Dynamic area matching and strategy automatic association method based on camera PTZ value

Abstract

The present invention provides a method for automatically associating dynamic area matching and strategies based on camera PTZ values, comprising the steps of: obtaining the camera's installation posture; calculating a three-dimensional coordinate set of the camera's illumination area based on the installation posture and the camera's PTZ value; matching the three-dimensional coordinate set of the illumination area with the camera's illumination scene three-dimensional model, and associating the camera's recognition strategy based on the matching result; and when the camera's PTZ value changes, using the PTZ change value to calculate and update the matching result between the three-dimensional coordinate set of the illumination area and the camera's illumination scene three-dimensional model, and dynamically invoking the corresponding recognition strategy. By automatically updating the camera's illumination area information based on the camera's PTZ value and dynamically associating the recognition strategy based on the scene three-dimensional modeling matching result, the present invention achieves automated scene adaptation, eliminating the need for manual configuration or later adjustment of the camera surveillance recognition strategy, thereby improving the system's intelligence level and resource utilization.

Description

Technical Field

[0001] The present invention relates to the field of intelligent security technology, and in particular to a method for automatically associating dynamic area matching and strategies based on camera PTZ values. Background Art

[0002] In the technical implementation of intelligent security monitoring systems, dynamic modeling of the camera's coverage area and precise configuration of recognition strategies are key elements in ensuring effective monitoring. Current mainstream systems generally rely on manual intervention to accomplish these two critical tasks, but this technical implementation suffers from significant efficiency bottlenecks and reliability risks, specifically inherent flaws in hardware parameter calibration, dynamic view response strategies, and spatial mapping.

[0003] The camera's illumination area is essentially the geometric mapping of an optical projection model within a three-dimensional scene. Its boundaries are determined by the camera's physical installation position, the three-dimensional coordinates of the optical axis, the lens' internal parameters, the focal length distortion coefficient, and the PTZ dynamic adjustment parameters: horizontal rotation angle, vertical pitch angle, and zoom factor. In traditional technical architectures, modeling the initial illumination area relies on manual offline calibration: technicians place calibration objects such as a checkerboard within the surveillance scene, measure the camera's installation height, horizontal offset distance, and optical axis pitch angle using a laser rangefinder, and, based on the focal length and field of view parameter table provided by the lens manufacturer, manually draw the illumination area boundaries on the two-dimensional image plane within the surveillance management software, such as the vertex pixel coordinates of a rectangular warning zone. This geometric modeling method based on empirical formulas has two sources of error: first, the limited manual measurement accuracy of physical parameters. The height measurement error is usually above ±5cm, and the horizontal tilt error exceeds ±2°, resulting in a spatial deviation of 10% to 15% between the theoretically calculated illumination area boundary and the actual optical projection range; second, the influence of lens nonlinear distortion. In actual imaging, the radial distortion coefficient of the wide-angle lens is not effectively compensated, and the geometric distortion rate of the edge field of view can reach 5% to 8%, causing the manually configured regular-shaped illumination area, such as a rectangle, to have obvious geometric deformation at the edge of the image, directly affecting the positioning accuracy of subsequent target detection.

[0004] When a camera adjusts its viewing angle using the PTZ function, such as by rotating horizontally 30° to zoom in to 200mm, the spatial extent of the illuminated area and its projection on the image plane undergo complex transformations. Existing systems lack automated geometric transformation calculations, preventing real-time updates of the illuminated area based on PTZ parameters. Technicians must bring calibration equipment to the site, remeasure the installation parameters at the new viewing angle, and repeat the offline calibration process. A single configuration typically takes over 15 minutes. For large-scale security projects deploying over 100 PTZ cameras, such as airport campus surveillance systems, the number of viewing angle adjustments required annually due to changes in equipment inspection scenarios can reach 3,000 to 5,000. The resulting manual configuration costs account for 25% to 35% of total O&M expenses. Furthermore, the lag in manual configuration leaves the surveillance system in a strategic vacuum for several hours after the viewing angle adjustment. During this period, if an intrusion occurs, the inaccurate boundary of the illuminated area can easily lead to missed or false alarms. This delay can pose a serious security risk, especially in critical, real-time scenarios such as perimeter security.

