Camera scheduling method and system adaptive to environment change

Abstract

The invention provides a camera scheduling method and system adaptive to environment change, and relates to the technical field of camera scheduling. A basic Kalman filtering model and an environment compensation model are fused, a state equation is corrected through a visibility compensation coefficient and an illumination abrupt change compensation coefficient, and a multi-stage weighted weight dynamic adjustment mechanism is combined, so that the camera scheduling efficiency is improved. According to the method, trajectory prediction deviation caused by environmental interference is effectively inhibited, and then a screening strategy of combining direction matching and adjacent camera priority is adopted, so that the screening accuracy is improved while the screening efficiency is ensured; then, by pre-preempting resource allocation, synchronizing target tracking parameters to relay candidate cameras, eliminating parameter jump and initialization delay during scheduling relay switching, and in addition, introducing a relay validity verification and anomaly rollback mechanism, realizing rapid recovery when equipment fails or environment is abnormal, improving system robustness, and improving system performance. The beneficial effects of continuity and high reliability of multi-camera cooperative scheduling tracking in a complex environment are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of camera scheduling, and particularly to a camera scheduling method and system adaptable to environmental changes. Background Art

[0002] In the field of intelligent video surveillance, existing camera scheduling methods mostly perform target handover based on static coverage models and fixed thresholds, and it is difficult to adapt to the impact of complex environmental changes on surveillance effectiveness. In traditional solutions, the coverage range of cameras is usually preset using the nominal surveillance radius, without considering the attenuation effect of dynamic environmental factors such as haze, rain, snow, and sudden changes in day and night illumination on the actual effective surveillance distance, resulting in misjudgment of handover timing and tracking loss; target trajectory prediction mostly relies on a single motion model, which is prone to prediction deviation under environmental interference, and lacks a compensation mechanism based on real-time environmental perception, making it difficult to ensure the smoothness of cross-camera handover; in terms of resource scheduling, candidate camera screening often uses simple distance threshold judgment, without dynamically evaluating multi-dimensional factors such as motion direction and device priority, which is prone to resource competition and handover failure. In addition, existing methods lack a systematic design for camera optical parameter synchronization, computing resource pre-allocation, and exception recovery mechanisms during the handover process, resulting in prominent problems such as parameter jumps and initialization delays. How to achieve environment-adaptive dynamic coverage modeling, multi-model collaborative trajectory prediction, intelligent candidate set screening, and resource pre-emption has become a key challenge for improving the robustness and handover efficiency of video surveillance systems in complex scenarios.

[0003] Therefore, it is necessary to provide a camera scheduling method and system adaptable to environmental changes to solve the above technical problems. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a camera scheduling method and system adaptable to environmental changes, achieving the beneficial effects of continuity, low latency, and high reliability in multi-camera collaborative scheduling and tracking of targets in complex environments.

[0005] The present invention provides a camera scheduling method adaptable to environmental changes, including:

[0006] S1: Construct a coverage relationship matrix based on the spatial coverage parameters of each camera. Meanwhile, obtain environmental perception data and the motion state data of the monitored target;

[0007] S2: Construct a basic state vector based on the motion state data, input it into a basic prediction model to obtain basic prediction data, construct an extended state vector based on the motion state data and environmental perception data, input it into a compensation prediction model to obtain compensation prediction data, and dynamically weight and fuse the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitored target;

[0008] S3: Based on the trajectory prediction data and the coverage relationship matrix, filter out a candidate camera set including at least one candidate camera according to the preset relay screening conditions;

[0009] S4: Calculate the real-time distance and the predicted minimum distance between the central coordinates of each candidate camera and the monitoring target in real time, and determine the relay candidate camera in combination with the preset preemption relay conditions, and at the same time trigger the preemption resource allocation;

[0010] S5: When the real-time distance between the monitoring target and the central coordinates of the relay candidate camera reaches the preset relay execution condition, the relay candidate camera executes the relay task.

