Camera layout position and preset position determination method

Abstract

The application provides a camera layout position and preset position determination method, and relates to the technical field of video monitoring. The method comprises the following steps: obtaining a three-dimensional model of a region to be monitored, and dividing the orthographic projection of the three-dimensional model on a horizontal plane into multiple grids; obtaining scene engineering costs of the grids; obtaining each target to be monitored and a monitoring requirement corresponding to each target to be monitored; obtaining camera costs of each preselected camera; respectively calculating monitoring costs of each grid to each target to be monitored; and determining camera layout positions and preset positions according to the scene engineering costs, the camera costs and the monitoring costs, so that the camera layout positions and the preset positions are determined at low cost and high efficiency, the design efficiency of a video monitoring project is effectively improved, manual input is reduced, and the construction period is significantly shortened.

Description

Technical Field

[0001] This invention relates to the field of video surveillance technology, and more specifically, to a method for determining the location and preset position of cameras. Background Technology

[0002] With the continuous growth of electricity demand, the safety requirements of power equipment are increasing, making monitoring systems for power equipment increasingly important. The rationality of the camera placement and preset positions within the monitoring system directly determines its effectiveness in monitoring power grid operations. Traditional methods of manually selecting camera locations and preset positions are labor-intensive and prone to blind spots or redundant placement due to differences in experience, often resulting in unsatisfactory results. Therefore, exploring a low-cost, high-efficiency method for determining camera placement and preset positions has become an urgent technical problem to be solved. Summary of the Invention

[0003] In view of this, the purpose of this application is to overcome the shortcomings of the prior art and provide a method for determining the location and preset position of a camera. The present invention provides the following technical solution:

[0004] This invention provides a method for determining the location and preset position of a camera, the method comprising:

[0005] Step S01: Obtain a 3D model of the area to be monitored, wherein the area to be monitored includes: the target to be monitored. ,in, ;

[0006] Step S02: Divide the orthographic projection of the three-dimensional model onto the horizontal plane into M×N grids;

[0007] Step S03: Determine the grid based on the scene information of the area to be monitored. Scene engineering cost ,in, The x-coordinate of the grid is represented. The vertical coordinate of the grid is represented. Indicates the camera's installation height;

[0008] Step S04: Obtain the target to be monitored. and the target to be monitored Corresponding monitoring requirements;

[0009] Step S05: Obtain the pre-selected camera Corresponding camera cost , in, express One of the different models of pre-selected cameras, , ;

[0010] Step S06: Based on the three-dimensional model and the target to be monitored... The corresponding monitoring requirements are determined in the grid. The pre-selected camera is installed inside. Installation height At that time, the grid For the target to be monitored The cost of surveillance ;

[0011] Step S07, determine in the grid The pre-selected camera is installed inside. And the installation height is At that time, the grid For the target to be monitored The cost of placement is calculated using the following formula:

[0012]

[0013] in, This represents the cost of the deployment. The grid is At installation height Install the pre-selected camera at the location. The set of targets to be monitored selected at that time. Indicates the set of targets to be monitored The target number in Represents the grid For the set of targets to be monitored The sum of the surveillance costs of each of the targets to be monitored, , A pre-defined positive integer representing the pre-selected camera. The maximum number of monitorable targets;

[0014] Step S08, will Determined as the grid The minimum deployment cost;

[0015] Step S09: Select the minimum placement cost Minimum grid As a camera placement grid, the pre-selected cameras are subsequently determined. and its installation height and including A set of targets to be monitored. ;

[0016] Step S10: Remove the selected camera placement grid and its selected... For each of the targets to be monitored, repeat steps S07 to S09 until all of the targets to be monitored are monitored.

[0017] Step S11: Based on the coordinate positions of all the camera grid points in the 3D model, the coordinate positions of each target to be monitored in each target set in the 3D model, and the camera and its installation height, determine the preset position of each target to be monitored by the camera.

[0018] In one embodiment, acquiring the three-dimensional model of the area to be monitored includes:

[0019] Acquire a scene image set of the area to be monitored, the scene image set including: multiple scene images of the area to be monitored;

[0020] Based on the scene image set, point cloud reconstruction is performed to obtain the three-dimensional model.

