A video security monitoring system and method based on deep learning

Abstract

The present invention discloses a video security monitoring system and method based on deep learning, which relate to the field of video monitoring technology. The key points of its technical solution include a marking module, which marks all monitoring devices that can monitor a mobile target object within a video monitoring area as execution monitoring devices, analyzes the video image of the mobile target object's behavior to determine whether the mobile target object has a security abnormality; if the mobile target object has a security abnormality, the time when the mobile target object has the security abnormality is marked as the first time; the video image of the mobile target object's behavior monitored at the first time is divided into a monitoring abnormality area, a monitoring safety area one and a monitoring safety area two; the effect is that when the mobile target object has a security abnormality, the security risk factors that cause the security abnormality to the mobile target object can be quickly analyzed.

Description

Technical Field

[0001] The present invention relates to the field of video surveillance technology, and more specifically, to a video security surveillance system and method based on deep learning. Background Art

[0002] With the continuous advancement of science and technology, video security monitoring systems are widely used in urban monitoring, traffic management, commercial security and other fields. With the development of deep learning technology, video security monitoring systems based on deep learning have gradually become mainstream. Video security monitoring systems based on deep learning mainly include computer vision technology, deep learning algorithms, large-scale data sets, hardware acceleration technology and other aspects. The development of these technologies provides strong support for the implementation of video security monitoring systems, making them play an increasingly important role in various application scenarios. Existing video security monitoring systems based on deep learning can only monitor and record moving target objects in the video monitoring area. When a moving target object has a security abnormality, the cause of the moving target object's security abnormality can be manually analyzed by watching the video screen containing the moving target object. Therefore, the existing video security monitoring systems based on deep learning cannot automatically identify and analyze whether a moving target object has a security abnormality, and cannot quickly analyze the security risk factors that cause the security abnormality when a moving target object has a security abnormality. Summary of the invention

[0003] In view of the shortcomings of the prior art, the object of the present invention is to provide a video security monitoring system and method based on deep learning.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A video security monitoring system based on deep learning, comprising:

[0006] The marking module marks all monitoring devices within the video monitoring area that can monitor the moving target object as execution monitoring devices.

[0007] The monitoring and judgment module obtains the video images of the behavior of the mobile target object monitored by each monitoring device in the corresponding monitoring direction, analyzes the video images of the behavior of the mobile target object and determines whether there is any abnormal safety condition of the mobile target object.

[0008] The division module, if there is a security anomaly in the mobile target object, marks the time when the security anomaly occurs in the mobile target object as the first time; divides the video screen of the monitoring mobile target object behavior at the first time into a monitoring anomaly area, a monitoring safety area one and a monitoring safety area two.

[0009] The extraction module extracts the first safety risk factor that can cause a safety abnormal state to the mobile target object within the monitoring abnormal area, extracts the second safety risk factor that can cause a safety abnormal state to the mobile target object within the monitoring safety area one, and extracts the third safety risk factor that can cause a safety abnormal state to the mobile target object within the monitoring safety area two.

[0010] The monitoring and locking module determines the security risk factors that cause security abnormality to the mobile target object within the video monitoring area according to the morphological changes of security impact risk factor one, security impact risk factor two and security impact risk factor three.

[0011] Preferably, analyzing the video footage of the mobile target object's behavior to determine whether the mobile target object has a safety abnormality specifically includes the following steps:

[0012] The moving speed information of the moving target object in the video screen is continuously obtained at a previous moment and a next moment, and the moving speed information of the moving target object in the video screen at a previous moment and a next moment is calculated and processed to obtain a speed change value.

[0013] Compare the speed change value of the moving target object with the preset speed change threshold:

[0014] If the speed change value of the moving target object is greater than a preset speed change threshold, it is determined that there is no safety abnormality condition of the moving target object.

[0015] If the speed change value of the moving target object is less than or equal to a preset speed change threshold, it is obtained whether the moving direction of the moving target object changes during the moving process.

[0016] When the speed change value of the moving target object is less than or equal to a preset speed change threshold, and the moving direction of the moving target object changes during the moving process, it is determined that there is no safety abnormality of the moving target object.

[0017] When the speed change value of the moving target object is less than or equal to a preset speed change threshold, and the moving direction of the moving target object does not change during the moving process, the state feature of the moving target object in the video picture is obtained.

[0018] Determine whether the moving target object has a security anomaly based on the state characteristics of the moving target object in the video image.

[0019] Preferably, judging whether the moving target object has a safety abnormality according to the state characteristics of the moving target object in the video picture specifically includes the following steps:

[0020] Determine whether the state characteristics of the mobile target object meet the standard change characteristics. If the state characteristics of the mobile target object meet the standard change characteristics, the mobile target object does not have a security abnormality. If the state characteristics of the mobile target object do not meet the standard change characteristics, the mobile target object has a security abnormality.

[0021] Preferably, the first time monitoring mobile target object behavior video screen is divided into monitoring abnormal area, monitoring safety area 1 and monitoring safety area 2, specifically including the following steps:

[0022] Get the number of video images showing abnormal security conditions at the first moment.

[0023] If there is only one video frame with abnormal security conditions at the first time, the area in the video frame where the moving target object has an abnormal security condition is divided and processed to obtain a monitoring abnormality area, the area in the video frame where the moving target object does not have an abnormal security condition is divided and processed to obtain a monitoring safety area one, and the remaining video frames where the moving target object does not have an abnormal security condition are divided and processed to obtain a monitoring safety area two.

[0024] If the number of video frames with abnormal security conditions at the first time is at least two, the areas where the moving target object in the corresponding video frames has an abnormal security condition are spliced ​​and combined to obtain a monitoring abnormality area, the areas where the moving target object in the corresponding video frames does not have an abnormal security condition are divided and processed to obtain a monitoring safety area one, and the remaining video frames where the moving target object does not have an abnormal security condition are divided and processed to obtain a monitoring safety area two.

[0025] Preferably, extracting a first safety impact risk factor that can cause a safety abnormal state to a mobile target object within the monitoring abnormal area, extracting a second safety impact risk factor that can cause a safety abnormal state to a mobile target object within the monitoring safety area one, and extracting a third safety impact risk factor that can cause a safety abnormal state to a mobile target object within the monitoring safety area two specifically includes the following steps:

[0026] The factors directly affecting the moving target object inside the monitored abnormal area are obtained and marked as the first influencing factors.

[0027] Obtain the factors affecting the mobile target object by radiation inside the monitoring abnormal area and mark them as first potential influencing factors, and obtain first indirect influencing information and first direct influencing information inside the monitoring abnormal area;

[0028] Determine the first change degree of the first indirect influence information or the first direct influence information on the first potential influence factor, and determine whether the first potential influence factor has a direct effect on the mobile target object through the first change degree; if the first change degree determines that the first potential influence factor has a direct effect on the mobile target object, mark the first potential influence factor as a second influence factor.