[0011] Preferably, in step S1, the construction of the coverage relationship matrix includes the following sub-steps:

[0012] S101: Obtain the monitoring coverage range, central coordinates and reference monitoring radius of each camera;

[0013] S102: At the same time, collect the environmental perception data and device health status data within the monitoring range of each camera in real time;

[0014] S103: Dynamically calculate the effective monitoring radius of each camera based on the environmental perception data and the reference monitoring radius;

[0015] S104: Determine the adjacent relationship based on the overlap degree of the monitoring coverage ranges of the cameras, generate the initial priority based on the historical relay success rate, and correct the initial priority according to the device health status and the overlap degree of the monitoring coverage ranges of the cameras to obtain the adjacent camera priority;

[0016] S105: Output a coverage relationship matrix including the central coordinates, effective monitoring radius, adjacent relationship and adjacent camera priority of all cameras.

[0017] Preferably, in step S2, the basic prediction model adopts the conventional Kalman filter algorithm, and adjusts the process noise covariance matrix and the observation noise covariance matrix of the filter of the Kalman filter algorithm based on the environmental perception data.

[0018] Preferably, in step S2, the compensation prediction model calculates the visibility compensation coefficient and the light mutation compensation coefficient respectively through the visibility and the light change amount in the environmental perception data, and uses the visibility compensation coefficient and the light mutation compensation coefficient as the correction terms of the state prediction equation of the Kalman filter algorithm to generate the compensation prediction data.

[0019] Preferably, in step S2, the step of dynamically weighted fusion of the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target includes:

[0020] Calculate the second weighted weight of the basic prediction data by combining the preset first weighted weight based on the basic prediction data with the visibility and the amount of light change in the environmental perception data;

[0021] Obtain the historical average light intensity, and calculate the light mutation intensity by combining the amount of light change;

[0022] Calculate the third weighted weight of the basic prediction data based on the second weighted weight, the light mutation intensity and the visibility, and limit the third weighted weight within the preset weight range through the Sigmoid function;

[0023] Dynamically weighted-fuse the basic prediction data and the compensation prediction data based on the third weighted weight of the basic prediction data to obtain the trajectory prediction data of the monitoring target.

[0024] Preferably, in step S3, screening out a candidate camera set including at least one candidate camera according to the preset relay screening conditions includes the following steps:

[0025] Based on the trajectory prediction data of the monitoring target and the coverage relationship matrix, screen out candidate cameras whose coverage areas are within the preset angular range threshold of the trajectory prediction direction of the monitoring target, and combine them into an initial candidate camera set;

[0026] Based on the adjacent camera priority of the coverage relationship matrix, screen out at least one candidate camera whose adjacent camera priority is greater than the preset priority threshold from the initial candidate camera set and combine them into a candidate camera set.

[0027] Preferably, step S4 includes the following steps:

[0028] Based on the coverage relationship matrix and the motion state data of the monitoring target, calculate the real-time distance and the predicted minimum distance between each candidate camera and the monitoring target, where the predicted minimum distance is the distance between the candidate camera and the monitoring target calculated based on the predicted trajectory of the monitoring target within a preset future period of time;

[0029] Determine that the candidate camera with a real-time distance less than the preset actual distance threshold and a predicted minimum distance less than the preset predicted distance threshold is a relay candidate camera, and allocate computing resources to the relay candidate camera.

[0030] Preferably, in step S4, the pre-emptive resource allocation further includes real-time synchronization of the target tracking parameters of the current camera associated with the monitoring target to the relay candidate camera.

[0031] Preferably, step S5 further includes a relay verification step. Within a continuous preset frame period after the relay candidate camera executes the relay task, calculate the root mean square error between the actual observed position and the predicted position of the monitoring target. If the root mean square error is lower than the preset prediction deviation threshold, it is determined that the relay is successful.

[0032] The present invention provides a camera scheduling system that adapts to environmental changes, which is applied to a camera scheduling method that adapts to environmental changes, including:

[0033] A data acquisition and preprocessing module, which is used to construct a coverage relationship matrix based on the spatial coverage parameters of each camera. At the same time, obtain environmental perception data and the motion state data of the monitoring target;

[0034] A monitoring target trajectory prediction module, which is used to construct a basic state vector based on the motion state data, input it into the basic prediction model to obtain basic prediction data, construct an extended state vector based on the motion state data and environmental perception data, input it into the compensation prediction model to obtain compensation prediction data, and dynamically weight and fuse the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target;

[0035] A candidate camera set generation module, which is used to screen out a candidate camera set including at least one candidate camera based on the trajectory prediction data and the coverage relationship matrix according to the preset relay screening conditions;

[0036] A relay preemption module, which is used to calculate the real-time distance and the predicted minimum distance between the center coordinates of each candidate camera and the monitoring target in real time, and determine the relay candidate camera in combination with the preset preemption relay conditions, and at the same time trigger preemption resource allocation;

[0037] A relay execution module, which is used to execute the relay task when the real-time distance between the monitoring target and the center coordinates of the relay candidate camera reaches the preset relay execution condition.

