction difference, key trends are extracted, the consistency and volatility of boundary shift are analyzed, and preliminary analysis results of boundary stability are obtained;

[0115] The time series data of rate and direction difference are called. First, the rate data is divided into time intervals, and the standard deviation of each interval is calculated as the volatility indicator. At the same time, the average value of each interval is calculated, and the standard deviation is divided by the average value to get the coefficient of variation. Then, the direction difference sequence is analyzed, and the direction offset angle values ​​are grouped by intervals. The distribution range and central trend of each group are statistically analyzed. Further, by comparing the distribution characteristics of rate volatility and direction difference, the correlation and difference between the two in time series are extracted. Finally, the key trend information about rate fluctuation and direction consistency is generated, and the preliminary analysis results of boundary stability are obtained.

[0116] Based on the preliminary boundary stability analysis results, compared with known geological data, combined with the rock material and activity frequency, the stability level of the rock formation is determined to obtain the stability assessment index;

[0117] By comparing with known geological data, extracting rock material attribute data and historical activity frequency records, dividing rock materials into different types, and statistically analyzing the historical stability levels corresponding to each material, matching the volatility and directional characteristics in the preliminary analysis results, screening the activity frequency curves associated with the material characteristics, combining the volatility characteristics with the activity frequency data, calculating the trend coefficient of frequency change, and comparing the stability data with historical records item by item to obtain the difference indicators associated with the current rock material and activity frequency, evaluate the boundary stability level, and obtain the stability assessment index.

[0118] like Figure 6 As shown in the figure, the analysis steps of the overlap coefficient are as follows:

[0119] Based on the stability assessment index, the stability level of the dangerous area is evaluated, and the fracture data of the associated rock formations, including the size, shape and direction of the fractures, are collected to obtain the fracture parameters of the dangerous area;

[0120] By extracting the stability assessment indicators of each dangerous area, combining the historical stability change records and current monitoring data in the area, the average stability level and fluctuation range of the current dangerous area are statistically analyzed, and the monitoring results of different time periods are further compared to screen out areas with significant stability changes. The areas are grouped according to the preset stability level thresholds, and high-risk areas are marked. At the same time, the relevant data of rock formation cracks are called, and the geometric information of cracks in the dangerous area is collected, including the length, width and direction of the cracks. The geometric parameters of the cracks are matched with the spatial characteristics of the area, and the crack data are statistically summarized to obtain the crack parameters of the dangerous area.

[0121] Based on the crack parameters of the dangerous area, the spatial overlap between the crack and the dangerous area is calculated using the formula:

[0122]

[0123] Get the overlap coefficient CY, where A overlap It is the overlapping area of ​​the crack and the dangerous area, and the overlapping area of ​​the physical space of the crack and the defined dangerous area in the plane or space. total is the total area of ​​the hazardous zone, representing the overall area of ​​the specific area under consideration;

[0124] In a geological assessment, it is determined by measurement that the total area of ​​a specific hazardous area is A. total = 1500 square meters, the area A where the crack overlaps with this area overlap =450 square meters, substitute into the formula to calculate the overlap coefficient CY:

[0125]

[0126] The results show that the cracks cover 0.3 of the danger zone, or 30%, indicating that this area has a higher risk of increased instability due to the cracks.

[0127] like Figure 7 As shown in FIG. 1 , the steps for obtaining collapse risk information are as follows:

[0128] The stability level and overlap coefficient are weighted, the potential collapse risk of multiple areas is calculated, and the regional risk score is obtained;

[0129] The stability level is divided into multiple subsets according to the regional data source. Each subset represents the stability level value of a region. At the same time, the overlap coefficient is expressed as the degree of mutual influence between regions. The weighted values ​​are calculated one by one by analyzing the stability level value and the overlap coefficient of each region one by one. The setting of the weights is based on the results of historical data statistics or experimental data calibration. After that, a comprehensive assessment is made on the potential collapse risk of each region. The results of the weighted calculation are normalized with the frequency of collapse events that have occurred in the region in history. During the standardization process, the frequencies of all regions are normalized to avoid error amplification between data. Finally, the potential collapse risk score of each region is obtained.