[0029] Among them, the first influencing factor and the second influencing factor are combined to form safety influencing risk factor one.

[0030] The factors directly affecting the moving target object in the monitoring abnormal area are obtained and marked as the third influencing factors.

[0031] The factors affecting the monitoring abnormal area caused by radiation inside the monitoring safety area are obtained and marked as second potential influencing factors, and the second indirect influencing information and the second direct influencing information inside the monitoring safety area are obtained.

[0032] Determine the second change degree of the second indirect influence information and the second direct influence information on the second potential influence factor, and determine whether the second potential influence factor has a direct effect on the mobile target object through the second change degree; if the second change degree determines that the second potential influence factor has a direct effect on the mobile target object, mark the second potential influence factor as the fourth influence factor.

[0033] Among them, the third influencing factor and the fourth influencing factor are combined to form the second safety influencing risk factor.

[0034] The factors in the second monitoring safety area that directly affect the moving target object in the monitoring abnormal area are obtained and marked as the fifth influencing factors.

[0035] The factors affecting the monitoring abnormal area caused by radiation inside the second monitoring safety area are obtained and marked as third potential influencing factors, and the third indirect influence information and the third direct influence information inside the second monitoring safety area are obtained.

[0036] Determine the third degree of change of the third indirect influence information and the third direct influence information on the third potential influence factor, and determine whether the third potential influence factor has a direct effect on the mobile target object through the third degree of change; if the third degree of change determines that the third potential influence factor has a direct effect on the mobile target object, mark the third potential influence factor as the sixth influence factor.

[0037] Among them, the fifth influencing factor and the sixth influencing factor are combined to form safety influencing risk factor three.

[0038] Preferably, determining the security risk factors causing a security abnormality state to the mobile target object inside the video surveillance area according to the morphological changes of the security impact risk factor 1, the security impact risk factor 2 and the security impact risk factor 3 specifically includes the following steps:

[0039] The first morphological characteristic and the second morphological characteristic of the safety influencing risk factor one are obtained at the first time and the second time respectively, the third morphological characteristic and the fourth morphological characteristic of the safety influencing risk factor two are obtained at the first time and the second time respectively, and the fifth morphological characteristic and the sixth morphological characteristic of the safety influencing risk factor three are obtained at the first time and the second time respectively; wherein the second time is earlier than the first time.

[0040] Compare the first morphological feature and the second morphological feature to obtain the first degree of change of the morphological feature of the internal safety impact risk factor one in the monitored abnormal area, compare the third morphological feature and the fourth morphological feature to obtain the second degree of change of the morphological feature of the internal safety impact risk factor two in the monitored safe area one, and compare the fifth morphological feature and the sixth morphological feature to obtain the third degree of change of the morphological feature of the internal safety impact risk factor two in the monitored safe area two.

[0041] The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined through the morphological feature change degree one, morphological feature change degree two and morphological feature change degree three.

[0042] Preferably, determining the security risk factors causing a security abnormality state to the mobile target object inside the video surveillance area through the first degree of change of the morphological feature, the second degree of change of the morphological feature, and the third degree of change of the morphological feature specifically includes the following steps:

[0043] If only one of the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exists, the corresponding item will be marked as the characteristic target morphological feature change degree, and the security risk factor causing security abnormality to the mobile target object within the video surveillance area will be determined based on the characteristic target morphological feature change degree.

[0044] If only two of the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exist, the corresponding two items will be marked as the first target morphological feature change degree and the second target morphological feature change degree respectively, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined in combination with the magnitude of the first target morphological feature change degree and the second target morphological feature change degree.

[0045] If all three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exist, then the sizes of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three are judged, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined by combining the sizes of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three.

[0046] Preferably, the security risk factors causing a security abnormality to the mobile target object within the video surveillance area are determined in combination with the magnitude of the change degree of the first target morphological feature and the change degree of the second target morphological feature, specifically:

[0047] Compare the change degree of the first target morphological feature and the change degree of the second target morphological feature.

[0048] If the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are equal, then the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are both determined as the degree of change of the characteristic target morphological feature, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the degree of change of the characteristic target morphological feature.

[0049] If the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are not equal, the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are marked in order from large to small as the first degree of change and the second degree of change, and the first safety and security factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the first degree of change, and the second safety and security factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the second degree of change, wherein the first safety factor and the second safety factor are combined to form a safety risk factor.

[0050] Preferably, the security risk factors causing a security abnormality state to the mobile target object inside the video surveillance area are determined in combination with the magnitudes of the first morphological feature change degree, the second morphological feature change degree and the third morphological feature change degree, specifically including the following steps:

[0051] Compare the sizes of the first degree of morphological feature change, the second degree of morphological feature change, and the third degree of morphological feature change.

[0052] If the sizes of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are equal, then the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are determined as the characteristic target morphological feature change degree, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the characteristic target morphological feature change degree.

[0053] If only two of the morphological feature change degrees one, two and three are equal in size, the largest of the morphological feature change degrees one, two and three is marked as the first change degree, and the rest of the morphological feature change degrees one, two and three are marked as the second change degree. The first safety and danger factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the first change degree, and the second safety and danger factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the second change degree, wherein the first safety and danger factor and the second safety and danger factor are combined to form a safety risk factor.

[0054] If the sizes of three of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are not equal, then the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are marked as the first change degree, the second change degree and the third change degree in order from large to small, and the first safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the first change degree, the second safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the second change degree, and the third safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the third change degree, wherein the first safety and danger factor, the second safety and danger factor and the third safety and danger factor are combined to form a safety risk factor.

[0055] A video security monitoring method based on deep learning, the method comprising the following steps:

[0056] All monitoring devices within the video monitoring area that can monitor the moving target object are marked as execution monitoring devices.

[0057] The video images of the behaviors of the mobile target objects monitored by each monitoring device in the corresponding monitoring direction are obtained, and the video images of the behaviors of the mobile target objects are analyzed to determine whether there are any abnormal safety conditions of the mobile target objects.

[0058] If the mobile target object has a security anomaly, the time when the mobile target object has the security anomaly is marked as the first time; the video screen of the mobile target object behavior monitored at the first time is divided into a monitoring anomaly area, a monitoring safety area one and a monitoring safety area two.

[0059] A first safety risk factor that can cause a safety abnormal state to the mobile target object is extracted within the monitoring abnormal area. A second safety risk factor that can cause a safety abnormal state to the mobile target object is extracted within the monitoring safety area one. A third safety risk factor that can cause a safety abnormal state to the mobile target object is extracted within the monitoring safety area two.