[0038] Compared with the related technology, the camera scheduling method and system provided by the present invention have the following beneficial effects:

[0039] The present invention effectively suppresses the trajectory prediction deviation caused by environmental interference by integrating a basic Kalman filter model with an environmental compensation model, modifying the state equation through a visibility compensation coefficient and a light mutation compensation coefficient, and combining a multi-stage weighted weight dynamic adjustment mechanism, providing a basis for the subsequent screening of relay cameras. Then, a screening strategy combining direction matching and adjacent camera priority is adopted to ensure the screening efficiency while increasing the screening accuracy. Next, through pre-emptive resource allocation, the target tracking parameters are synchronized to the relay candidate cameras to eliminate parameter jumps and initialization delays during relay switching. In addition, a relay effectiveness verification and exception rollback mechanism is introduced to achieve rapid recovery in case of equipment failure or environmental anomalies, improving the system robustness, and overall achieving the beneficial effects of continuity, low latency, and high reliability of multi-camera collaborative tracking in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flowchart of a method for camera scheduling that adapts to environmental changes according to the present invention;

[0041] Figure 2 is a module structure diagram of a system for camera scheduling that adapts to environmental changes according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The present invention will be further described in detail below with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only for explaining the present invention, rather than limiting the present invention. In addition, it should be noted that only parts related to the present invention rather than all structures are shown in the drawings for the convenience of description. In addition, the embodiments in the present invention and the features in the embodiments can be combined with each other without conflict.

[0043] In addition, it should be noted that only parts related to the present invention rather than all content are shown in the drawings for the convenience of description. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as being processed sequentially, many of the operations can be performed in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operations are completed, but there can also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0044] Embodiment 1

[0045] A method for camera scheduling that adapts to environmental changes. In the specific implementation process, as Figure 1 shown, it shows a flowchart of a method for camera scheduling that adapts to environmental changes, including:

[0046] Step S1: Construct a coverage relationship matrix based on the spatial coverage parameters of each camera. Meanwhile, obtain environmental perception data and the motion state data of the monitoring target.

[0047] Specifically, in step S1, the construction of the coverage relationship matrix includes the following sub-steps:

[0048] S101: Obtain the monitoring coverage range, central coordinates, and reference monitoring radius of each camera;

[0049] S102: Meanwhile, collect in real time the environmental perception data and the device health status data within the monitoring range of each camera;

[0050] S103: Dynamically calculate the effective monitoring radius of each camera based on the environmental perception data and the reference monitoring radius;

[0051] S104: Determine the adjacent relationship based on the overlap degree of the monitoring coverage ranges of the cameras, generate an initial priority based on the historical relay success rate, and correct the initial priority according to the device health status and the overlap degree of the monitoring coverage ranges of the cameras to obtain the adjacent camera priorities;

[0052] S105: Output a coverage relationship matrix including the central coordinates, effective monitoring radius, adjacent relationship, and adjacent camera priorities of all cameras.

[0053] During the specific implementation process, first, the static spatial parameters of each camera, including the monitoring coverage range and center coordinates, are obtained through including but not limited to the positioning module or preset configuration of each camera. The monitoring coverage range is pre-calibrated based on the hardware parameters of the camera's lens focal length and the pan-tilt rotation range. At the same time, for example, the built-in sensors of the camera are used to collect environmental perception data in real time, including but not limited to visibility measured by optical sensors and light changes detected by photosensors. The device health status data is obtained through temperature sensors and system monitoring interfaces. The device health status data includes but is not limited to the real-time temperature of the device, CPU load rate and network communication delay. Subsequently, the effective monitoring radius of each camera is dynamically adjusted according to the environmental perception data. For example, when the visibility is lower than the preset visibility threshold, the monitoring radius is reduced to match the actual visual distance. If a drastic change in light intensity is detected, further At the same time, if the real-time temperature of the device or the CPU load rate exceeds the normal operating threshold, the effective monitoring radius is reduced according to the preset rules to reflect the impact of the device performance degradation on the monitoring capability; next, based on the corrected effective monitoring radius and center coordinates, the adjacent relationship between cameras is determined, and the initial priority is generated based on the historical relay success rate. The initial priority is corrected according to the overlap of the device health status and the camera's monitoring coverage to obtain the priority of the adjacent camera. If the real-time temperature of a camera exceeds the preset operating temperature threshold, the priority of its associated adjacent cameras is lowered; the final output coverage relationship matrix contains the center coordinates of all cameras, the dynamically adjusted effective monitoring radius, the adjacent relationship and the priority points of the adjacent cameras, forming a complete spatial topology data network, which provides data support for environmental adaptation and device status perception for subsequent steps.