[0130] Based on the regional risk score, the risk level change trend in the future period is predicted using the formula:

[0131] RH t =RH t-1 ×e k·(1-CY) ;

[0132] Get collapse risk information, where RH t is the risk level at the prediction time, RH t-1is the risk level at the previous moment, indicating the collapse risk level at the last known time point before the prediction, k is the adjustment coefficient, indicating the sensitivity of time changes to risk, CY is the overlap coefficient, and t is the time;

[0133] First, select a suitable time range for data sliding analysis, divide the risk score data within the time period into multiple small intervals, calculate the average risk value of each interval one by one, and compare the risk value changes in two consecutive time periods, analyze the proportion of the changes, and mark the change trend. At the same time, through the analysis of the historical risk score change trend of each area, select the prediction method or data regression model, use the historical score and trend analysis results as input data, predict the risk score change trend of each area in the future time range, divide the risk level according to the distribution of the predicted score, and finally form the collapse risk information in the future time period. Let RH t-1 =0.5, indicating that the risk level at the previous moment was 0.5, the overlap coefficient CY = 0.3, and the adjustment coefficient k = 0.1, then RH is calculated according to the formula t :

[0134] RH t =0.5×e 0.1·(1-0.3) ;

[0135] RH t =0.5×e 0.07 ≈0.5×1.072;

[0136] RH t =0.536;

[0137] The results show that after considering the current overlap coefficient and time sensitivity, the predicted risk level has increased slightly, providing a quantitative prediction of future risk development.

[0138] like Figure 8 As shown in FIG. 1 , the steps for obtaining the behavior hidden danger warning results are as follows:

[0139] Based on the collapse risk information, the movement trajectory of the miners in the monitoring video is analyzed, and the movement trajectory is extracted, including the moving path and timestamp, to obtain the movement trajectory information;

[0140] The surveillance video is segmented into frames, the dynamic target area in each frame is extracted, the spatial position of each target area is marked, and the timestamp is recorded. The timestamp is matched with the spatial coordinates to generate the initial motion trajectory of each miner in the video. Then, by removing the static area and background data, all the trajectory points of continuous motion are retained and arranged in chronological order into complete path information. Finally, the data of the trajectory points are integrated to establish the complete motion trajectory data of each miner, including all moving paths and timestamps. The motion trajectory information is obtained after batch processing of all video data.

[0141] Based on the motion trajectory information, the movement amplitude and frequency of each miner are calculated, the deviation from the standard safety behavior is evaluated, and the behavior pattern data is obtained;

[0142] The spatial distance between each trajectory point is extracted, the moving distance within the time interval is calculated, and the amplitude of the miner's movement is obtained. At the same time, the number of trajectory points is counted, and the number of repetitions of the movement per unit time is calculated in combination with the time length to obtain the movement frequency. When calculating the amplitude, the distance of all trajectory points is averaged and the standard deviation is recorded to analyze the variation range of the miner's movement amplitude. When calculating the frequency, the number of movements of each miner is accumulated and combined with the time span statistics to obtain the frequency value. Finally, the movement amplitude and movement frequency data of each miner are recorded and integrated to generate the miner's behavior pattern data.

[0143] Based on the behavior pattern data, the formula is used:

[0144]

[0145] Identify potential safety hazards and obtain behavioral hazard warning results, where DF is the behavioral deviation percentage, DB is the current behavior value, which comes from the movement amplitude and movement frequency data obtained by analyzing the miner's movement trajectory, and DS is the safety behavior standard value, which is the safety standard defined in the safety operating procedures;

[0146] DB is the actual measured action frequency of 0.4 times / second, while DS (safety standard value) is 0.3 times / second. Substitute it into the formula for calculation:

[0147]

[0148] The results showed that the miners' behavior deviated from the safety standard by 33.33%, and further monitoring or intervention was needed to prevent potential accidents.

[0149] An intrinsically safe intelligent AI network camera comprises a processor, a memory, a lens and a power supply. The memory stores a computer program, and the processor implements the above-mentioned image analysis system when executing the computer program.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, not to limit them; although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention. In particular, as long as there is no structural conflict, the various technical features mentioned in each embodiment can be combined in any way. The present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions that fall within the scope of the claims.

Claims

1. An image analysis system for an intrinsically safe intelligent AI network camera, characterized in that: The system comprises: The image feature analysis module extracts the crack pixel area in the rock formation image based on the image data collected by the camera, calculates the width value, depth value and pixel distribution mean of the crack area, constructs a crack distribution table, analyzes the expansion trend of cracks in adjacent areas based on the crack distribution table, determines the expansion direction and expansion rate, and obtains crack trend data; The rock formation monitoring module analyzes the displacement of the rock formation area boundary in the image frame sequence based on the crack trend data, calculates the boundary offset rate and direction difference, and extracts the displacement change characteristics in the continuous frames to determine the stability level of the rock formation and obtain the stability evaluation index; The risk assessment module evaluates the stability level of the dangerous area based on the stability assessment index, analyzes the overlap coefficient between the rock formation fracture parameters and the dangerous area, calculates the potential collapse risk of the area based on the stability level and the overlap coefficient, and predicts the risk level change trend in the future time period to obtain collapse risk information; Based on the collapse risk information, the behavioral safety module analyzes the movement trajectory of the miners in the monitoring video, calculates the movement amplitude and movement frequency, and evaluates the behavior pattern and movement deviation value to determine potential safety hazards and obtain behavioral hazard warning results.

2. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The steps for constructing the crack distribution table are specifically as follows: Based on the image data collected by the camera, the rock formation image is processed to extract the crack pixel area and determine the crack width and depth of each area using the formula: Calculate the average characteristic size M of the crack area and obtain the preliminary characteristic data of the crack, where w i represents the width of the i-th crack, d i represents the depth value of the i-th crack, and n represents the total number of cracks; Based on the preliminary crack feature data, the pixel distribution mean of the cracks is integrated to construct a crack distribution table.

3. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The steps for obtaining the crack trend data are specifically as follows: Based on the crack distribution table, the relative position of each crack and its adjacent cracks is analyzed, and the distance and angle difference between the cracks are calculated to obtain geometric relationship data; Based on the geometric relationship data, the vector field analysis method is used to determine the crack extension direction, and the formula is used: The crack expansion rate v is calculated to obtain the crack trend data, where Δd represents the change in crack length and Δt is the time interval between two monitorings.

4. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The calculation steps of the boundary offset rate and the direction difference are specifically as follows: Based on the fracture trend data, the rock formation boundary positions of the continuous image frames are extracted, and the position change between each frame and the previous frame is analyzed to obtain the boundary position change data; Applying a differential algorithm to process the boundary position change data, determining the boundary offset rate between consecutive frames, and obtaining boundary offset rate information for each time interval; Based on the boundary offset rate information, the formula is adopted: Calculate the offset direction difference of the boundary, where θ represents the angle between the two vectors. represents the boundary offset vector at the initial time point, represents the boundary offset vector at the subsequent time point, and Represent the magnitude of the two vectors, cos -1 Represents the arccosine function, which is used to calculate the angle between two vectors from the result of the dot product.

5. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The steps for obtaining the stability evaluation index are specifically as follows: Based on the boundary shift rate and direction difference, key trends are extracted, the consistency and volatility of boundary shift are analyzed, and preliminary analysis results of boundary stability are obtained; Based on the preliminary boundary stability analysis results, compared with known geological data, combined with the rock material and activity frequency, the stability level of the rock formation is judged to obtain a stability assessment index.

6. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The analysis steps of the overlap coefficient are specifically as follows: Based on the stability evaluation index, the stability level of the dangerous area is evaluated, and at the same time, the fracture data of the associated rock formations, including the size, shape and direction of the fractures, are collected to obtain the fracture parameters of the dangerous area; Based on the crack parameters of the hazardous area, the spatial overlap between the crack and the hazardous area is calculated using the formula: Get the overlap coefficient CY, where A overlap is the area where the crack overlaps with the danger zone, A total is the total area of ​​the danger zone.

7. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The steps for obtaining the collapse risk information are specifically as follows: The stability levels and overlap coefficients are weighted, and the potential collapse risks of multiple regions are calculated to obtain regional risk scores; Based on the regional risk score, the risk level change trend in the future time period is predicted using the formula: RH t =RH t-1 ×e k·(1-CY) ; Get collapse risk information, where RH t is the risk level at the prediction time, RH t-1 is the risk level at the previous moment, k is the adjustment coefficient, which indicates the sensitivity of risk to time changes, CY is the overlap coefficient, and t is the time.

8. The image analysis system of the intrinsically safe intelligent AI network camera according to claim 1, characterized in that: The steps for obtaining the behavior hidden danger warning result are specifically as follows: Based on the collapse risk information, the movement trajectory of the miners in the monitoring video is analyzed, and the movement trajectory of each miner is extracted, including the moving path and timestamp, to obtain the movement trajectory information; Based on the motion trajectory information, the movement amplitude and movement frequency of each miner are calculated, the deviation from the standard safety behavior is evaluated, and the behavior pattern data is obtained; Based on the behavior pattern data, the formula is used: Identify potential safety hazards and obtain behavioral hazard warning results, where DF is the behavioral deviation percentage, DB is the current behavior value, and DS is the safety behavior standard value.

9. An intrinsically safe intelligent AI network camera, comprising a processor, a memory, a lens and a power supply, characterized in that: The memory stores a computer program, and when the processor executes the computer program, the image analysis system of the intrinsically safe intelligent AI network camera according to any one of claims 1 to 8 is implemented.

Images