Visual detection early warning system and method based on computer

By constructing the feature data set of vision sensors and calculating similarity, accurate positioning and chain abnormality warning of monitoring targets is achieved, and the problems of difficulty in detecting and tracking targets and inability to warning chain abnormalities in complex environments in the existing technology are solved, and monitoring is improved System accuracy and security.

CN119049234AInactive Publication Date: 2024-11-29HAINAN YIPI BIOTECHNOLOGY DEVELOPMENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411121343.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing monitoring and early warning systems are difficult to accurately detect and track multiple targets in complex environments, especially when there are occlusion, lighting changes and viewing angle changes, and it is impossible to warning of chain abnormalities in advance, resulting in an increase in possible accidents.

Method used

By constructing the feature data set of vision sensors, the similarity between each object and the feature data set is calculated, the monitoring target is determined and real-time tracking is performed, the abnormality index threshold and dangerous distance are calculated, and the precise positioning of the monitoring target and the chain abnormality warning are achieved.

Benefits of technology

It realizes rapid search and positioning of monitoring targets in complex environments, improves detection accuracy, avoids the occurrence of chain abnormalities, and reduces economic losses and dangers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119049234A_ABST
    Figure CN119049234A_ABST
Patent Text Reader

Abstract

The invention discloses a visual detection early warning system and method based on a computer, and belongs to the technical field of intelligent early warning. The method comprises the following steps: collecting all environmental data in a visual range by using a visual sensor, analyzing according to the environmental data, extracting a feature data set, and determining a monitoring target in the visual range of the visual sensor; after the monitoring target in the visible range is determined, the mark of the monitoring target is used for positioning, and real-time monitoring target tracking is realized; analyzing a visual image of a monitoring target in history, extracting associated data when the monitoring target is abnormal, and calculating an abnormal index threshold value of the monitoring target; monitoring the associated data of the monitoring target in real time, judging whether the monitoring target is abnormal or not according to the calculated associated data threshold, and giving out early warning; after judgment and early warning are carried out on the monitoring targets, the relative distance is calculated according to real-time positioning of all the monitoring targets, and chain abnormity early warning is carried out.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent early warning technology, and in particular to a computer-based visual detection early warning system and method. Background Art

[0002] With the continuous advancement of sensor technology and the reduction of costs, various sensor devices including cameras, radars, and lidars have become more popular and practical. Various sensors are used to monitor the environment and targets in real time and provide data support for monitoring and early warning systems. With the rapid development of computer vision and image processing technology, algorithms in areas such as target detection, target tracking, and image recognition have been continuously improved, making real-time analysis and processing of monitoring images more feasible and effective. With the development of society and the increasing attention people pay to security and monitoring, monitoring and early warning technology has been widely used. This includes the demand for public safety, such as traffic monitoring and urban security, as well as the demand for personal and private property safety, such as home monitoring and vehicle monitoring. Therefore, intelligent monitoring and early warning based on visual technology has been well known to the public. With the development of artificial intelligence and automation technology, monitoring and early warning systems are also developing in the direction of intelligence and automation. For example, deep learning algorithms are used to realize intelligent analysis and recognition of monitoring images, and automatic control systems are used to realize automatic response and processing of abnormal events. However, due to the complexity and variability of real-world environmental conditions, target detection and tracking become more difficult when there are a large number of targets, cross-motion, or dense crowds in the monitoring scene. The algorithm is easily affected by factors such as occlusion, illumination change, and perspective change, resulting in reduced accuracy of detection and tracking. In addition, there may be multiple monitoring targets during monitoring, and most monitoring and early warning systems only perform intelligent monitoring and early warning on multiple monitoring targets separately, but do not analyze the relationship between multiple monitoring targets, resulting in only warning of abnormal situations of a single monitoring target, and unable to provide early warning of chain abnormal situations caused by chain relationships to avoid major accidents. Therefore, the present invention designs a computer-based visual detection and early warning system and method for such situations. Summary of the invention

[0003] The purpose of the present invention is to provide a computer-based visual detection warning system and method to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0005] A computer-based visual detection early warning method, the method comprising the following steps:

[0006] S100, using a visual sensor to collect all environmental data within the visible range, analyzing the environmental data, extracting a feature data set, and determining a monitoring target within the visible range of the visual sensor;

[0007] Furthermore, the specific method for determining the monitoring target within the visible range of the visual sensor is:

[0008] S101, collect all the data of the monitoring target in history, and construct the monitoring target data set A, A={a1, a2, a3...a i}, a1, a2, a3...a i is the 1st, 2nd, 3rd, ..., ith data type of the monitoring target, where i is a positive integer; the monitoring target is a general term for objects of the same type monitored by the visual sensor;

[0009] S102, repeat the step of S101 n times to obtain n monitoring target data sets A, compare the n monitoring target data sets, extract the data types that exist in the n monitoring target data sets, integrate the data types that exist in the n monitoring target data sets, and construct a feature data set B of the monitoring target, B = {b1, b2, b3...b p}, b1, b2, b3...b p is the 1st, 2nd, 3rd, ...pth characteristic data of the monitoring target, where p is a positive integer;