[0060] The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the morphological changes of security impact risk factor one, security impact risk factor two and security impact risk factor three.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] In the present invention, different execution monitoring devices are used to monitor the moving target object in the video monitoring area in the corresponding monitoring directions to form different video images, and the present invention can analyze the video image of the behavior of the moving target object to determine whether the moving target object has a security abnormality. When the mobile target object has a security abnormality, the video image is divided into a monitoring abnormality area, a monitoring safety area one and a monitoring safety area two, and the security impact risk factor one, the security impact risk factor two and the security impact risk factor three that can cause a security abnormality to the mobile target object are extracted from the monitoring abnormality area, the monitoring safety area one and the monitoring safety area two respectively. The security risk factors that cause a security abnormality to the mobile target object in the video monitoring area are determined by changing the morphology of the security impact risk factor one, the security impact risk factor two and the security impact risk factor three. Therefore, the present invention can quickly analyze the security risk factors that cause a security abnormality to the mobile target object when the mobile target object has a security abnormality. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A module schematic diagram of a video security monitoring system based on deep learning is proposed for the present invention;

[0064] Figure 2 A flowchart of a deep learning-based video security monitoring method is proposed for the present invention. DETAILED DESCRIPTION

[0065] Reference Figure 1 to Figure 2 .

[0066] Embodiment 1 further illustrates a deep learning-based video security monitoring system proposed in the present invention.

[0067] A video security monitoring system based on deep learning, comprising:

[0068] The marking module marks all monitoring devices within the video monitoring area that can monitor the moving target object as execution monitoring devices.

[0069] The monitoring and judgment module obtains the video images of the behavior of the mobile target object monitored by each monitoring device in the corresponding monitoring direction, analyzes the video images of the behavior of the mobile target object and determines whether there is any abnormal safety condition of the mobile target object.

[0070] The division module, if there is a security anomaly in the mobile target object, marks the time when the security anomaly occurs in the mobile target object as the first time; divides the video screen of the monitoring mobile target object behavior at the first time into a monitoring anomaly area, a monitoring safety area one and a monitoring safety area two.

[0071] The extraction module extracts the first safety risk factor that can cause a safety abnormal state to the mobile target object within the monitoring abnormal area, extracts the second safety risk factor that can cause a safety abnormal state to the mobile target object within the monitoring safety area one, and extracts the third safety risk factor that can cause a safety abnormal state to the mobile target object within the monitoring safety area two.

[0072] The monitoring and locking module determines the security risk factors that cause security abnormality to the mobile target object within the video monitoring area according to the morphological changes of security impact risk factor one, security impact risk factor two and security impact risk factor three.

[0073] In this application, all monitoring devices that can monitor mobile target objects within the video monitoring area are marked as execution monitoring devices, where the mobile target objects can be pedestrians or moving vehicles within the video monitoring area. Since there are multiple monitoring devices within the video monitoring area to monitor pedestrians or moving vehicles at different angles, all monitoring devices that can monitor mobile target objects such as pedestrians or moving vehicles are marked as execution monitoring devices.

[0074] Since each executive monitoring device has a corresponding monitoring direction, different executive monitoring devices monitor the moving target object in the video monitoring area in the corresponding monitoring direction to form different video images. By analyzing the video images of the moving target object's behavior, it can be determined whether there is a security anomaly in the moving target object within the video monitoring area.

[0075] If a security abnormality occurs to a mobile target object inside the video surveillance area, the time when the security abnormality occurs to the mobile target object will be marked as the first time, and the video screen will be divided into a monitoring abnormality area, a monitoring safety area one, and a monitoring safety area two at this time, and security impact risk factor one, security impact risk factor two, and security impact risk factor three that can cause a security abnormality to the mobile target object will be extracted from the monitoring abnormality area, the monitoring safety area one, and the monitoring safety area two respectively, and it will be determined whether the forms of security impact risk factor one, security impact risk factor two, and security impact risk factor three have changed; if the forms of security impact risk factor one, security impact risk factor two, and security impact risk factor three have changed, then the security risk factors that cause a security abnormality to the mobile target object inside the video surveillance area are determined.

[0076] For example, if a pedestrian in the video surveillance area falls down in a safety abnormality state, then at this time, the safety impact risk factor one, safety impact risk factor two and safety impact risk factor three of the monitoring abnormality area, monitoring safety area one and monitoring safety area two are judged respectively. If the safety impact risk factor two (such as broken branches) inside the monitoring safety area one causes the pedestrian to fall in a safety abnormality state, then it can be determined that the safety risk factor that causes the safety abnormality state of the moving target object inside the video surveillance area is the broken branch.

[0077] Analyzing the video footage of the mobile target object's behavior to determine whether the mobile target object has a safety abnormality, specifically includes the following steps:

[0078] The moving speed information of the moving target object in the video screen is continuously obtained at a previous moment and a next moment, and the moving speed information of the moving target object in the video screen at a previous moment and a next moment is calculated and processed to obtain a speed change value.

[0079] Compare the speed change value of the moving target object with the preset speed change threshold:

[0080] If the speed change value of the moving target object is greater than a preset speed change threshold, it is determined that there is no safety abnormality condition of the moving target object.

[0081] If the speed change value of the moving target object is less than or equal to a preset speed change threshold, it is obtained whether the moving direction of the moving target object changes during the moving process.

[0082] When the speed change value of the moving target object is less than or equal to a preset speed change threshold, and the moving direction of the moving target object changes during the moving process, it is determined that there is no safety abnormality of the moving target object.

[0083] When the speed change value of the moving target object is less than or equal to a preset speed change threshold, and the moving direction of the moving target object does not change during the moving process, the state feature of the moving target object in the video picture is obtained.

[0084] Determine whether the moving target object has a security anomaly based on the state characteristics of the moving target object in the video image.

[0085] It should be noted that the present application first determines whether there is any abnormal safety condition in the moving target object by continuously acquiring the moving speed information of the moving target object in the video screen at the previous moment and the next moment, and compares the speed change value of the moving target object in the video screen at the previous moment and the next moment with the preset speed change threshold. The preset speed change threshold here can be adjusted according to the actual situation. For example, the moving target object here is a pedestrian. The moving speed of the pedestrian in the video screen of the previous second is 1.2m / s, and the moving speed of the pedestrian in the video screen of the next second is 1m / s. Here, the speed change value of the pedestrian in the walking state is -0.2m / s. 2 Here, the preset speed change threshold is set to -0.6m / s 2 , it can be seen that the speed change value of the pedestrian in the walking state is -0.2m / s 2 Greater than the preset speed change threshold -0.6m / s 2 , it is judged that there is no abnormal safety condition of the moving target object, that is, the pedestrian is in a safe state in the video surveillance area.