[0054] Step S2: Construct a basic state vector based on the motion state data, input the basic prediction model to obtain basic prediction data, construct an extended state vector based on the motion state data and environmental perception data, input the compensation prediction model to obtain compensation prediction data, and dynamically weighted fuse the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitored target.

[0055] Specifically, in step S2, the basic prediction model adopts a conventional Kalman filter algorithm, and adjusts the process noise covariance matrix and the observation noise covariance matrix of the filter of the Kalman filter algorithm based on the environmental perception data.

[0056] Specifically, in step S2, the compensation prediction model calculates the visibility compensation coefficient and the illumination mutation compensation coefficient respectively through the visibility and illumination change in the environmental perception data, and uses the visibility compensation coefficient and the illumination mutation compensation coefficient as correction terms of the state prediction equation of the Kalman filter algorithm to generate compensation prediction data.

[0057] Specifically, in step S2, the step of dynamically weighted fusion of the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target includes:

[0058] Based on the preset first weighting weight of the basic prediction data, calculate the second weighting weight of the basic prediction data by combining the visibility and the amount of light change in the environmental perception data;

[0059] Obtain the historical average light intensity, and calculate the light mutation intensity by combining the amount of light change;

[0060] Calculate the third weighting weight of the basic prediction data based on the second weighting weight, the light mutation intensity and the visibility, and limit the third weighting weight within the preset weight range through the Sigmoid function;

[0061] Dynamically weighted fusion of the basic prediction data and the compensation prediction data based on the third weighting weight of the basic prediction data to obtain the trajectory prediction data of the monitoring target.

[0062] In the specific implementation process, first, based on the real-time monitored motion state data of the target, including but not limited to speed, acceleration, and position, a basic state vector is constructed, and the basic prediction model using the conventional Kalman filter algorithm is input to generate basic prediction data. The basic prediction model dynamically adjusts its process noise covariance matrix and observation noise covariance matrix through environmental perception data. Exemplarily, in a low visibility environment, the process noise covariance is increased to reflect the impact of environmental uncertainty on the target motion. At the same time, the observation noise covariance is adjusted according to the amount of light change to match the observation error characteristics of the sensor under complex lighting conditions, thereby enhancing the environmental adaptability of the basic prediction. Subsequently, an extended state vector is constructed based on the motion state data and the visibility and light change amount in the environmental perception data, and the compensated prediction model is input to generate compensated prediction data. The compensated prediction model calculates the visibility compensation coefficient by analyzing the visibility to enhance the prediction robustness in low visibility scenarios. At the same time, a light mutation compensation coefficient is generated based on the difference between the current light intensity and the historical average value to offset the interference of sudden light changes on the monitoring target tracking, and the two types of compensation coefficients are integrated into the state prediction equation of the Kalman filter algorithm as correction terms, thereby generating the compensated trajectory prediction data, that is, the compensated prediction data. Further, the basic prediction data and the compensated prediction data are integrated into the trajectory prediction data of the final monitoring target through a dynamic weighted fusion mechanism. First, a preset first weighting weight is assigned to the basic prediction data, and the second weighting weight is calculated in combination with the real-time visibility and light change amount. Then, the light mutation intensity is calculated based on the historical average light intensity, and the second weighting weight is dynamically adjusted to obtain the third weighting weight to enhance its dominant role in a stable environment or suppress its prediction deviation in an environment mutation. Subsequently, the third weighting weight is range-constrained through the Sigmoid function to avoid the negative impact of extreme weight values on the fusion result. Finally, the two types of prediction data are weighted and fused based on the constrained third weighting weight, and the trajectory prediction data of the monitoring target is output, providing a reliable basis for subsequent relay screening and resource scheduling.