[0010] S103, assuming that the number of objects within the visual range of the visual sensor is m, obtain m object data sets; use the visual sensor to collect all data types within the monitoring range in real time, and construct corresponding data sets separately for the data types of all objects within the monitoring range;

[0011] S104, assuming that the data set of each object within the visible range of the visual sensor is {c1, c2, c3...c k}, c1, c2, c3…c k is the 1st, 2nd, 3rd...kth data in the object dataset, where k is a positive integer; the similarity between each object dataset and the feature dataset is calculated using the following formula:

[0012]

[0013] In the formula, ρ is the similarity between each object dataset and feature dataset, and s is the number of identical data in each object dataset and feature dataset;

[0014] The similarity threshold θ_s is calculated based on the historical monitoring target data set and the feature data set. The calculation formula is:

[0015]

[0016] S105. When the similarity between the object dataset and the feature dataset is ρ≥θ_s, the object corresponding to the object dataset is judged as a monitoring target and marked. On the contrary, when ρ<θ_s, the object corresponding to the object dataset is judged not to be a monitoring target and ignored. The above m object datasets are calculated and judged, and finally the monitoring target set is obtained as {j1, j2, j3...j u}, j1, j2, j3...j u is the 1st, 2nd, 3rd, ..., uth monitoring target in the m object data set, where u is a positive integer and u <m。

[0017] By collecting and analyzing the historical data of the monitored targets, different feature data can be obtained according to different objects, so that the intelligent early warning method can be flexibly applied in a variety of monitoring scenarios, and the monitored targets within the visible range of the visual sensor can be found quickly and conveniently.

[0018] S200, after determining the monitoring target within the visible range, use the marker of the monitoring target to locate it and realize real-time monitoring target tracking;

[0019] Furthermore, the specific steps of using the marker of the monitored target for real-time tracking are as follows:

[0020] S201, using the visual sensor as the origin to establish a spatial rectangular coordinate system, assuming that the front of the visual sensor is the positive direction of the x-axis of the spatial rectangular coordinate system, the bottom is the positive direction of the z-axis of the spatial rectangular coordinate system, and the left is the positive direction of the y-axis of the spatial rectangular coordinate system;

[0021] S202, when it is determined that the object within the visual range of the visual sensor is a monitoring target, the distance between the monitoring target and the visual sensor is measured as L, the line segment between the monitoring target and the visual sensor is set as L_m, the angle between the line segment L_m and the positive direction of the z-axis is set as α_1, the mapping line segment of the horizontal plane formed by the line segment L_m on the x-axis and the y-axis in the spatial rectangular coordinate system is L_n, and the angle between the line segment L_n and the positive direction of the y-axis is measured as α_;

[0022] S203, calculate the coordinates (X, Y, Z) of the monitored target in the spatial rectangular coordinate system, and the calculation formula is:

[0023] X=L×sin(α_1)×sin(α_2);

[0024] Y=L×sin(α_1)×cos(α_2);

[0025] Z = L × cos(α_1);

[0026] In the formula, X, Y, and Z are the coordinates of the x-axis, y-axis, and z-axis of the monitored target in the spatial rectangular coordinate system, respectively;

[0027] S204. According to the above method, when the monitored target moves within the visible range of the visual sensor, after real-time measurement of L, α_1 and α_2, the real-time positioning and tracking of the monitored target is achieved using a spatial rectangular coordinate system.

[0028] Using the spatial rectangular coordinate system to locate and track the monitoring target and calculating the three-dimensional spatial coordinate system of the monitoring target can avoid the positioning ambiguity of the monitoring target in a complex environment, eliminate the inaccurate positioning caused by object occlusion, and achieve accurate positioning of the monitoring target in the three-dimensional spatial coordinate system.

[0029] S300, analyzing the visual images of the monitored targets in history, extracting the associated data when the monitored targets are abnormal, and calculating the abnormal index threshold of the monitored targets;

[0030] Furthermore, the specific steps for calculating the abnormal index threshold of the monitoring target are as follows:

[0031] S301, collect all the data when the monitoring target has abnormality in history, and form an abnormal data set C, C = {s1, s2, s3...s h}, s1, s2, s3...s h is the 1st, 2nd, 3rd...hth data type when the monitoring target is abnormal, h is a positive integer; collect all the data when the monitoring target is not abnormal in the history, and form a data set D, D = {s′1, s′2, s′3...s′ g}, s′1, s′2, s′3...s′ g The 1st, 2nd, 3rd, ... gth data types when the monitoring target has no abnormality, where g is a positive integer;

[0032] S302, compare the data types in the abnormal data set C and the data set D, extract different data types to construct an abnormal associated data set {w1, w2, w3...w e}, w1, w2, w3...w e are the 1st, 2nd, 3rd, ...eth data types that are different after comparison, where e is a positive integer;

[0033] S303, collect the values ​​of all associated data when the monitored target is abnormal in history, calculate the average value of all associated data after d collections, use the calculated average value as the threshold of each associated data, and calculate the abnormal index threshold of the monitored target. The formula is:

[0034]

[0035] In the formula, θ_y is the abnormal index threshold of the monitoring target, Q r is the threshold for each associated data, Q′ rTo monitor the value of each associated data when no abnormality occurs in the target, r takes a value from 1 to e.