[0086] For example, the moving target object here is a pedestrian. The pedestrian's walking speed in the previous second of the video is 1.2m / s, and the pedestrian's walking speed in the next second of the video is 0.1m / s. Here, the speed change value of the pedestrian in the walking state is -1.1m / s 2 Here, the preset speed change threshold is set to -0.6m / s 2 , it can be seen that the speed change value of the pedestrian in the walking state is -1.1m / s 2 Less than the preset speed change threshold -0.6m / s 2 , it is necessary to determine whether the pedestrian's walking direction changes during walking. If the pedestrian's walking direction changes during walking, it is normal for pedestrians to reduce their speed when turning. Therefore, it is determined that there is no abnormal safety condition in the moving target object, that is, the pedestrian is in a safe state in the video surveillance area. If the pedestrian's speed change value in the walking state is -1.1m / s 2 Less than the preset speed change threshold -0.6m / s 2 , and the pedestrian's walking direction does not change during walking, it is necessary to combine the state characteristics of the pedestrian in the video image to determine whether the pedestrian has a safety abnormality.

[0087] Judging whether the moving target object has a security abnormality according to the state characteristics of the moving target object in the video screen specifically includes the following steps:

[0088] Determine whether the state characteristics of the mobile target object meet the standard change characteristics. If the state characteristics of the mobile target object meet the standard change characteristics, the mobile target object does not have a security abnormality. If the state characteristics of the mobile target object do not meet the standard change characteristics, the mobile target object has a security abnormality.

[0089] It should be noted that if the pedestrian's speed changes by -1.1m / s 2 Less than the preset speed change threshold -0.6m / s 2 , and at this time, the pedestrian's walking direction has not changed during the walking process, it is necessary to combine the state characteristics of the pedestrian in the video screen to determine whether the pedestrian has a safety abnormality. Here, the standard change characteristics of the pedestrian can be conventional safe action characteristics such as squatting, long jumping, and spreading arms. If the state characteristics of the pedestrian are conventional safe action characteristics of spreading arms, it means that the pedestrian does not have a safety abnormality, that is, the pedestrian is in a safe state in the video surveillance area. If the state characteristics of the pedestrian are the state characteristics of wrestling, since the state characteristics of wrestling do not meet the standard change characteristics, it means that the pedestrian has a safety abnormality, that is, the pedestrian is in a dangerous state in the video surveillance area.

[0090] The first-time monitoring video screen of the mobile target object behavior is divided into a monitoring abnormal area, a monitoring safety area 1 and a monitoring safety area 2, specifically including the following steps:

[0091] Get the number of video images showing abnormal security conditions at the first moment.

[0092] If there is only one video frame with abnormal security conditions at the first time, the area in the video frame where the moving target object has an abnormal security condition is divided and processed to obtain a monitoring abnormality area, the area in the video frame where the moving target object does not have an abnormal security condition is divided and processed to obtain a monitoring safety area one, and the remaining video frames where the moving target object does not have an abnormal security condition are divided and processed to obtain a monitoring safety area two.

[0093] If the number of video frames with abnormal security conditions at the first time is at least two, the areas where the moving target object in the corresponding video frames has an abnormal security condition are spliced ​​and combined to obtain a monitoring abnormality area, the areas where the moving target object in the corresponding video frames does not have an abnormal security condition are divided and processed to obtain a monitoring safety area one, and the remaining video frames where the moving target object does not have an abnormal security condition are divided and processed to obtain a monitoring safety area two.

[0094] It should be noted that here, according to the number of video images of safety abnormalities existing in the first time, the video images of monitoring the behavior of the moving target object at the first time are divided into monitoring abnormality areas, monitoring safety area one and monitoring safety area two. Since the number of monitoring devices that can monitor pedestrians in safety abnormality states may be different in different video monitoring areas, if the pedestrian's safety abnormality state is that the front side of the body is hit by a stone, and there are front execution monitoring devices and rear execution monitoring devices, only the front execution monitoring device can capture the video image of the pedestrian in safety abnormality state. Therefore, at this time, the number of video images of safety abnormality is one. Then, the area in the video image where the moving target object has a safety abnormality state is divided and processed to obtain the monitoring abnormality area, and the area in the video image where the moving target object does not have a safety abnormality state is divided and processed to obtain the monitoring safety area one. The rear monitoring device captures the video image where there is no pedestrian safety abnormality state, so these video images are marked as monitoring safety area two.

[0095] If the pedestrian's abnormal safety state is that the right arm of the body is hit by a stone, and there are front, rear, left and right execution monitoring devices, then the front, rear and right execution monitoring devices can all capture video footage of the pedestrian in an abnormal safety state. Therefore, there are three video footages of the abnormal safety state at this time. The areas in the three video footages where the pedestrian is in an abnormal safety state are spliced ​​and combined to obtain an abnormal monitoring area, and the areas in the three video footages where the moving target object does not have an abnormal safety state are divided and processed to obtain a first monitoring safety area. Since the left side execution monitoring device did not capture the video footage of the pedestrian in an abnormal safety state, the video footage captured by the left side execution monitoring device is marked as a second monitoring safety area.

[0096] Extracting a first safety impact risk factor that can cause a safety abnormal state to a mobile target object within the monitoring abnormal area, extracting a second safety impact risk factor that can cause a safety abnormal state to a mobile target object within the monitoring safety area one, and extracting a third safety impact risk factor that can cause a safety abnormal state to a mobile target object within the monitoring safety area two, specifically comprising the following steps:

[0097] The factors directly affecting the moving target object inside the monitored abnormal area are obtained and marked as the first influencing factors.

[0098] Obtain the factors affecting the mobile target object by radiation inside the monitoring abnormal area and mark them as first potential influencing factors, and obtain first indirect influencing information and first direct influencing information inside the monitoring abnormal area;

[0099] Determine the first change degree of the first indirect influence information or the first direct influence information on the first potential influence factor, and determine whether the first potential influence factor has a direct effect on the mobile target object through the first change degree; if the first change degree determines that the first potential influence factor has a direct effect on the mobile target object, mark the first potential influence factor as a second influence factor.

[0100] Among them, the first influencing factor and the second influencing factor are combined to form the first safety influencing risk factor.

[0101] It should be noted that there are factors directly affecting the moving target object (pedestrian) within the monitoring abnormal area. For example, if a pedestrian is walking on a bumpy road surface, he may fall. Therefore, the road surface can be regarded as the first influencing factor.