[0005] The configuration of recognition strategies relies on accurate modeling of the illumination area, and the two exhibit a complex geometric coupling in their spatial mapping relationship. In traditional systems, security personnel must manually delineate recognition areas on the 2D surveillance screen, such as virtual cordoned areas and no-stop zones. These pixel-based 2D areas must be mapped to the 3D scene space using camera calibration parameters. However, existing technologies lack an effective spatial coordinate conversion mechanism. Firstly, the camera's internal and external parameters, especially the distortion coefficient, are not fully incorporated into the mapping model. This results in a significant increase in the conversion error between 2D pixel coordinates and 3D world coordinates with increasing field of view, with positioning deviations of up to 0.5 to 1 meter in edge areas. Secondly, when PTZ parameters change, the camera's field of view along the optical axis shifts, completely invalidating the original mapping between the 2D recognition area and the 3D scene. Consequently, the system cannot automatically trigger a policy update, requiring manual re-delineation of the areas on the new screen. This process is not only labor-intensive but also prone to inconsistent policy configurations due to subjective operator judgment. For example, in a cross-camera linkage scenario, if the spatial overlap of recognition areas under different viewing angles is less than 70%, target tracking will be disrupted, creating blind spots.

[0006] A deeper logical analysis of the technical implementation reveals that the core flaw of traditional solutions lies in their lack of dynamic geometric modeling capabilities and automated mapping mechanisms for policy spaces. Existing systems rely on static data from offline calibration, making it impossible to calculate the three-dimensional boundaries of the illuminated area in real time after changes in PTZ parameters. Furthermore, at the software architecture level, the recognition policy configuration module and the camera parameter module are independent of each other, and a spatial coordinate conversion engine based on multi-view geometry is not established. Consequently, the mapping relationship between the two-dimensional image area and the three-dimensional monitoring space relies on manual calibration. This technical architectural limitation makes it difficult for the system to meet the core real-time accuracy requirements of intelligent security when faced with complex scene changes caused by dynamic perspective adjustments.

[0007] In summary, the manual configuration of illumination zones and recognition strategies in traditional security monitoring systems is essentially a temporary technical solution, limited by insufficient geometric modeling automation and a lack of multi-source data fusion capabilities. The resulting inefficiencies and risk of missed reports have become key bottlenecks hindering the intelligent upgrade of security systems. Therefore, it is necessary to develop a dynamic illumination zone modeling algorithm based on real-time calculation of internal and external camera parameters. Combined with automatic mapping technology between 3D scene coordinates and 2D image coordinates, this algorithm can achieve intelligent and adaptive configuration of recognition strategies, fundamentally improving the reliability and operational efficiency of monitoring systems. Summary of the Invention

[0008] In view of the problems existing in the prior art, the present invention provides a method which is simple to implement, low in cost and can automatically achieve camera illumination area matching.

[0009] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0010] The present invention provides a method for automatically associating dynamic area matching and strategy based on camera PTZ values, which mainly includes the following steps:

[0011] Get the installation position of the camera;

[0012] Calculating a three-dimensional coordinate set of the camera's illumination area based on the installation posture and the camera's PTZ value;

[0013] Matching the three-dimensional coordinate set of the illumination area with the three-dimensional model of the illumination scene of the camera, and associating the recognition strategy of the camera according to the matching result;

[0014] When the PTZ value of the camera changes, the PTZ change value is used to calculate and update the matching result between the three-dimensional coordinate set of the illumination area and the three-dimensional model of the illumination scene of the camera, and the corresponding recognition strategy is dynamically called.

[0015] Optionally, the method for modeling the three-dimensional model of the scene illuminated by the camera includes the following steps:

[0016] After the camera is installed, an initial set of images of the illuminated scene is captured by the camera;

[0017] Acquire a set of fixed-point images of the scene illuminated by the camera through multi-point shooting of a drone;

[0018] Based on the initial image set and the fixed-point image set, a three-dimensional model of the scene illuminated by the camera is obtained through a three-dimensional reconstruction method.

[0019] Optionally, the drone patrols along the edge of the scene illuminated by the camera in a fixed-altitude mode, and sequentially captures scene images at multiple fixed points on the edge of the scene;

[0020] The scene graphs form the fixed-point image set in a shooting order, and at least one scene graph contains a corresponding camera image.