[0036] S400, real-time monitoring of the associated data of the monitoring target is performed, and whether an abnormality occurs in the monitoring target is determined according to the calculated associated data threshold value, and an early warning is issued;

[0037] Furthermore, the specific steps to determine whether the monitored target is abnormal and issue an early warning are:

[0038] S401, using visual sensors to collect relevant data of the monitored target in real time, and obtain a real-time relevant data set {w′1, w′2, w′3...w′ e}, w′1, w′2, w′3...w′ e are the 1st, 2nd, 3rd, ...eth real-time associated data values ​​respectively, where e is a positive integer;

[0039] S402: Compare the collected real-time correlation data value with the correlation data threshold.

[0040] w′ r >Q r , the associated data is judged as abnormal associated data. Finally, after e times of judgment, t abnormal associated data are obtained, and the abnormal index θ_s of the real-time monitoring target is calculated. The formula is:

[0041]

[0042] When θ_s≥θ_y, the monitored target is judged to be abnormal and an abnormal warning is issued.

[0043] S500: After judging and warning the monitoring target, calculate the relative distance according to the real-time positioning of all monitoring targets and issue a chain abnormality warning.

[0044] Furthermore, the specific steps for chain abnormality warning are as follows:

[0045] S501, collect the relative distances L_x of all monitoring targets with abnormalities in history, and construct a set {L_x1, L_x2, L_x3...L_x f}, L_x1, L_x2, L_x3...L_x f is the relative distance of the 1st, 2nd, 3rd...fth monitoring targets where abnormalities occur, where f is a positive integer; calculate the average value of the f relative distances, and take the average value L_x_p as the dangerous distance of the monitoring target.

[0046] S502: When it is determined that an abnormality occurs in a monitoring target, the distance between the abnormal monitoring target and other monitoring targets within the visual range of the visual sensor is calculated. The spatial coordinates of the abnormal monitoring target are (X1, Y1, Z1), and the spatial coordinates of the other monitoring targets are (X v , Y v , Z v ), v ranges from 1 to u-1, and the formula is:

[0047]

[0048] In the formula, L_s is the distance between the abnormal monitoring target and other monitoring targets within the visual range of the visual sensor;

[0049] S503, when there is a monitoring target with L_s≤L_x_p, it is determined that the abnormal monitoring target will cause an abnormality to occur when the distance is less than L_x_p, resulting in a chain abnormality, and a chain abnormality warning is issued;

[0050] S504. When issuing an abnormal warning and a chain warning of a monitoring target, simultaneously send the monitoring target positioning data to the maintenance personnel, so that the maintenance personnel can handle the abnormality in time.

[0051] When an abnormality occurs in a monitoring target, the abnormal monitoring target may affect the remaining monitoring targets, causing the remaining monitoring targets to also become abnormal, resulting in a chain of abnormalities in an area. For example, when an abnormality occurs in a moving vehicle during traffic, it may affect the remaining vehicles that are closer to the abnormal vehicle, causing the remaining vehicles to also become abnormal, resulting in a chain of traffic accidents, causing greater losses and harm. Therefore, early warning of chain abnormalities can avoid the occurrence of chain abnormalities.

[0052] A computer-based visual detection and early warning system, the visual detection and early warning system includes a data collection module, a monitoring target search module, a monitoring target tracking module, a threshold calculation module, a distance calculation module and an early warning module;

[0053] The data collection module is used to collect all data of the monitoring target in history, and to monitor the monitoring target in real time and collect real-time data;

[0054] The monitoring target search module is used to search for monitoring targets within the visual range of the visual sensor;

[0055] The monitoring target tracking module is used to mark the monitoring target after finding the monitoring target within the visual range of the visual sensor, and track the coordinates of the monitoring target in the spatial rectangular coordinate system in real time;

[0056] The threshold calculation module is used to calculate the similarity threshold when searching for the monitoring target within the visual range of the visual sensor and the threshold for judging whether the monitoring target has an abnormality or a chain abnormality;

[0057] The distance calculation module is used to calculate the relative distance between the abnormal monitoring target and the other monitoring targets;

[0058] The warning module is used to issue corresponding warnings when abnormalities and chain abnormalities occur in the monitored target, and output the real-time coordinates of the monitored target during the warning.

[0059] The monitoring target search module is used to obtain the characteristic data of the monitoring target after analyzing all the data of the monitoring target in history, judge all objects within the visible range of the visual sensor according to the characteristic data of the monitoring target, and search for the monitoring target within the visible range of the visual sensor.

[0060] The threshold calculation module includes a similarity threshold calculation unit, an abnormal index threshold calculation unit and a danger distance calculation module;

[0061] The similarity threshold calculation unit is used to calculate the similarity threshold according to the historical monitoring target data set and the feature data set;

[0062] The abnormal index threshold calculation unit is used to collect the values ​​of all associated data when an abnormality occurs in the monitoring target in the history, calculate the average value of all associated data after d times of collection, use the calculated average value as the threshold of each associated data, and calculate the abnormal index threshold of the monitoring target according to the threshold of each associated data and the value of each associated data when the monitoring target does not have an abnormality;

[0063] The danger distance calculation module is used to collect all relative distances of the monitoring targets that have abnormalities in history, calculate the average value of the relative distances, and use the average value as the danger distance of the monitoring target.