[0102] If the moving target object inside the monitoring abnormal area is a pedestrian, the pedestrian may be injured due to the breakage or shaking of the branches, and therefore the branches are the factor that has a radiation impact on the pedestrian, that is, the first potential influencing factor is the branches, and the indirect impact information on the breakage of the branches inside the monitoring abnormal area may be gas flow, where the first indirect impact information is gas flow. Since there may be a situation where people shake the branches to pick fruits inside the monitoring abnormal area, the first direct impact information here may be the shaking of the branches by humans.

[0103] When the first indirect influencing information is gas flow, the first degree of change of the gas flow on the branches is judged. Here, the first degree of change is the shaking of the branches. If the gas flow can cause the branches to shake, the first degree of change determines that the first potential influencing factor can have a direct effect on the moving target object. Here, the second influencing factor is the branches.

[0104] When the direct influencing information may be artificial shaking of branches, the first degree of change of the branches caused by artificial shaking of branches is determined. Here, the first degree of change is the shaking of branches. If artificial shaking of branches can cause the branches to shake (sometimes because the branches are relatively thick and the force of artificial shaking of branches is relatively small, the branches do not shake), then the first degree of change determines that the first direct influencing factor can have a direct effect on the moving target object. Here, the second influencing factor is artificial shaking of branches.

[0105] The factors directly affecting the moving target object in the monitoring abnormal area are obtained and marked as the third influencing factors.

[0106] The factors affecting the monitoring abnormal area caused by radiation inside the monitoring safety area are obtained and marked as second potential influencing factors, and the second indirect influencing information and the second direct influencing information inside the monitoring safety area are obtained.

[0107] Determine the second change degree of the second indirect influence information and the second direct influence information on the second potential influence factor, and determine whether the second potential influence factor has a direct effect on the mobile target object through the second change degree; if the second change degree determines that the second potential influence factor has a direct effect on the mobile target object, mark the second potential influence factor as the fourth influence factor.

[0108] Among them, the third influencing factor and the fourth influencing factor combine to form the second safety risk factor.

[0109] It should be noted that the method for determining the second safety impact risk factor within the monitored safety area one is the same as the method for determining the first safety impact risk factor within the monitored abnormal area, so examples will not be repeated here.

[0110] The factors in the second monitoring safety area that directly affect the moving target object in the monitoring abnormal area are obtained and marked as the fifth influencing factors.

[0111] The factors affecting the monitoring abnormal area caused by radiation inside the second monitoring safety area are obtained and marked as third potential influencing factors, and the third indirect influence information and the third direct influence information inside the second monitoring safety area are obtained.

[0112] Determine the third degree of change of the third indirect influence information and the third direct influence information on the third potential influence factor, and determine whether the third potential influence factor has a direct effect on the mobile target object through the third degree of change; if the third degree of change determines that the third potential influence factor has a direct effect on the mobile target object, mark the third potential influence factor as the sixth influence factor.

[0113] Among them, the fifth influencing factor and the sixth influencing factor combine to form the safety risk factor three.

[0114] It should be noted that the method for determining the safety impact risk factor three within the monitoring safety area two is the same as the method for determining the safety impact risk factor one within the monitoring abnormal area, so the examples will not be repeated here.

[0115] According to the morphological changes of the first safety impact risk factor, the second safety impact risk factor and the third safety impact risk factor, the safety risk factors causing the abnormal safety state of the mobile target object in the video surveillance area are determined, which specifically includes the following steps:

[0116] The first morphological characteristic and the second morphological characteristic of the safety influencing risk factor one are obtained at the first time and the second time respectively, the third morphological characteristic and the fourth morphological characteristic of the safety influencing risk factor two are obtained at the first time and the second time respectively, and the fifth morphological characteristic and the sixth morphological characteristic of the safety influencing risk factor three are obtained at the first time and the second time respectively; wherein the second time is earlier than the first time.

[0117] Compare the first morphological feature and the second morphological feature to obtain the first degree of change of the morphological feature of the internal safety impact risk factor one in the monitored abnormal area, compare the third morphological feature and the fourth morphological feature to obtain the second degree of change of the morphological feature of the internal safety impact risk factor two in the monitored safe area one, and compare the fifth morphological feature and the sixth morphological feature to obtain the third degree of change of the morphological feature of the internal safety impact risk factor two in the monitored safe area two.

[0118] The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined through the morphological feature change degree one, morphological feature change degree two and morphological feature change degree three.

[0119] It should be noted that here, the security risk factors that cause security abnormal states to the mobile target objects within the video surveillance area are determined through the first degree of change of morphological characteristics, the second degree of change of morphological characteristics and the third degree of change of morphological characteristics. The security risk factors that cause security abnormal states to the mobile target objects within the video surveillance area can be determined based on the first degree of change of morphological characteristics of the security impact risk factor one within the monitoring abnormal area, the second degree of change of morphological characteristics of the security impact risk factor two within the monitoring safety area one and the third degree of change of morphological characteristics of the security impact risk factor three within the monitoring safety area two.

[0120] For example, if there is only a degree of change in the morphological characteristics of the security risk factor 1 inside the monitoring abnormal area, the security risk factor that causes the security abnormal state to the mobile target object inside the video monitoring area is determined by the degree of change in the morphological characteristics of the security risk factor 1 inside the monitoring abnormal area.

[0121] For example, if the first safety risk factor influencing the abnormal area of ​​the monitoring is the walking surface, and the first morphological feature of the safety risk factor influencing the first at the first time is the flat walking surface, and the second morphological feature of the safety risk factor influencing the first at the second time is the potholes in the walking surface, then by comparing the first morphological feature and the second morphological feature, it can be obtained that there is a degree of morphological feature change of the safety risk factor influencing the abnormal area of ​​the monitoring, then it can be determined that the safety risk factor causing the safety abnormal state of the moving target object in the video monitoring area is the safety risk factor influencing the first, that is, the safety risk factor causing the safety abnormal state of the moving target object is the walking surface.

[0122] The security risk factors that cause security abnormality to the mobile target object in the video surveillance area are determined by the first degree of change of the morphological feature, the second degree of change of the morphological feature, and the third degree of change of the morphological feature, specifically including the following steps:

[0123] If only one of the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exists, the corresponding item will be marked as the characteristic target morphological feature change degree, and the security risk factor causing security abnormality to the mobile target object within the video surveillance area will be determined based on the characteristic target morphological feature change degree.

[0124] If only two of the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exist, the corresponding two items will be marked as the first target morphological feature change degree and the second target morphological feature change degree respectively, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined in combination with the magnitude of the first target morphological feature change degree and the second target morphological feature change degree.

[0125] If all three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exist, then the sizes of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three are judged, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined by combining the sizes of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three.

[0126] It should be noted that by conducting situation analysis on different numbers of morphological feature change degrees one, morphological feature change degrees two and morphological feature change degrees three, it is possible to more accurately determine the security risk factors that cause security abnormalities to mobile target objects within the video surveillance area.