[0021] Optionally, the three-dimensional reconstruction method includes the following steps:

[0022] Performing image preprocessing on the initial image set and the fixed-point image set, and merging them to obtain a scene multi-view image set;

[0023] Establishing a global coordinate system based on the installation position of the camera;

[0024] Extracting image features from the multi-view image set using the SIFT algorithm, and sequentially matching adjacent images based on the image features to obtain matching pairs;

[0025] Based on the global coordinate system and the matching pairs, batch-generating three-dimensional space points through a structure-from-motion technology framework;

[0026] A three-dimensional model of the scene illuminated by the camera is established based on the three-dimensional space points.

[0027] Optionally, the image preprocessing includes geometric distortion correction and illumination equalization processing.

[0028] Optionally, calculating the three-dimensional coordinate set of the illumination area of the camera includes the following steps:

[0029] Calculating the optical axis direction vector of the camera;

[0030] Calculating an intersection point between the optical axis and the ground based on the optical axis direction vector;

[0031] Calculating the boundary coordinates of the illumination area of the camera according to the intersection of the optical axis and the ground by the inner lens of the camera;

[0032] The illumination area of the camera is marked by the boundary coordinates, and a three-dimensional coordinate set is obtained.

[0033] Optionally, a spatial position identifier is assigned to the three-dimensional model of the illumination scene, and a security rule library based on the spatial position identifier is established;

[0034] An image recognition algorithm library is established, and association rules between the security rule library and the image recognition algorithm library are constructed based on a mapping method.

[0035] Optionally, obtaining the recognition strategy of the camera includes the following steps:

[0036] Calculating the intersection of the camera's illumination area and the camera's illumination scene three-dimensional model based on a spatial geometry algorithm; obtaining the matching result based on the intersection;

[0037] Obtaining a spatial position identifier in the three-dimensional coordinate set of the irradiated area based on the matching result, and correspondingly obtaining a matching entry in the security rule library;

[0038] extracting a corresponding image recognition algorithm from the image recognition algorithm library based on the matching entry and the association rule;

[0039] The corresponding image recognition algorithms are combined to form the recognition strategy of the camera.

[0040] Optionally, when the PTZ value of the camera changes, firstly comparing the matching results before and after the PTZ value change;

[0041] When the matching results before and after the PTZ value change are consistent, the recognition strategy is not called again;

[0042] When the matching results before and after the PTZ value changes are inconsistent, the recognition strategy is re-called and the original recognition strategy is replaced.

[0043] Optionally, the following steps are also included:

[0044] When the target moves out of the camera's illumination area, it is determined whether target tracking is required based on preset conditions; if target tracking is required, the drone is triggered to track the target, and the corresponding image recognition strategy is selected based on the drone's captured image to identify the area where the target is located.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The method of the present invention realizes automatic scene adaptation by automatically updating the camera's illumination area information according to the camera's PTZ value, and dynamically associating the recognition strategy based on the scene three-dimensional modeling matching result. It no longer requires manual configuration or later adjustment of the camera monitoring recognition strategy, thereby improving the system's intelligence level and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 is a flow chart of a method in a specific embodiment of the present invention;

[0049] Figure 2 It is a schematic diagram of a scene in a specific embodiment of the present invention.

[0050] In the picture: 1. PTZ camera, 2. Monitoring scene, 3. Water area. DETAILED DESCRIPTION

[0051] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0052] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0053] In the description of the present invention, “plurality” means two or more, unless otherwise clearly defined.

[0054] It is worth noting that the methods used in the present invention are all conventional methods unless otherwise specified; the raw materials and devices used are all conventional commercially available products, and their sources are not specifically limited unless otherwise specified.

[0055] like Figure 1 As shown, this embodiment provides a method for automatically associating dynamic area matching and strategy based on camera PTZ values, which mainly includes the following steps:

[0056] Combine Figure 2As shown, at least one PTZ camera 1 (hereinafter referred to as the camera) is installed in each camera's monitoring scene 2. This camera acquires monitoring data and, during installation, determines its installation position, i.e., its location and orientation. The position is represented by the camera's height H above the ground and its horizontal position; the orientation is the orientation of the camera's base, expressed as a vector. Since the camera is typically installed horizontally, it is typically a polar coordinate value.