[0064] The early warning module is used to issue an abnormality warning when it is judged that an abnormality occurs in the monitoring target. When it is judged that an abnormality occurs in the monitoring target, the relative distance between the abnormal monitoring target and the remaining monitoring targets is calculated. When it is judged that the relative distance between the monitoring target and the remaining monitoring targets does not exceed the dangerous distance, it is judged that a chain abnormality occurs, and a chain abnormality warning is issued, and the real-time coordinates of the monitoring target are output during the warning.

[0065] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0066] 1. The present invention analyzes all data of the monitored targets in history, collects feature data of the monitored targets, forms a feature data set, and determines the monitored targets within the visible range of the visual sensor based on the feature data set. This can achieve quick search of the monitored targets within the visible range of the visual sensor and prevent missing the monitored targets.

[0067] 2. The present invention locates the monitoring target within the visible range of the visual sensor by constructing a spatial rectangular coordinate system, calculates the spatial coordinates of the monitoring target, and tracks the monitoring target. The position of the monitoring target can be determined in a complex environment.

[0068] 3. The present invention determines whether chain abnormalities will occur by calculating the relative distance between the abnormal monitoring target and the other monitoring targets, and issues early warning to avoid the occurrence of chain abnormalities, which may lead to more economic losses and dangers. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0070] Figure 1 It is a module distribution diagram of a computer-based visual detection and early warning system of the present invention;

[0071] Figure 2 It is a schematic diagram of the steps of a computer-based visual detection and early warning method of the present invention. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0073] See also Figure 1-Figure 2 , the present invention provides a technical solution:

[0074] A computer-based visual detection early warning method, the method comprising the following steps:

[0075] S100, using a visual sensor to collect all environmental data within the visible range, analyzing the environmental data, extracting a feature data set, and determining a monitoring target within the visible range of the visual sensor;

[0076] The specific method for determining the monitoring target within the visual range of the visual sensor is:

[0077] S101, collect all the data of the monitoring target in history, and construct the monitoring target data set A, A={a1, a2, a3...a i}, a1, a2, a3...a i is the 1st, 2nd, 3rd, ..., ith data type of the monitoring target, where i is a positive integer; the monitoring target is a general term for objects of the same type monitored by the visual sensor;

[0078] S102, repeat the step of S101 n times to obtain n monitoring target data sets A, compare the n monitoring target data sets, extract the data types that exist in the n monitoring target data sets, integrate the data types that exist in the n monitoring target data sets, and construct a feature data set B of the monitoring target, B = {b1, b2, b3...b p}, b1, b2, b3...b p is the 1st, 2nd, 3rd, ...pth characteristic data of the monitoring target, where p is a positive integer;

[0079] S103, assuming that the number of objects within the visual range of the visual sensor is m, obtain m object data sets; use the visual sensor to collect all data types within the monitoring range in real time, and construct corresponding data sets separately for the data types of all objects within the monitoring range;

[0080] S104: Assume that the data set of each object within the visible range of the visual sensor is

[0081] {c1, c2, c3...c k}, c1, c2, c3...c k is the 1st, 2nd, 3rd...kth data in the object dataset, where k is a positive integer; the similarity between each object dataset and the feature dataset is calculated using the following formula:

[0082]

[0083] In the formula, ρ is the similarity between each object dataset and the feature dataset, and s is the number of identical data in each object dataset and the feature dataset. The similarity threshold θ_ is calculated based on the historical monitoring target dataset and the feature dataset, and the calculation formula is:

[0084]

[0085] S105. When the similarity between the object dataset and the feature dataset is ρ≥θ_s, the object corresponding to the object dataset is judged as a monitoring target and marked. On the contrary, when ρ<θ_s, the object corresponding to the object dataset is judged not to be a monitoring target and ignored. The above m object datasets are calculated and judged, and finally the monitoring target set is obtained as {j1, j2, j3...j u}, j1, j2, j3...j u is the 1st, 2nd, 3rd, ..., uth monitoring target in the m object data set, where u is a positive integer and u <m。

[0086] By collecting and analyzing the historical data of the monitored targets, different feature data can be obtained according to different objects, so that the intelligent early warning method can be flexibly applied in a variety of monitoring scenarios, and the monitored targets within the visible range of the visual sensor can be found quickly and conveniently.

[0087] S200, after determining the monitoring target within the visible range, use the marker of the monitoring target to locate it and realize real-time monitoring target tracking;

[0088] The specific steps of real-time tracking using the markers of the monitored target are as follows:

[0089] S201, using the visual sensor as the origin to establish a spatial rectangular coordinate system, assuming that the front of the visual sensor is the positive direction of the x-axis of the spatial rectangular coordinate system, the bottom is the positive direction of the z-axis of the spatial rectangular coordinate system, and the left is the positive direction of the y-axis of the spatial rectangular coordinate system;

[0090] S202, when it is determined that the object within the visual range of the visual sensor is a monitoring target, the distance between the monitoring target and the visual sensor is measured as L, the line segment between the monitoring target and the visual sensor is set as L_m, the angle between the line segment L_m and the positive direction of the z-axis is set as α_1, the mapping line segment of the horizontal plane formed by the line segment L_m on the x-axis and the y-axis in the spatial rectangular coordinate system is L_n, and the angle between the line segment L_n and the positive direction of the y-axis is measured as α_;