[0063] Step S3: Based on the trajectory prediction data and the coverage relationship matrix, a candidate camera set including at least one candidate camera is screened out according to the preset relay screening conditions.

[0064] Specifically, in step S3, screening out a candidate camera set including at least one candidate camera according to the preset relay screening conditions includes the following steps:

[0065] Based on the trajectory prediction data of the monitoring target and the coverage relationship matrix, candidate cameras whose coverage areas are within the preset direction angle range threshold of the trajectory prediction direction of the monitoring target are screened out and combined into an initial candidate camera set;

[0066] Based on the priority of adjacent cameras in the coverage relationship matrix, at least one candidate camera combination with a priority of adjacent cameras greater than the preset priority threshold is selected from the initial candidate camera set as the candidate camera set.

[0067] In the specific implementation process, first, using the motion direction and speed information of the monitored target in the trajectory prediction data of the monitored target, cameras with monitoring coverage within a preset included angle range of the motion direction of the monitored target are selected as the initial candidate set to ensure a reasonable spatial match between the monitoring coverage of the candidate cameras and the target motion trend; subsequently, combined with the adjacent camera priority data recorded in the coverage relationship matrix, the initial candidate set is screened again, and adjacent cameras with priorities lower than the preset priority threshold are removed, and only candidate cameras with a high historical handover success rate, good device health status, and coverage overlap meeting the requirements are retained to form the final candidate camera set.

[0068] Step S4: Calculate the real-time distance and predicted minimum distance between the central coordinates of each candidate camera and the monitored target in real time, and determine the handover candidate camera in combination with the preset preemption handover condition, and at the same time trigger preemption resource allocation.

[0069] Specifically, step S4 includes the following steps:

[0070] Based on the coverage relationship matrix and the motion state data of the monitored target, calculate the real-time distance and predicted minimum distance between each candidate camera and the monitored target, where the predicted minimum distance is the distance between the candidate camera and the monitored target calculated based on the predicted trajectory of the monitored target within a preset future period of time;

[0071] Determine that the candidate camera with a real-time distance less than the preset actual distance threshold and a predicted minimum distance less than the preset predicted distance threshold is the handover candidate camera, and allocate computing resources to the handover candidate camera.

[0072] Specifically, in step S4, the preemption resource allocation further includes real-time synchronization of the target tracking parameters of the current camera associated with the monitored target to the handover candidate camera.

[0073] During the specific implementation process, based on the camera center coordinates and effective monitoring radius recorded in the coverage relationship matrix, combined with the real-time motion state data of the monitoring target, calculate the real-time distance between each candidate camera and the target; at the same time, based on the trajectory prediction data of the monitoring target, deduce the motion path of the monitoring target within a preset future time period, and calculate the predicted minimum distance between the candidate camera and the monitoring target along this path. Select the candidate cameras whose real-time distance is less than the actual distance threshold and the predicted minimum distance meets the predicted distance threshold as the relay candidate cameras; subsequently, trigger pre-emptive resource allocation to reserve dedicated computing resources for the selected relay candidate cameras. Exemplarily, such as GPU computing power and communication bandwidth, to ensure that the relay candidate cameras have sufficient real-time processing capabilities when executing the relay task. At the same time, the optical parameters of the current camera that are tracking the monitoring target, exemplarily, such as exposure time, zoom ratio, and the algorithm model parameters, exemplarily, such as target feature templates and tracking confidence thresholds, are synchronized to the relay candidate cameras in real time through a low-latency communication protocol, so that the optical parameters and algorithm states of the relay candidate cameras are seamlessly adapted to the current tracking scenario, avoiding target loss or image quality degradation caused by parameter jumps. Through the above operations, ensure that the relay candidate cameras complete resource readiness and parameter synchronization before the target enters its monitoring range, laying the foundation for smooth relay.

[0074] Step S5: When the real-time distance between the monitoring target and the center coordinates of the relay candidate camera reaches the preset relay execution condition, the relay candidate camera executes the relay task.