[0127] If only one of the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exists, the corresponding item will be marked as the characteristic target morphological feature change degree, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the characteristic target morphological feature change degree. For example, if only the morphological feature change degree one of the security impact risk factor one within the abnormal monitoring area exists, the morphological feature change degree one will be marked as the characteristic target morphological feature change degree, that is, the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the morphological feature change degree one.

[0128] For example, the first safety impact risk factor inside the monitoring abnormal area is the walking surface, and the first morphological feature of the safety impact risk factor one at the first time is the flat walking surface, and the second morphological feature of the safety impact risk factor one at the second time is the potholes on the walking surface. By comparing the first morphological feature and the second morphological feature, it is obtained that there is a degree of morphological feature change one for the safety impact risk factor one inside the monitoring abnormal area. At this time, the degree of morphological feature change one is marked as the characteristic target morphological feature change degree, and it can be determined that the safety impact risk factor one is the walking surface, that is, the safety risk factor that causes the safety abnormal state of the mobile target object is the walking surface in the monitoring abnormal area.

[0129] The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined by combining the change degree of the first target morphological feature and the change degree of the second target morphological feature, specifically:

[0130] Compare the change degree of the first target morphological feature and the change degree of the second target morphological feature.

[0131] If the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are equal, then the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are both determined as the degree of change of the characteristic target morphological feature, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the degree of change of the characteristic target morphological feature.

[0132] It should be noted that when only two of the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exist, firstly, two of the existing ones are screened out from the three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three, and the two existing ones are respectively marked as the first target morphological feature change degree and the second target morphological feature change degree, and then the sizes of the first target morphological feature change degree and the second target morphological feature change degree are compared. If the sizes of the first target morphological feature change degree and the second target morphological feature change degree are equal, then the first target morphological feature change degree and the second target morphological feature change degree are both determined as characteristic target morphological feature change degrees, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined according to the characteristic target morphological feature change degrees.

[0133] For example, when there is only a first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormal area and a second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area, they are marked as the first target morphological characteristics change degree and the second target morphological characteristics change degree respectively. If the first target morphological characteristics change degree of the walking road surface inside the monitoring abnormal area and the second target morphological characteristics change degree of the walking road surface inside the monitoring safety area are the same, the first target morphological characteristics change degree and the second target morphological characteristics change degree are determined as the target morphological characteristics change degree, thereby determining that the walking road surface inside the monitoring abnormal area and the walking road surface inside the monitoring safety area are safety risk factors that cause a safety abnormality state to the mobile target object;

[0134] If the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are not equal, the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are marked in order from large to small as the first degree of change and the second degree of change, and the first safety and security factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the first degree of change, and the second safety and security factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the second degree of change, wherein the first safety factor and the second safety factor are combined to form a safety risk factor.

[0135] It should be noted that, when the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are not equal, it is necessary to mark the degree of change of the first target morphological feature and the degree of change of the second target morphological feature in order from large to small as the first degree of change and the second degree of change, so that it can be judged that the first security factor corresponding to the first degree of change has a greater impact on the security abnormality state caused to the mobile target object within the video surveillance area, while the first security factor corresponding to the second degree of change has a smaller impact on the security abnormality state caused to the mobile target object within the video surveillance area.

[0136] For example, when there is only a first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormal area and a second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area, they are marked as the first target morphological characteristic change degree and the second target morphological characteristic change degree respectively. If the first target morphological characteristic change degree of the walking road surface inside the monitoring abnormal area and the second target morphological characteristic change degree of the walking road surface inside the monitoring safety area are not equal, the first target morphological characteristic change degree and the second target morphological characteristic change degree are marked as the first change degree and the second change degree in order from large to small. If the first target morphological characteristic change degree of the walking road surface inside the monitoring abnormal area is greater than the second target morphological characteristic change degree of the walking road surface inside the monitoring safety area, the first safety factor is determined to be the walking road surface inside the monitoring abnormal area, and the second safety factor is the walking road surface inside the monitoring safety area.

[0137] The security risk factors causing security abnormality to the mobile target object within the video surveillance area are determined by combining the magnitudes of the first, second and third morphological feature change degrees, specifically including the following steps:

[0138] Compare the sizes of the first degree of morphological feature change, the second degree of morphological feature change, and the third degree of morphological feature change.

[0139] If the sizes of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are equal, then the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are determined as the characteristic target morphological feature change degree, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the characteristic target morphological feature change degree.

[0140] For example, if the first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormal area, the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one, and the third degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area two are all the same, then the first degree of change in the morphological characteristics, the second degree of change in the morphological characteristics, and the third degree of change in the morphological characteristics are determined as the degree of change in the morphological characteristics of the characteristic targets, thereby determining the safety risk factors of the walking road surface inside the monitoring abnormal area, the walking road surface inside the monitoring safety area one, and the walking road surface inside the monitoring safety area two causing safety abnormality to the moving target object.

[0141] If only two of the morphological feature change degrees one, two and three are equal in size, the largest of the morphological feature change degrees one, two and three is marked as the first change degree, and the rest of the morphological feature change degrees one, two and three are marked as the second change degree. The first safety and danger factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the first change degree, and the second safety and danger factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the second change degree, wherein the first safety and danger factor and the second safety and danger factor are combined to form a safety risk factor.

[0142] For example, when only two of the first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormality area, the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one, and the third degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area two are equal, and if the first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormality area is equal to the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one, and the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one is greater than the third degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area two, then the walking road surfaces inside the monitoring abnormality area and the monitoring safety area one are determined as the first safety and risk factor, and the walking road surface inside the monitoring safety area two is determined as the second safety and risk factor.

[0143] If the sizes of three of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are not equal, then the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are marked as the first change degree, the second change degree and the third change degree in order from large to small, and the first safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the first change degree, the second safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the second change degree, and the third safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the third change degree, wherein the first safety and danger factor, the second safety and danger factor and the third safety and danger factor are combined to form a safety risk factor.

[0144] For example, the first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormality area, the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one, and the third degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area two are not equal, and if the first degree of change in the morphological characteristics of the walking road surface inside the monitoring abnormality area is greater than the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one, if the second degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area one is greater than the third degree of change in the morphological characteristics of the walking road surface inside the monitoring safety area two, then the walking road surface inside the monitoring abnormality area is determined as the first safety and risk factor, the walking road surface inside the monitoring safety area one is determined as the second safety and risk factor, and the walking road surface inside the monitoring safety area two is determined as the third safety and risk factor.

[0145] Example 2 further illustrates a deep learning-based video security monitoring method proposed in the present invention.