[0091] S203, calculate the coordinates (X, Y, Z) of the monitored target in the spatial rectangular coordinate system, and the calculation formula is:

[0092] X=L×sin(α_1)×sin(α_2);

[0093] Y=L×sin(α_1)×cos(α_2);

[0094] Z = L × cos(α_1);

[0095] In the formula, X, Y, and Z are the coordinates of the x-axis, y-axis, and z-axis of the monitored target in the spatial rectangular coordinate system, respectively;

[0096] S204. According to the above method, when the monitored target moves within the visible range of the visual sensor, after real-time measurement of L, α_1 and α_2, the real-time positioning and tracking of the monitored target is achieved using a spatial rectangular coordinate system.

[0097] Using the spatial rectangular coordinate system to locate and track the monitoring target and calculating the three-dimensional spatial coordinate system of the monitoring target can avoid the positioning ambiguity of the monitoring target in a complex environment, eliminate the inaccurate positioning caused by object occlusion, and achieve accurate positioning of the monitoring target in the three-dimensional spatial coordinate system.

[0098] S300, analyzing the visual images of the monitored targets in history, extracting the associated data when the monitored targets are abnormal, and calculating the abnormal index threshold of the monitored targets;

[0099] S301, collect all the data when the monitoring target has abnormality in history, and form an abnormal data set C, C = {s1, s2, s3...s h}, s1, s2, s3...s h is the 1st, 2nd, 3rd...hth data type when the monitoring target is abnormal, h is a positive integer; collect all the data when the monitoring target is not abnormal in the history, and form a data set D, D = {s′1, s′2, s′3...s′ g}, s′1, s′2, s′3...s′ g The 1st, 2nd, 3rd, ... gth data types when the monitoring target has no abnormality, where g is a positive integer;

[0100] S302, compare the data types in the abnormal data set C and the data set D, extract different data types to construct an abnormal associated data set {w1, w2, w3...w e}, w1, w2, w3...w e are the 1st, 2nd, 3rd, ...eth data types that are different after comparison, where e is a positive integer;

[0101] S303, collect the values ​​of all associated data when the monitored target is abnormal in history, calculate the average value of all associated data after d collections, use the calculated average value as the threshold of each associated data, and calculate the abnormal index threshold of the monitored target. The formula is:

[0102]

[0103] In the formula, θ_y is the abnormal index threshold of the monitoring target, Q r is the threshold for each associated data, Q′ r To monitor the value of each associated data when no abnormality occurs in the target, r takes a value from 1 to e.

[0104] S400, real-time monitoring of the associated data of the monitoring target is performed, and whether an abnormality occurs in the monitoring target is determined according to the calculated associated data threshold value, and an early warning is issued;

[0105] The specific steps to determine whether the monitoring target is abnormal and issue an early warning are:

[0106] S401, using visual sensors to collect relevant data of the monitored target in real time, and obtain a real-time relevant data set {w′1, w′2, w′3...w′ e}, w′1, w′2, w′3...w′ e are the 1st, 2nd, 3rd, ...eth real-time associated data values ​​respectively, where e is a positive integer;

[0107] S402: Compare the collected real-time associated data value with the associated data threshold. r >Q r , the associated data is judged as abnormal associated data. Finally, after e times of judgment, t abnormal associated data are obtained, and the abnormal index θ_s of the real-time monitoring target is calculated. The formula is:

[0108]

[0109] When θ_s≥θ_y, the monitored target is judged to be abnormal and an abnormal warning is issued.

[0110] S500: After judging and warning the monitoring target, calculate the relative distance according to the real-time positioning of all monitoring targets and issue a chain abnormality warning.

[0111] The specific steps for chain abnormality warning are:

[0112] S501, collect the relative distances L_x of all monitoring targets with abnormalities in history, and construct a set {L_x1, L_x2, L_x3...L_x f}, L_x1, L_x2, L_x3...L_x f is the relative distance of the 1st, 2nd, 3rd...fth monitoring targets where abnormalities occur, where f is a positive integer; calculate the average value of the f relative distances, and take the average value L_x_p as the dangerous distance of the monitoring target.

[0113] S502: When it is determined that an abnormality occurs in a monitoring target, the distance between the abnormal monitoring target and other monitoring targets within the visual range of the visual sensor is calculated. The spatial coordinates of the abnormal monitoring target are (X1, Y1, Z1), and the spatial coordinates of the other monitoring targets are (X v , Y v , Z v ), v ranges from 1 to u-1, and the formula is:

[0114]

[0115] In the formula, L_s is the distance between the abnormal monitoring target and other monitoring targets within the visual range of the visual sensor;

[0116] S503, when there is a monitoring target with L_s≤L_x_p, it is determined that the abnormal monitoring target will cause an abnormality to occur when the distance is less than L_x_p, resulting in a chain abnormality, and a chain abnormality warning is issued;

[0117] S504. When issuing an abnormal warning and a chain warning of a monitoring target, simultaneously send the monitoring target positioning data to the maintenance personnel, so that the maintenance personnel can handle the abnormality in time.