[0057] A camera's PTZ value collectively describes the three core parameters of a surveillance camera: pan, tilt, and zoom. These parameters directly affect the camera's monitoring range, viewing angle adjustment, and image clarity. Pan controls the camera's horizontal rotation angle, determining horizontal coverage. Tilt adjusts the vertical angle, affecting the pitch monitoring range. Zoom determines how far the camera can zoom in or out. Specific PTZ values can be obtained through system-level parameter settings.

[0058] By obtaining the above installation posture and PTZ value, the specific spatial position of the camera can be described from a spatial perspective.

[0059] Thus, the three-dimensional coordinate set of the camera's illumination area is calculated according to the installation posture and the PTZ value of the camera; optionally, calculating the three-dimensional coordinate set of the camera's illumination area includes the following steps:

[0060] First, calculate the camera's optical axis direction vector. This can be determined based on the camera's installation posture and the camera's PTZ values (Pan and Tilt).

[0061] Afterwards, the intersection of the optical axis and the ground is calculated based on the optical axis direction vector; given the camera installation height, the projection method is used to solve the intersection (x0, y0) of the extended line of the optical axis and the ground, thereby obtaining the required intersection of the optical axis and the ground.

[0062] According to the intersection of the zoom in the camera's internal parameters and the optical axis and the ground, the boundary coordinates of the camera's illumination area are calculated through a geometric algorithm, such as Figure 2 The black dots in the middle represent some boundary points.

[0063] Finally, the camera's illumination area is marked by boundary coordinates, and a three-dimensional coordinate set is obtained.

[0064] The three-dimensional coordinate set of the illuminated area is matched with the three-dimensional model of the camera's illuminated scene, and the camera's recognition strategy is associated with the matching result. Optionally, in this embodiment, it is also necessary to pre-acquire the three-dimensional data of the monitoring scene 2 to perform the corresponding modeling operation. There are many executable methods for scene three-dimensional reconstruction in the existing technology. Among them, the three-dimensional modeling of simple scenes can be restored in a computer with reference to engineering drawings, such as leveling the site. For complex scenes, especially those that change over time and have a large number of irregularly shaped equipment and objects, manual mapping, automated mapping, etc. can be used. Specifically, in automated surveying, drones equipped with radar or binocular vision systems can complete surveying during inspections. The survey data can then be used to perform computer-generated 3D reconstruction / restoration of the scene. This approach offers advantages in mature technology and high accuracy. However, it requires specialized drones and ancillary equipment, which are expensive to purchase and configure separately. Furthermore, it requires specialized personnel for operation, making it suitable for 3D surveying of large natural scenes, such as canyons, hills, and rivers. While leasing these systems can reduce operational costs, complex and time-varying scenes, such as those during construction, experience frequent changes in both topography and the movement of large amounts of heavy equipment. This leads to frequent re-surveying, significantly increasing rental costs and impacting overall work progress. Furthermore, complex re-surveying logic must be specified to adapt to specific scenarios, including when to perform re-surveying and modeling, and how to select camera recognition strategies.

[0065] In summary, this embodiment further provides a method for modeling a three-dimensional model of a scene illuminated by a camera, which includes the following steps:

[0066] After the camera is installed, an initial image set of the illuminated scene is captured by the camera; that is, after the camera is installed, images of a portion of the scene are captured in an initial state, and the images are aggregated to form an initial image set.

[0067] Afterwards, a fixed-point image set of the scene illuminated by the camera is obtained through multi-point shooting by a conventional drone equipped with a monocular camera; during this period, this embodiment requires the use of a drone with a fixed height function, because most commercially available products have a fixed height function, such as air pressure height, ultrasonic height, laser height, GPS height or combined height, so it can be selected according to the accuracy requirements, and this embodiment does not impose specific restrictions. In the fixed height mode, the drone patrols along the edge of the scene in the scene illuminated by the camera, and periodically shoots scene images at multiple fixed points at the edge of the scene along the route; the scene images are composed of a fixed-point image set in the order of shooting, and at least one scene image contains the corresponding camera image. Among them, the camera image should be understood as the image taken by the drone at at least one point during the fixed-point shooting process, in which the elements of at least one image contain a complete camera. In this embodiment, the drone's fixed-point photography is performed with a 10m difference between the drone's fixed-altitude flight path and the camera's height H, rather than a constant-height inspection. This design allows for a certain degree of parallax. Subsequently, a circular image is taken, centered around the fixed camera, or along a manually defined GPS path, along the edge of the scene, at horizontal intervals of 15°-30°. Vertical overhead images are also supplemented to acquire multi-angle parallax data. When performing horizontal interval photography, the drone's geometric center is connected to the scene's or the camera's mounting point's geometric center, with the angle between the adjacent lines being the angle between them. Furthermore, the drone's fixed-point photography direction is oriented toward the scene's horizontal geometric center. Furthermore, the drone's onboard POS system (GPS + IMU) records each image's external parameters, such as 3D position, pitch, yaw, and roll angles, in real time, as well as internal parameters, such as focal length and distortion coefficient. This information can be acquired in advance through factory calibration or offline checkerboard calibration.