[0146] A video security monitoring method based on deep learning, the method comprising the following steps:

[0147] All monitoring devices within the video monitoring area that can monitor the moving target object are marked as execution monitoring devices.

[0148] The video images of the behavior of the mobile target object monitored by each monitoring device in the corresponding monitoring direction are obtained, and the video images of the behavior of the mobile target object are analyzed to determine whether there is a safety abnormality in the mobile target object.

[0149] If the mobile target object has a security anomaly, the time when the mobile target object has the security anomaly is marked as the first time; the video screen of the mobile target object behavior monitored at the first time is divided into a monitoring anomaly area, a monitoring safety area one and a monitoring safety area two.

[0150] A first safety risk factor that can cause a safety abnormal state to the mobile target object is extracted within the monitoring abnormal area. A second safety risk factor that can cause a safety abnormal state to the mobile target object is extracted within the monitoring safety area one. A third safety risk factor that can cause a safety abnormal state to the mobile target object is extracted within the monitoring safety area two.

[0151] The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the morphological changes of security impact risk factor one, security impact risk factor two and security impact risk factor three.

[0152] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0153] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A video security monitoring system based on deep learning, characterized in that: include: A marking module marks all monitoring devices within the video monitoring area that can monitor the moving target object as execution monitoring devices; The monitoring and judgment module obtains the video images of the behaviors of the mobile target objects monitored by each monitoring device in the corresponding monitoring direction, and analyzes the video images of the behaviors of the mobile target objects to determine whether there is any abnormal safety condition of the mobile target objects; The division module marks the time when the mobile target object has the security abnormality as the first time if the mobile target object has the security abnormality; Divide the first-time monitoring video screen of the mobile target object behavior into a monitoring abnormality area, a monitoring safety area 1 and a monitoring safety area 2; An extraction module extracts a first safety impact risk factor that can cause a safety abnormal state to a mobile target object within a monitoring abnormal area, extracts a second safety impact risk factor that can cause a safety abnormal state to a mobile target object within a monitoring safety area one, and extracts a third safety impact risk factor that can cause a safety abnormal state to a mobile target object within a monitoring safety area two; Obtain the factors that directly affect the moving target object in the monitored abnormal area and mark them as the first influencing factors; Obtain the factors affecting the mobile target object by radiation inside the monitoring abnormal area and mark them as first potential influencing factors, and obtain first indirect influencing information and first direct influencing information inside the monitoring abnormal area; Determine a first change degree of the first indirect influence information or the first direct influence information on the first potential influence factor, determine whether the first potential influence factor has a direct effect on the mobile target object according to the first change degree, and mark the first potential influence factor as a second influence factor if the first change degree determines that the first potential influence factor has a direct effect on the mobile target object; Wherein, the first influencing factor and the second influencing factor are combined to form a safety influencing risk factor 1; Obtain the factors in the first monitoring safety area that directly affect the moving target object in the monitoring abnormal area and mark them as the third influencing factors; Obtain the factors affecting the monitoring abnormal area caused by radiation inside the monitoring safety area 1 and mark them as second potential influencing factors, and obtain the second indirect influencing information and the second direct influencing information inside the monitoring safety area 1; Determine a second change degree of the second indirect influence information and the second direct influence information on the second potential influence factor, determine whether the second potential influence factor has a direct effect on the mobile target object according to the second change degree, and mark the second potential influence factor as a fourth influence factor if the second change degree determines that the second potential influence factor has a direct effect on the mobile target object; The third influencing factor and the fourth influencing factor are combined to form a second safety influencing risk factor; Obtain the factors in the second monitoring safety area that directly affect the moving target object in the monitoring abnormal area and mark them as the fifth influencing factors; Obtain the factors affecting the monitoring abnormal area caused by radiation inside the monitoring safety area 2 and mark them as third potential influencing factors, and obtain the third indirect influence information and the third direct influence information inside the monitoring safety area 2; Determine a third degree of change of the third indirect influence information and the third direct influence information on the third potential influence factor, determine whether the third potential influence factor has a direct effect on the mobile target object according to the third degree of change, and mark the third potential influence factor as a sixth influence factor if the third degree of change determines that the third potential influence factor has a direct effect on the mobile target object; The fifth influencing factor and the sixth influencing factor are combined to form a safety influencing risk factor three; The monitoring and locking module determines the security risk factors that cause security abnormality to the mobile target object within the video monitoring area according to the morphological changes of security impact risk factor one, security impact risk factor two and security impact risk factor three.

2. A video security monitoring system based on deep learning according to claim 1, characterized in that: Analyzing the video footage of the mobile target object's behavior to determine whether the mobile target object has a safety abnormality, specifically includes the following steps: Continuously acquiring the moving speed information of the moving target object in the video screen at a previous moment and a next moment, and calculating and processing the moving speed information of the moving target object in the video screen at a previous moment and a next moment to obtain a speed change value; Compare the speed change value of the moving target object with the preset speed change threshold: If the speed change value of the moving target object is greater than the preset speed change threshold, it is determined that there is no safety abnormality in the moving target object; If the speed change value of the moving target object is less than or equal to a preset speed change threshold, obtaining whether the moving direction of the moving target object changes during the moving process; When the speed change value of the moving target object is less than or equal to the preset speed change threshold value, and the moving direction of the moving target object changes during the moving process, it is determined that there is no safety abnormality of the moving target object; When the speed change value of the moving target object is less than or equal to a preset speed change threshold value, and the moving direction of the moving target object does not change during the moving process, the state feature of the moving target object in the video picture is obtained; Determine whether the moving target object has a security anomaly based on the state characteristics of the moving target object in the video image.

3. A video security monitoring system based on deep learning according to claim 2, characterized in that: Judging whether the moving target object has a security abnormality according to the state characteristics of the moving target object in the video screen specifically includes the following steps: Determine whether the state characteristics of the mobile target object meet the standard change characteristics. If the state characteristics of the mobile target object meet the standard change characteristics, the mobile target object does not have a security abnormality. If the state characteristics of the mobile target object do not meet the standard change characteristics, the mobile target object has a security abnormality.

4. The deep learning-based video security monitoring system according to claim 1, characterized in that: The first-time monitoring video screen of the mobile target object behavior is divided into a monitoring abnormal area, a monitoring safety area 1 and a monitoring safety area 2, specifically including the following steps: Obtain the number of video images with abnormal safety conditions at the first time; If there is one video screen with a security anomaly at the first time, the area in the video screen where the moving target object has a security anomaly is divided and processed to obtain a monitoring anomaly area, the area in the video screen where the moving target object does not have a security anomaly is divided and processed to obtain a monitoring safety area 1, and the remaining video screens where the moving target object does not have a security anomaly are divided and processed to obtain a monitoring safety area 2; If the number of video frames with abnormal security conditions at the first time is at least two, the areas where the moving target object in the corresponding video frames has an abnormal security condition are spliced ​​and combined to obtain a monitoring abnormality area, the areas where the moving target object in the corresponding video frames does not have an abnormal security condition are divided and processed to obtain a monitoring safety area one, and the remaining video frames where the moving target object does not have an abnormal security condition are divided and processed to obtain a monitoring safety area two.