[0118] When an abnormality occurs in a monitoring target, the abnormal monitoring target may affect the other monitoring targets, causing the other monitoring targets to also become abnormal, resulting in chain abnormalities in an area. For example, when a moving vehicle has an abnormality during traffic driving, it may affect the other vehicles that are closer to the abnormal vehicle, causing the other vehicles to also become abnormal, resulting in chain traffic accidents, causing greater losses and harm. Therefore, early warning of chain abnormalities can avoid the occurrence of chain abnormalities. A computer-based visual detection and warning system, the visual detection and warning system includes a data collection module, a monitoring target search module, a monitoring target tracking module, a threshold calculation module, a distance calculation module and a warning module;

[0119] The data collection module is used to collect all data of the monitoring target in history, and to monitor the monitoring target in real time and collect real-time data;

[0120] The monitoring target search module is used to search for monitoring targets within the visual range of the visual sensor;

[0121] The monitoring target tracking module is used to mark the monitoring target after finding the monitoring target within the visual range of the visual sensor, and track the coordinates of the monitoring target in the spatial rectangular coordinate system in real time;

[0122] The threshold calculation module is used to calculate the similarity threshold when searching for the monitoring target within the visual range of the visual sensor and the threshold for judging whether the monitoring target has an abnormality or a chain abnormality;

[0123] The distance calculation module is used to calculate the relative distance between the abnormal monitoring target and the other monitoring targets;

[0124] The warning module is used to issue corresponding warnings when abnormalities and chain abnormalities occur in the monitored target, and output the real-time coordinates of the monitored target during the warning.

[0125] The monitoring target search module is used to obtain the characteristic data of the monitoring target after analyzing all the data of the monitoring target in history, judge all objects within the visible range of the visual sensor according to the characteristic data of the monitoring target, and search for the monitoring target within the visible range of the visual sensor.

[0126] The threshold calculation module includes a similarity threshold calculation unit, an abnormal index threshold calculation unit and a danger distance calculation module;

[0127] The similarity threshold calculation unit is used to calculate the similarity threshold according to the historical monitoring target data set and the feature data set;

[0128] The abnormal index threshold calculation unit is used to collect the values ​​of all associated data when an abnormality occurs in the monitoring target in the history, calculate the average value of all associated data after d times of collection, use the calculated average value as the threshold of each associated data, and calculate the abnormal index threshold of the monitoring target according to the threshold of each associated data and the value of each associated data when the monitoring target does not have an abnormality;

[0129] The danger distance calculation module is used to collect all relative distances of the monitoring targets that have abnormalities in history, calculate the average value of the relative distances, and use the average value as the danger distance of the monitoring target.

[0130] The early warning module is used to issue an abnormality warning when it is judged that an abnormality occurs in the monitoring target. When it is judged that an abnormality occurs in the monitoring target, the relative distance between the abnormal monitoring target and the remaining monitoring targets is calculated. When it is judged that the relative distance between the monitoring target and the remaining monitoring targets does not exceed the dangerous distance, it is judged that a chain abnormality occurs, and a chain abnormality warning is issued, and the real-time coordinates of the monitoring target are output during the warning.

[0131] Example 1

[0132] First, use visual sensors to monitor passing vehicles on a certain road. First, collect the data of vehicles in history, and obtain the characteristic data of vehicles as {size, structure, license plate}. First, judge the objects within the visible range of the visual sensor. First, collect the data of three objects, namely {size, structure}, {size, license plate}, and {size, structure license plate}. Use the formula to calculate the similarity of each object. The formula is:

[0133]

[0134] According to the above formula, the similarities of the three objects are calculated to be 99.5%, 99.5% and 100% respectively; the similarity threshold is set to 75%, so there are three final monitoring targets, which are set to be A1, A2 and A3 respectively;

[0135] In the spatial rectangular coordinate system, the three-dimensional spatial coordinates of the three monitoring targets are (3, 2, 5), (9, 4, 5) and (13, 1, 5). After analyzing the historical data, the associated data of the vehicle abnormality is {speed, tire temperature}, and the thresholds of the associated data are 160km / h and 76 degrees Celsius, respectively. The three monitoring targets are now monitored in real time, the speed and tire temperature are measured, and the abnormal index is calculated. The designed abnormal index threshold is, and the formula is:

[0136]

[0137] According to the above formula, the abnormal indexes of the three monitoring targets are calculated to be 80, 75, and 94 respectively. After comparison, it is found that the monitoring target A3 may be abnormal, and an abnormal warning is issued;

[0138] The relative distances between the abnormal monitoring target and the two monitoring targets A1 and A2 are calculated to be 10 and 5 respectively. According to historical data, the danger distance is 6m. Therefore, it is obtained that monitoring targets A1 and A3 may have chain abnormalities, and a chain abnormality warning is issued, and the positioning coordinates of A1 and A3 are output.