[0075] Specifically, step S5 further includes a relay verification step. Calculate the root mean square error between the actual observed position and the predicted position of the monitoring target within a continuous preset frame period after the relay candidate camera executes the relay task. If the root mean square error is lower than the preset prediction deviation threshold, it is determined that the relay is successful.

[0076] In the specific implementation process, first, based on the central coordinates of the relay candidate camera and the real-time distance of the monitored target, it is determined whether the monitored target enters the effective monitoring range of the relay candidate camera. If the condition is met, the candidate camera takes over the tracking task. At the same time, through the pre-emption resource allocation mechanism, it is ensured that the relay camera has synchronized the target tracking parameters of the current camera, including, but not limited to, exposure time, zoom ratio, white balance setting, and target feature template, and transmits the target motion trajectory prediction data in real time through a low-latency communication protocol, so that the relay camera pre-loads the tracking algorithm model in advance and warms up to a stable state. Subsequently, within a continuous preset frame period after the relay execution, the root mean square error is calculated by comparing the actual observed position with the predicted trajectory data. If the error value continuously remains lower than the preset deviation threshold, the relay is determined to be successful; otherwise, an exception recovery mechanism is triggered, including, but not limited to, falling back to the original camera and expanding the monitoring radius of the original camera to re-capture the target, and releasing the computing resources allocated to the relay candidate camera at the same time. In addition, if the number of relay verification failures exceeds the preset threshold, a full-system circular scanning process is started, and the scanning range is dynamically adjusted based on the priorities of adjacent cameras in the coverage relationship matrix, and the monitoring coverage areas of high-priority cameras are preferentially searched to ensure the rapid restoration of the continuity of target tracking in complex environments or equipment anomalies.

[0077] The working principle of a camera scheduling method for adapting to environmental changes provided by the present invention is as follows:

[0078] The present invention constructs a coverage relationship matrix through environmental perception data and the spatial coverage parameters of cameras to real-time correct the effective monitoring radius of each camera and the priorities of adjacent cameras, forming an environment-adaptive spatial topology network; based on the motion state data of the monitored target, a basic prediction model composed of the Kalman filtering algorithm is used to predict and generate a basic trajectory prediction, and the prediction equation of the compensation prediction model is corrected by combining the visibility compensation coefficient and the illumination mutation compensation coefficient to generate compensation prediction data; two types of prediction data are integrated through a dynamic weighted fusion mechanism to output high-precision monitored target trajectory prediction data; based on a dual screening strategy of trajectory direction matching and adjacent camera priorities, a set of candidate cameras is selected from the coverage relationship matrix; the central coordinates of the candidate camera and the real-time distance and predicted minimum distance of the monitored target are calculated in real time, and the relay candidate camera is determined based on the preset pre-emption relay conditions, and at the same time, pre-emption resource allocation is triggered. Combining resource pre-allocation and tracking parameter synchronization ensures that the resources of the relay candidate camera are ready; finally, when the monitored target enters the relay range, the relay candidate camera executes the relay task, and the relay effect is verified through the root mean square error of consecutive frames. When an exception occurs, a fallback mechanism and a system-level scan recovery are triggered to achieve a smooth transition and high reliability of multi-camera collaborative tracking in complex environments.

[0079] Embodiment 2

[0080] An adaptive camera scheduling system for environmental changes. During the specific implementation process, such as Figure 2 shown, it shows a module structure diagram of an adaptive camera scheduling system for environmental changes, including:

[0081] A data acquisition and preprocessing module 100, which is used to construct a coverage relationship matrix based on the spatial coverage parameters of each camera. At the same time, it acquires environmental perception data and the motion state data of the monitoring target;

[0082] A monitoring target trajectory prediction module 200, which is used to construct a basic state vector based on the motion state data, input the basic prediction model to obtain basic prediction data, construct an extended state vector based on the motion state data and environmental perception data, input the compensation prediction model to obtain compensation prediction data, and dynamically weight and fuse the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target;

[0083] A candidate camera set generation module 300, which is used to screen out a candidate camera set including at least one candidate camera based on the trajectory prediction data and the coverage relationship matrix according to the preset relay screening conditions;

[0084] A relay preemption module 400, which is used to calculate the real-time distance and the predicted minimum distance between the center coordinates of each candidate camera and the monitoring target in real time, and determine the relay candidate camera in combination with the preset preemption relay conditions, and at the same time trigger the preemption resource allocation;

[0085] A relay execution module 500, which is used to execute the relay task when the real-time distance between the monitoring target and the center coordinates of the relay candidate camera reaches the preset relay execution condition.