[0068] Based on the initial image set and the fixed-point image set, a 3D model of the scene illuminated by the camera is obtained through a 3D reconstruction method. The 3D reconstruction method specifically includes the following steps:

[0069] The initial and fixed-point image sets undergo image preprocessing, which includes geometric distortion correction and illumination balancing. Specifically, using image processing libraries such as OpenCV, based on known camera and drone internal parameters, all images are geometrically corrected to eliminate lens distortion, such as radial and tangential distortion. Illumination balancing, such as histogram equalization, is performed on overexposed and underexposed images to enhance feature detectability in weakly textured areas.

[0070] The pre-processed images are merged to obtain a multi-view image set of the scene;

[0071] A global coordinate system is established based on the camera's installation location. Internal parameters such as focal length, optical center coordinates, and distortion coefficients are obtained through checkerboard calibration or device parameters, along with external parameters such as 3D position and horizontal and vertical orientation angles, as the origin of the global reference coordinate system. Specifically, the camera's installation location is used as the world coordinate system origin, defining the X, Y, and Z axes. The GPS coordinates taken by the drone are converted to local coordinates with the camera as the origin using a geographic coordinate conversion tool.

[0072] The SIFT algorithm is used to extract image features from the multi-view image set, extracting key points and descriptors from all images. Each image contains approximately 5,000 feature points, and is robust to scale and rotation changes. Within the multi-view image set, the images are arranged sequentially, starting with the camera images and subsequently sorted according to the order of the drone's fixed points to form the multi-view image set. Adjacent images are then matched based on their image features. Using the FLANN fast matching algorithm, a one-to-one feature match is performed between the drone image and the fixed camera image. The RANSAC algorithm is then used to eliminate incorrect matches, retaining over 80% of correct matches to ensure accurate cross-device perspective association.

[0073] Based on the global coordinate system and matching pairs, a structure-from-motion framework is used to batch generate 3D points. Specifically, triangulation is performed directly using known camera and drone extrinsic parameters to generate the first batch of 3D points. The bundle adjustment (BA) algorithm is then used to optimize the drone's pose, adjusting only the drone's rotation and translation parameters to minimize the reprojection error of the 3D points onto the image and improve global reconstruction accuracy. This completes depth map estimation and point cloud generation. A 3D model of the scene illuminated by the camera is established based on the 3D points. Specifically, the TSDF (Truncated Signed Distance Function) fusion technique is used to merge all sparse point clouds into a dense point cloud. Outlier noise points are filtered out by setting a distance threshold to generate a smooth and continuous point cloud model. The dense point cloud is converted into a triangular mesh model using a Poisson reconstruction algorithm or a Ball Pivoting algorithm, thus obtaining a 3D model of the scene illuminated by the camera.

[0074] In order to realize the flexible calling of camera recognition strategy, this embodiment gives a spatial position identifier in the three-dimensional model of the illumination scene, such as Figure 2 As shown in the figure, a feasible approach is to automatically mark regions based on semantic segmentation. For example, if water surface features are extracted from the image, then if there is water area 3, the spatial position range is marked and the label "water area" is added. For example, if there are flammable materials in area A, the spatial position range is marked and the label "flammable area" is added. Afterwards, a safety rule library based on the spatial position identification is established according to the labels in the spatial position identification. For example, "water area" corresponds to no entry for personnel, and "flammable area" corresponds to no smoking and no open flames.