[0139] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0140] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A computer-based visual detection and early warning method, characterized in that: The method comprises the following steps: S100, using a visual sensor to collect all environmental data within the visible range, analyzing the environmental data, extracting a feature data set, and determining a monitoring target within the visible range of the visual sensor; S200, after determining the monitoring target within the visible range, use the marker of the monitoring target to locate it and realize real-time monitoring target tracking; S300, analyzing the visual images of the monitored targets in history, extracting the associated data when the monitored targets are abnormal, and calculating the abnormal index threshold of the monitored targets; S400, real-time monitoring of the associated data of the monitoring target is performed, and whether an abnormality occurs in the monitoring target is determined according to the calculated associated data threshold value, and an early warning is issued; S500: After judging and warning the monitoring target, calculate the relative distance according to the real-time positioning of all monitoring targets and issue a chain abnormality warning.

2. A computer-based visual detection and early warning method according to claim 1, characterized in that: The specific method for determining the monitoring target within the visible range of the visual sensor in S100 is: S101, collect all the data of the monitoring target in history, and construct the monitoring target data set A, A={a1, a2, a3...a i }, a1, a2, a3...a i is the 1st, 2nd, 3rd, ..., ith data type of the monitoring target, where i is a positive integer; the monitoring target is a general term for objects of the same type monitored by the visual sensor; S102, repeat the step of S101 n times to obtain n monitoring target data sets A, compare the n monitoring target data sets, extract the data types that exist in the n monitoring target data sets, integrate the data types that exist in the n monitoring target data sets, and construct a feature data set B of the monitoring target, B = {b1, b2, b3...b p }, b1, b2, b3...b p is the 1st, 2nd, 3rd, ...pth characteristic data of the monitoring target, where p is a positive integer; S103, assuming that the number of objects within the visual range of the visual sensor is m, obtain m object data sets; use the visual sensor to collect all data types within the monitoring range in real time, and construct corresponding data sets separately for the data types of all objects within the monitoring range; S104, assuming that the data set of each object within the visible range of the visual sensor is {c1, c2, c3...c k }, c1, c2, c3...c k is the 1st, 2nd, 3rd...kth data in the object dataset, where k is a positive integer; the similarity between each object dataset and the feature dataset is calculated using the following formula: In the formula, ρ is the similarity between each object dataset and feature dataset, and s is the number of identical data in each object dataset and feature dataset; The similarity threshold θ_s is calculated based on the historical monitoring target data set and the feature data set. The calculation formula is: S105. When the similarity between the object dataset and the feature dataset is ρ≥θ_s, the object corresponding to the object dataset is judged as a monitoring target and marked. On the contrary, when ρ<θ_s, the object corresponding to the object dataset is judged not to be a monitoring target and ignored. The above m object datasets are calculated and judged, and finally the monitoring target set is obtained as {j1, j2, j3...j u }, j1, j2, j3...j u is the 1st, 2nd, 3rd, ..., uth monitoring target in the m object data set, where u is a positive integer and u <m。 3. The computer-based visual detection and early warning method according to claim 1, characterized in that: The specific steps of using the marker of the monitoring target for real-time tracking in S200 are: S201, using the visual sensor as the origin to establish a spatial rectangular coordinate system, assuming that the front of the visual sensor is the positive direction of the x-axis of the spatial rectangular coordinate system, the bottom is the positive direction of the z-axis of the spatial rectangular coordinate system, and the left is the positive direction of the y-axis of the spatial rectangular coordinate system; S202, when it is determined that the object within the visual range of the visual sensor is a monitoring target, the distance between the monitoring target and the visual sensor is measured as L, the line segment between the monitoring target and the visual sensor is set as L_m, the angle between the line segment L_m and the positive direction of the z-axis is set as α_1, the mapping line segment of the horizontal plane formed by the line segment L_m on the x-axis and the y-axis in the spatial rectangular coordinate system is set as L_n, and the angle between the line segment L_n and the positive direction of the y-axis is measured as α_2; S203, calculate the coordinates (X, Y, Z) of the monitored target in the spatial rectangular coordinate system, and the calculation formula is: X=L×sin(α_1)×sin(α_2); Y=L×sin(α_1)×cos(α_2); Z = L × cos(α_1); In the formula, X, Y, and Z are the coordinates of the x-axis, y-axis, and z-axis of the monitored target in the spatial rectangular coordinate system, respectively; S204. According to the above method, when the monitored target moves within the visible range of the visual sensor, after real-time measurement of L, α_1 and α_2, the real-time positioning and tracking of the monitored target is achieved using a spatial rectangular coordinate system.

4. The computer-based visual detection and early warning method according to claim 1, characterized in that: The specific steps of calculating the monitoring target abnormality index threshold in S300 are: S301, collect all the data when the monitoring target has abnormality in history, and form an abnormal data set C, C = {s1, s2, s3...s h }, s1, s2, s3...s h is the 1st, 2nd, 3rd...hth data type when the monitoring target is abnormal, h is a positive integer; collect all the data when the monitoring target is not abnormal in the history, and form a data set D, D = {s′1, s′2, s′3...s′ g }, s′1, s′2, s′3...s′ g The 1st, 2nd, 3rd, ... gth data types when the monitoring target has no abnormality, where g is a positive integer; S302, compare the data types in the abnormal data set C and the data set D, extract different data types to construct an abnormal associated data set {w1, w2, w3...w e }, w1, w2, w3...w e are the 1st, 2nd, 3rd, ...eth types of data that are different after comparison, where e is a positive integer; S303, collect the values ​​of all associated data when the monitored target is abnormal in history, calculate the average value of all associated data after d collections, use the calculated average value as the threshold of each associated data, and calculate the abnormal index threshold of the monitored target. The formula is: In the formula, θ_y is the abnormal index threshold of the monitoring target, Q r For each associated data threshold, Q' r To monitor the value of each associated data when no abnormality occurs in the target, r takes a value from 1 to e.