5. The deep learning-based video security monitoring system according to claim 1, characterized in that: According to the morphological changes of the first safety impact risk factor, the second safety impact risk factor and the third safety impact risk factor, the safety risk factors causing the abnormal safety state of the mobile target object in the video surveillance area are determined, which specifically includes the following steps: The first morphological feature and the second morphological feature of the safety impact risk factor 1 at the first time and the second time are obtained respectively, the third morphological feature and the fourth morphological feature of the safety impact risk factor 2 at the first time and the second time are obtained respectively, and the fifth morphological feature and the sixth morphological feature of the safety impact risk factor 3 at the first time and the second time are obtained respectively; wherein the second time is earlier than the first time; The first morphological feature and the second morphological feature are compared to obtain the first degree of change of the morphological feature of the first internal safety impact risk factor of the monitoring abnormal area, the third morphological feature and the fourth morphological feature are compared to obtain the second degree of change of the morphological feature of the second internal safety impact risk factor of the monitoring safety area one, and the fifth morphological feature and the sixth morphological feature are compared to obtain the third degree of change of the morphological feature of the second internal safety impact risk factor of the monitoring safety area two; The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined through the morphological feature change degree one, morphological feature change degree two and morphological feature change degree three.

6. A video security monitoring system based on deep learning according to claim 5, characterized in that: The security risk factors that cause security abnormality to the mobile target object in the video surveillance area are determined by the first degree of change of the morphological feature, the second degree of change of the morphological feature, and the third degree of change of the morphological feature, specifically including the following steps: If only one of the three items of morphological feature change degree 1, morphological feature change degree 2 and morphological feature change degree 3 exists, the corresponding item is marked as the characteristic target morphological feature change degree, and the security risk factor causing the security abnormal state of the mobile target object inside the video surveillance area is determined according to the characteristic target morphological feature change degree; If only two of the three items of morphological feature change degree 1, morphological feature change degree 2 and morphological feature change degree 3 exist, the corresponding two items are marked as the first target morphological feature change degree and the second target morphological feature change degree, respectively, and the security risk factors causing security abnormality to the mobile target object in the video surveillance area are determined by combining the magnitudes of the first target morphological feature change degree and the second target morphological feature change degree; If all three items of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three exist, then the sizes of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three are judged, and the security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined by combining the sizes of morphological feature change degree one, morphological feature change degree two and morphological feature change degree three.

7. A video security monitoring system based on deep learning according to claim 6, characterized in that: The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined by combining the change degree of the first target morphological feature and the change degree of the second target morphological feature, specifically: Compare the magnitude of the change degree of the first target morphological feature and the change degree of the second target morphological feature; If the magnitude of the change degree of the first target morphological feature and the change degree of the second target morphological feature are equal, the change degree of the first target morphological feature and the change degree of the second target morphological feature are both determined as the change degree of the characteristic target morphological feature, and the security risk factor causing the security abnormal state of the mobile target object inside the video surveillance area is determined according to the change degree of the characteristic target morphological feature; If the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are not equal, the degree of change of the first target morphological feature and the degree of change of the second target morphological feature are marked in order from large to small as the first degree of change and the second degree of change, and the first safety and security factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the first degree of change, and the second safety and security factor causing an abnormal safety state to the mobile target object within the video surveillance area is determined based on the second degree of change, wherein the first safety factor and the second safety factor are combined to form a safety risk factor.

8. The deep learning-based video security monitoring system according to claim 7, characterized in that: The security risk factors causing security abnormality to the mobile target object within the video surveillance area are determined by combining the magnitudes of the first, second and third morphological feature change degrees, specifically including the following steps: Compare the magnitudes of the first degree of change in morphological characteristics, the second degree of change in morphological characteristics, and the third degree of change in morphological characteristics; If the magnitudes of the first morphological feature change degree, the second morphological feature change degree and the third morphological feature change degree are all equal, the first morphological feature change degree, the second morphological feature change degree and the third morphological feature change degree are determined as the characteristic target morphological feature change degree, and the security risk factors causing the security abnormal state to the mobile target object inside the video surveillance area are determined according to the characteristic target morphological feature change degree; If only two of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are equal in magnitude, the largest of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three is marked as the first change degree, and the rest of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three is marked as the second change degree, and the first safety and danger factor causing the safety abnormal state of the mobile target object inside the video surveillance area is determined according to the first change degree, and the second safety and danger factor causing the safety abnormal state of the mobile target object inside the video surveillance area is determined according to the second change degree, wherein the first safety and danger factor and the second safety and danger factor are combined to form a safety risk factor; If the sizes of three of the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are not equal, then the morphological feature change degree one, the morphological feature change degree two and the morphological feature change degree three are marked as the first change degree, the second change degree and the third change degree in order from large to small, and the first safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the first change degree, the second safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the second change degree, and the third safety and danger factor causing the safety abnormal state to the mobile target object within the video surveillance area is determined according to the third change degree, wherein the first safety and danger factor, the second safety and danger factor and the third safety and danger factor are combined to form a safety risk factor.

9. A video security monitoring method based on deep learning, applied to a video security monitoring system based on deep learning according to any one of claims 1 to 8, characterized in that: The method comprises the following steps: Marking all monitoring devices within the video monitoring area that can monitor the moving target object as execution monitoring devices; Obtaining video images of the behavior of the mobile target object monitored by each execution monitoring device in the corresponding monitoring direction, and analyzing the video images of the behavior of the mobile target object to determine whether there is a safety abnormality of the mobile target object; If the mobile target object has a security anomaly, the time when the mobile target object has a security anomaly is marked as the first time; the video screen of the mobile target object behavior monitored at the first time is divided into a monitoring anomaly area, a monitoring safety area one, and a monitoring safety area two; Extract the first safety impact risk factor that can cause a safety abnormal state to the mobile target object within the monitoring abnormal area, extract the second safety impact risk factor that can cause a safety abnormal state to the mobile target object within the monitoring safety area one, and extract the third safety impact risk factor that can cause a safety abnormal state to the mobile target object within the monitoring safety area two; The security risk factors that cause security abnormality to the mobile target object within the video surveillance area are determined based on the morphological changes of security impact risk factor one, security impact risk factor two and security impact risk factor three.

Images