[0075] An image recognition algorithm library is established based on the entries in the security rule library. Different image recognition algorithms can efficiently and quickly identify corresponding image content. For example, a cigarette recognition algorithm trained on a convolutional neural network (CNN), commonly used in retail and security inspections, can be used for "flammable zone" detection. Thus, association rules are constructed between the security rule library and the image recognition algorithm library based on a mapping method, forming a data chain such as "spatial location identifier [flammable zone] - security rule library [no smoking, no open flames] - image recognition algorithm library [cigarette recognition algorithm, open flame detection algorithm]."

[0076] Therefore, the camera identification strategy acquisition includes the following steps:

[0077] Based on the spatial geometry algorithm, the intersection of the camera's illumination area and the three-dimensional model of the camera's illumination scene is calculated; a matching result is obtained based on the intersection. The matching result here is reflected in which spatial location identifiers are included in the illumination area, and the corresponding matching entries in the safety rule library are obtained; for example, when the camera is initially installed, the matching result shows that the illumination area covers the water area and area A, and the spatial location identifiers [water area] and [flammable area] are obtained. The system then searches the safety rule library and obtains the entries [No entry for personnel] and [No smoking, no open flames]. Based on the matching entries and association rules, the corresponding image recognition algorithms [personnel detection algorithm, cigarette recognition algorithm, open flame detection algorithm] are extracted from the image recognition algorithm library;

[0078] The corresponding three groups of image recognition algorithms are combined to form the camera's recognition strategy, and an alarm message is issued when the corresponding phenomenon is detected.

[0079] When the PTZ value of the camera changes, the PTZ change value is used to calculate and update the matching results between the three-dimensional coordinate set of the illuminated area and the three-dimensional model of the camera's illuminated scene, and the corresponding recognition strategy is dynamically called. Optionally, since the PTZ camera 1 can be adjusted according to remote settings, when the PTZ value of the camera changes, the matching results before and after the PTZ value change are first compared;

[0080] When the matching results before and after the PTZ value change are consistent, that is, the intersection of the camera's illumination area and the camera's illumination scene 3D model contains the same spatial location identifier, there is no need to update the camera's recognition strategy. Therefore, the system does not re-call the recognition strategy, which reduces subsequent operation steps and saves system resources.

[0081] When the matching results before and after the PTZ value changes are inconsistent, that is, the intersection of the calculated camera's illumination area and the camera's illumination scene three-dimensional model contains different spatial position identifiers, the security rule library - image recognition algorithm library is searched in sequence according to the matching results to obtain a new algorithm combination, and the recognition strategy called last time is replaced, thereby realizing dynamic calling of the algorithm as the PTZ value changes, so as to achieve matching between the automated algorithm and the camera's illumination area, save system resources, and no longer need to pre-store a large number of algorithms for parallel detection of camera images.

[0082] Optionally, in the specific monitoring process of this embodiment, there are certain blind spots in the scene due to the influence of ground height difference, equipment orientation, etc., and some blind spots can be used by people to evade monitoring. Therefore, considering that the drone used in the above embodiment is not a high-cost professional equipment, the method further includes the following steps:

[0083] When the target moves out of the camera's irradiation area, it is determined based on preset conditions whether target tracking is required; specifically, since the irradiation area of this embodiment includes a three-dimensional space, there are two situations: one is that the target continuously moves out of the irradiation area boundary; the other is that the target disappears in the irradiation area, such as when the target is not detected for N consecutive frames. The latter can be understood as the target entering a monitoring blind spot. Therefore, during the target tracking process, a position comparison method is used to determine that if the target disappears within the irradiation area, target tracking is required, and the drone is triggered to track the target. The drone converts the target disappearance position provided by the monitoring system into natural geographic coordinates, and automatically cruises to the location for multi-angle shooting. The corresponding image recognition strategy is selected based on the drone-photographed image to identify the area where the target is located. That is, after the system receives the drone-photographed image, it calculates the spatial position identification information in the drone-photographed area based on the image content, and refers to the strategy of the fixed camera to obtain the corresponding retrieval recognition strategy for recognition. I will not go into details here.

[0084] Through the above-mentioned drone collaborative monitoring, blind-spot monitoring can be achieved, and the maneuverability of drones can be reasonably utilized to ensure the necessary dispatch volume, reasonably utilize system resources, and reduce scene monitoring costs.