5. The computer-based visual detection and early warning method according to claim 1, characterized in that: The specific steps of determining whether an abnormality occurs in the monitoring target and issuing an early warning in S400 are: S401, using visual sensors to collect relevant data of the monitored target in real time, and obtain a real-time relevant data set {w'1, w'2, w'3...w' e }, w'1, w'2, w'3...w' e are the 1st, 2nd, 3rd, ..., eth types of real-time associated data values, respectively, where e is a positive integer; S402, compare the collected real-time associated data value with the associated data threshold, when w' r >Q r When the associated data is judged as abnormal associated data, after e times of judgment, t abnormal associated data are obtained, and the abnormal index of the real-time monitoring target is calculated. The formula is: When θ_s≥θ_y, the monitored target is judged to be abnormal and an abnormal warning is issued.

6. The computer-based visual detection and early warning method according to claim 1, characterized in that: The specific steps of performing chain abnormality warning in S500 are: S501, collect the relative distances L_x of all monitoring targets with abnormalities in history, and construct a set (L_x1, L_x2, L_x3...L_x f }, L_x1, L_x2, L_x3...L_x f is the relative distance of the 1st, 2nd, 3rd...fth monitoring targets with abnormalities, where f is a positive integer; calculate the average value of the f relative distances, and take the average value L_x_p as the dangerous distance of the monitoring target; S502: When it is determined that an abnormality occurs in a monitoring target, the distance between the abnormal monitoring target and other monitoring targets within the visual range of the visual sensor is calculated. The spatial coordinates of the abnormal monitoring target are (X1, Y1, Z1), and the spatial coordinates of the other monitoring targets are (X v , Y v , Z v ), v ranges from 1 to u-1, and the formula is: In the formula, L_s is the distance between the abnormal monitoring target and other monitoring targets within the visual range of the visual sensor; S503, when there is a monitoring target with L_s≤L_x_p, it is determined that the abnormal monitoring target will cause an abnormality to occur when the distance is less than L_x_p, resulting in a chain abnormality, and a chain abnormality warning is issued; S504. When issuing an abnormal warning and a chain warning of a monitoring target, simultaneously send the monitoring target positioning data to the maintenance personnel, so that the maintenance personnel can handle the abnormality in time.

7. A computer-based visual detection warning system, characterized in that: The visual detection warning system includes a data collection module, a monitoring target search module, a monitoring target tracking module, a threshold calculation module, a distance calculation module and a warning module; The data collection module is used to collect all data of the monitoring target in history, and to monitor the monitoring target in real time and collect real-time data; The monitoring target search module is used to search for monitoring targets within the visual range of the visual sensor; The monitoring target tracking module is used to mark the monitoring target after finding the monitoring target within the visual range of the visual sensor, and track the coordinates of the monitoring target in the spatial rectangular coordinate system in real time; The threshold calculation module is used to calculate the similarity threshold when searching for the monitoring target within the visual range of the visual sensor and the threshold for judging whether the monitoring target has an abnormality or a chain abnormality; The distance calculation module is used to calculate the relative distance between the abnormal monitoring target and the other monitoring targets; The warning module is used to issue corresponding warnings when abnormalities and chain abnormalities occur in the monitored target, and output the real-time coordinates of the monitored target during the warning.

8. A computer-based visual detection and early warning system according to claim 7, characterized in that: The monitoring target search module is used to obtain feature data of the monitoring target after analyzing all data of the monitoring target in history, judge all objects within the visible range of the visual sensor according to the feature data of the monitoring target, and search for the monitoring target within the visible range of the visual sensor.

9. A computer-based visual detection and early warning system according to claim 7, characterized in that: The threshold calculation module includes a similarity threshold calculation unit, an abnormality index threshold calculation unit and a danger distance calculation module; The similarity threshold calculation unit is used to calculate the similarity threshold according to the historical monitoring target data set and the feature data set; The abnormal index threshold calculation unit is used to collect the values ​​of all associated data when an abnormality occurs in the monitoring target in the history, calculate the average value of all associated data after d times of collection, use the calculated average value as the threshold of each associated data, and calculate the abnormal index threshold of the monitoring target according to the threshold of each associated data and the value of each associated data when the monitoring target does not have an abnormality; The danger distance calculation module is used to collect all relative distances of the monitoring targets that have abnormalities in history, calculate the average value of the relative distances, and use the average value as the danger distance of the monitoring target.

10. A computer-based visual detection and early warning system according to claim 7, characterized in that: The early warning module is used to issue an abnormality warning when it is judged that an abnormality occurs in the monitoring target, calculate the relative distance between the abnormal monitoring target and the remaining monitoring targets when it is judged that an abnormality occurs in the monitoring target, and when it is judged that the relative distance between the monitoring target and the remaining monitoring targets does not exceed the dangerous distance, it is judged that a chain abnormality occurs, and a chain abnormality warning is issued, and the real-time coordinates of the monitoring target are output during the warning.