[0086] The working principle of an adaptive camera scheduling system for environmental changes provided by the present invention is as follows:

[0087] The present invention obtains the spatial coverage parameters, environmental perception data, and motion state data of monitoring targets of each camera in real time through the data acquisition and preprocessing module 100, and dynamically constructs a coverage relationship matrix to reflect the effective monitoring radius, adjacent relationship, and priority of the cameras; the monitoring target trajectory prediction module 200 constructs a basic state vector based on the motion state data, generates a basic trajectory prediction through a basic prediction model, and at the same time constructs an extended state vector in combination with the environmental perception data and inputs it into a compensation prediction model to generate compensation prediction data corrected by visibility and light mutation compensation, and integrates the two types of prediction results through a dynamic weighted fusion mechanism to output a high-precision trajectory prediction; the candidate camera set generation module 300 screens a set of candidate cameras whose coverage areas match the target motion direction and whose priorities meet the standards according to the trajectory prediction direction and the coverage relationship matrix; the relay preemption module 400 calculates the real-time distance and predicted minimum distance between the candidate machine and the target in real time, determines the relay candidate machine in combination with the preemption conditions, and triggers resource pre-allocation, and synchronizes the optical parameters and algorithm states of the current camera to the candidate machine to eliminate parameter jumps; finally, the relay execution module 500 takes over the tracking task when the target enters the relay range, verifies the deviation between the actual trajectory and the predicted value through consecutive frames, and if the error exceeds the limit, triggers an abnormal recovery mechanism to dynamically expand the monitoring radius of the original camera or start circular scanning to ensure the continuity and robustness of multi-camera relay tracking in a complex environment.

[0088] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing in the process Figure 1 one process or multiple processes and / or blocks Figure 1 a device for the functions specified in one block or multiple blocks.

[0089] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium, which includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0090] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent in such process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

Claims

1. An adaptive camera scheduling method for environmental changes, characterized in that The camera scheduling method includes the following steps: S1: Construct a coverage relationship matrix based on the spatial coverage parameters of each camera. Meanwhile, obtain environmental perception data and the motion state data of the monitoring target; S2: Construct a basic state vector based on the motion state data, input it into a basic prediction model to obtain basic prediction data. Construct an extended state vector based on the motion state data and environmental perception data, input it into a compensation prediction model to obtain compensation prediction data, and dynamically weight and fuse the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target; S3: Based on the trajectory prediction data and the coverage relationship matrix, screen out a candidate camera set including at least one candidate camera according to the preset relay screening conditions; S4: Calculate in real time the real-time distance and the predicted minimum distance between the center coordinates of each candidate camera and the monitoring target, and determine the relay candidate camera in combination with the preset preemption relay conditions, and at the same time trigger preemption resource allocation; S5: When the real-time distance between the monitoring target and the center coordinates of the relay candidate camera reaches the preset relay execution condition, the relay candidate camera executes the relay task.

2. The camera scheduling method for adapting to environmental changes according to claim 1, characterized in that, In step S1, the construction of the coverage relationship matrix includes the following sub-steps: S101: Obtain the monitoring coverage range, center coordinates and reference monitoring radius of each camera; S102: Meanwhile, collect in real time the environmental perception data and device health status data within the monitoring range of each camera; S103: Dynamically calculate the effective monitoring radius of each camera based on the environmental perception data and the reference monitoring radius; S104: Determine the adjacent relationship based on the overlap degree of the monitoring coverage ranges of the cameras, generate an initial priority based on the historical relay success rate, and correct the initial priority according to the device health status and the overlap degree of the monitoring coverage ranges of the cameras to obtain the adjacent camera priority; S105: Output a coverage relationship matrix including the center coordinates, effective monitoring radius, adjacent relationship and adjacent camera priority of all cameras.

3. The camera scheduling method adaptable to environmental changes according to claim 2, wherein In step S2, the basic prediction model uses the conventional Kalman filter algorithm, and adjusts the process noise covariance matrix and the observation noise covariance matrix of the filter of the Kalman filter algorithm based on the environmental perception data.