[0085] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A method for automatically associating dynamic area matching and strategy based on camera PTZ values, characterized by: The steps include: Get the installation position of the camera; Calculating a three-dimensional coordinate set of the illumination area of the camera according to the installation posture and the PTZ value of the camera; Matching the three-dimensional coordinate set of the illumination area with the three-dimensional model of the illumination scene of the camera, and associating the recognition strategy of the camera according to the matching result; When the PTZ value of the camera changes, the matching result between the three-dimensional coordinate set of the illumination area and the three-dimensional model of the illumination scene of the camera is calculated and updated using the PTZ change value, and the corresponding recognition strategy is dynamically called; Assigning a spatial position identifier in the three-dimensional model of the irradiation scene and establishing a safety rule library based on the spatial position identifier; Establishing an image recognition algorithm library, and constructing association rules between the security rule library and the image recognition algorithm library based on a mapping method; The camera identification strategy acquisition includes the following steps: Calculating the intersection of the camera's illumination area and the camera's illumination scene three-dimensional model based on a spatial geometry algorithm; obtaining the matching result based on the intersection; Obtaining a spatial position identifier in the three-dimensional coordinate set of the irradiated area based on the matching result, and correspondingly obtaining a matching entry in the security rule library; extracting a corresponding image recognition algorithm from the image recognition algorithm library based on the matching entry and the association rule; The corresponding image recognition algorithms are combined to form the recognition strategy of the camera.

2. The method for automatically associating dynamic area matching and strategy based on camera PTZ values according to claim 1, characterized in that: The method for modeling the three-dimensional model of the camera illumination scene comprises the following steps: After the camera is installed, an initial set of images of the illuminated scene is captured by the camera; Acquire a set of fixed-point images of the scene illuminated by the camera through multi-point shooting of a drone; Based on the initial image set and the fixed-point image set, a three-dimensional model of the scene illuminated by the camera is obtained through a three-dimensional reconstruction method.

3. The method for automatically associating dynamic area matching and strategy based on camera PTZ values according to claim 2, characterized in that: The drone patrols along the edge of the scene illuminated by the camera in a fixed-altitude mode, and sequentially captures scene images at multiple fixed points on the edge of the scene; The scene graphs form the fixed-point image set in a shooting order, and at least one scene graph contains a corresponding camera image.

4. The method for automatically associating dynamic area matching and strategy based on camera PTZ values according to claim 2 or 3, characterized in that: The three-dimensional reconstruction method comprises the following steps: Performing image preprocessing on the initial image set and the fixed-point image set, and merging them to obtain a scene multi-view image set; Establishing a global coordinate system based on the installation position of the camera; Extracting image features from the multi-view image set using the SIFT algorithm, and sequentially matching adjacent images based on the image features to obtain matching pairs; Based on the global coordinate system and the matching pairs, batch-generating three-dimensional space points through a structure-from-motion technology framework; A three-dimensional model of the scene illuminated by the camera is established based on the three-dimensional space points.

5. The method for automatic association of dynamic area matching and strategy based on camera PTZ values according to claim 4, characterized in that: The image preprocessing includes geometric distortion correction and illumination equalization processing.

6. The method for automatic association of dynamic area matching and strategy based on camera PTZ values according to claim 1, characterized in that: Calculating the three-dimensional coordinate set of the camera's illumination area includes the following steps: Calculating the optical axis direction vector of the camera; Calculating an intersection point between the optical axis and the ground based on the optical axis direction vector; Calculating the boundary coordinates of the illumination area of the camera according to the intersection of the optical axis and the ground by the inner lens of the camera; The illumination area of the camera is marked by the boundary coordinates, and a three-dimensional coordinate set is obtained.

7. The method for automatic association of dynamic area matching and strategy based on camera PTZ values according to claim 1, characterized in that: When the PTZ value of the camera changes, first compare the matching results before and after the PTZ value change; When the matching results before and after the PTZ value change are consistent, the recognition strategy is not called again; When the matching results before and after the PTZ value changes are inconsistent, the recognition strategy is re-called and the original recognition strategy is replaced.

8. The method for automatic association of dynamic area matching and strategy based on camera PTZ values according to claim 1, characterized in that: The following steps are also included: When the target moves out of the camera's illumination area, it is determined whether target tracking is required based on preset conditions; if target tracking is required, the drone is triggered to track the target, and the corresponding image recognition strategy is selected based on the drone's captured image to identify the area where the target is located.

Images