4. An adaptive environment change camera scheduling method according to claim 3, characterized in that, In step S2, the compensation prediction model calculates the visibility compensation coefficient and the light mutation compensation coefficient respectively through the visibility and the light change amount in the environmental perception data, and uses the visibility compensation coefficient and the light mutation compensation coefficient as the correction terms of the state prediction equation of the Kalman filter algorithm to generate compensation prediction data.

5. An adaptive environment change camera scheduling method according to claim 4, characterized in that, In step S2, the steps of dynamically weighting and fusing the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target include: Based on the preset first weighting weight of the basic prediction data, calculate the second weighting weight of the basic prediction data in combination with the visibility and the light change amount in the environmental perception data; Obtain the historical average light intensity, and calculate the light mutation intensity in combination with the light change amount; Calculate the third weighting weight of the basic prediction data based on the second weighting weight, the light mutation intensity and the visibility, and limit the third weighting weight within the preset weight range through the Sigmoid function; The third weighted weight based on the basic prediction data dynamically weights and fuses the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target.

6. The camera scheduling method adaptable to environmental changes according to claim 5, characterized in that, In step S3, screening out a candidate camera set including at least one candidate camera according to the preset relay screening conditions includes the following steps: Based on the trajectory prediction data of the monitoring target and the coverage relationship matrix, screening out candidate cameras whose coverage areas are within the threshold range of the preset direction angle between the trajectory prediction direction of the monitoring target, and combining them into an initial candidate camera set; Based on the priority of adjacent cameras in the coverage relationship matrix, screening out at least one candidate camera with an adjacent camera priority greater than the preset priority threshold from the initial candidate camera set and combining them into a candidate camera set.

7. An adaptive environment change camera scheduling method according to claim 6, characterized in that, Step S4 includes the following steps: Based on the coverage relationship matrix and the motion state data of the monitoring target, calculating the real-time distance and the predicted minimum distance between each candidate camera and the monitoring target, where the predicted minimum distance is the distance between the candidate camera and the monitoring target calculated based on the predicted trajectory of the monitoring target within a preset future period of time; Determining the candidate cameras with a real-time distance less than the preset actual distance threshold and a predicted minimum distance less than the preset predicted distance threshold as relay candidate cameras, and allocating computing resources to the relay candidate cameras.

8. An adaptive environment change camera scheduling method according to claim 7, characterized in that, In step S4, the pre-emptive resource allocation further includes synchronizing the target tracking parameters of the current camera associated with the monitoring target to the relay candidate cameras in real time.

9. The camera scheduling method for adapting to environmental changes according to claim 8, characterized in that, Step S5 further includes a relay verification step. Calculate the root mean square error between the actual observed position of the monitoring target and the predicted position of the monitoring target within a continuous preset frame period after the relay candidate camera executes the relay task. If the root mean square error is lower than the preset prediction deviation threshold, it is determined that the relay is successful.

10. An adaptive camera scheduling system for environmental changes, characterized in that, Applied to a camera scheduling method for adapting to environmental changes as described in any one of claims 1-9, the camera scheduling system includes: A data acquisition and preprocessing module, configured to construct a coverage relationship matrix based on the spatial coverage parameters of each camera, and at the same time, obtain environmental perception data and the motion state data of the monitoring target; A monitoring target trajectory prediction module, configured to construct a basic state vector based on the motion state data, input the basic prediction model to obtain basic prediction data, construct an extended state vector based on the motion state data and the environmental perception data, input the compensation prediction model to obtain compensation prediction data, and dynamically weight and fuse the basic prediction data and the compensation prediction data to obtain the trajectory prediction data of the monitoring target; A candidate camera set generation module, configured to screen out a candidate camera set including at least one candidate camera according to the preset relay screening conditions based on the trajectory prediction data and the coverage relationship matrix; A relay pre-emption module, configured to calculate the real-time distance and the predicted minimum distance between the center coordinates of each candidate camera and the monitoring target in real time, and determine the relay candidate cameras in combination with the preset pre-emption relay conditions, and at the same time trigger the pre-emptive resource allocation; A relay execution module, configured to when the real-time distance between the monitoring target and the center coordinates of the relay candidate camera reaches the preset relay execution condition, the relay candidate camera executes the relay task.

Images