[0021] In one embodiment, the target to be monitored The corresponding monitoring requirements include: minimum number of pixels in the horizontal direction. and the smallest pixel in the vertical direction The three-dimensional model and the target to be monitored are used as the basis for this process. The corresponding monitoring requirements are determined in the grid. The pre-selected camera is installed inside. Installation height At that time, the grid For the target to be monitored The cost of surveillance ,include:

[0022] According to the target to be monitored and its position in the 3D model and the pre-selected camera The parameters are used to determine the pre-selected camera. The target to be monitored can be captured by camera. The image is used to determine the target to be monitored. The maximum possible horizontal pixels of the image and the maximum possible vertical pixels , , , and According to the pre-selected camera The parameters are obtained;

[0023] Based on the minimum number of pixels in the horizontal direction The minimum pixel in the vertical direction The maximum possible horizontal pixels and the maximum possible vertical pixel Determine the sharpness parameters ;

[0024] According to the clarity parameter Determine the cost index for clarity ;

[0025] Based on the aforementioned 3D model, the target to be monitored is determined. The normal to the pre-selected camera, and the pre-selected camera. The optical axis, based on the angle between the normal and the optical axis. Determine perspective distortion parameters ;

[0026] Determine the perspective distortion cost index based on the aforementioned perspective distortion parameters. ;

[0027] Based on the aforementioned 3D model, the target to be monitored is determined. With the pre-selected camera actual distance ;

[0028] Obtain the pre-selected camera For the target to be monitored Maximum surveillance distance and minimum monitoring distance ;

[0029] Based on the actual distance Maximum monitoring distance and minimum monitoring distance Determine the distance sharpness index ;

[0030] According to the clarity cost index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location For the target to be monitored The cost of surveillance ;

[0031] Alternatively, based on the aforementioned clarity cost index The distance sharpness index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location At the time of the target to be monitored The cost of surveillance .

[0032] In one embodiment, based on the minimum number of pixels in the horizontal direction The minimum pixel in the vertical direction The maximum possible horizontal pixels and the maximum possible vertical pixel Determine the sharpness parameters The formula is:

[0033] 。

[0034] In one embodiment, the step of basing the resolution parameter on... Determine the cost index for clarity The formula is:

[0035]

[0036] in, It is a logarithmic function. and It is a pre-set scaling factor constant. It is a pre-set threshold.

[0037] In one embodiment, the perspective distortion parameters are determined based on the angle between the normal and the optical axis. The formula is: ,in, This represents the perspective distortion parameter. This represents the angle between the normal and the optical axis.

[0038] In one embodiment, the step of determining the perspective distortion cost index based on the perspective distortion parameters is... The formula is:

[0039] .

[0040] In one embodiment, the step of basing the actual distance Maximum monitoring distance and minimum monitoring distance Determine the distance sharpness index The formula is:

[0041] 。

[0042] In one embodiment, the step of adjusting the resolution cost index... and the aforementioned perspective distortion cost index Determine the grid For the target to be monitored The cost of surveillance The formula is:

[0043] ;

[0044] in, It is a set constant.

[0045] In one embodiment, the step of adjusting the resolution cost index... The distance sharpness index and the aforementioned perspective distortion cost index Determine the grid For the target to be monitored The cost of surveillance The formula is:

[0046]

[0047] in, It is a set constant.

[0048] The camera placement and preset position determination method provided in this application embodiment can determine the camera placement and preset position at low cost and high efficiency, effectively improving the design efficiency of video surveillance projects, reducing manual input, and significantly shortening the construction period.

[0049] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This paper illustrates a flowchart of a method for determining camera placement locations and preset positions according to an embodiment of this application.

[0052] Figure 2 This paper illustrates another flowchart of the method for determining camera placement locations and preset positions provided in an embodiment of this application.

[0053] Figure 3This paper illustrates another flowchart of the method for determining camera placement locations and preset positions provided in an embodiment of this application. Detailed Implementation

[0054] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0055] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the template description is for the purpose of describing particular embodiments only and is not intended to limit the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0057] Example 1

[0058] This application provides a method for determining camera placement locations and preset positions. Please refer to [link to relevant documentation]. Figure 1 The method includes steps S01 to S11.

[0059] Step S01: Obtain a 3D model of the area to be monitored, wherein the area to be monitored includes: the target to be monitored. ,in, .

[0060] In this embodiment, the area to be monitored includes, but is not limited to, the substation and its surrounding area. The target to be monitored... This includes, but is not limited to, electrical equipment such as transformers, disconnectors, oil level gauges, insulators, and various instruments. A 3D model of the area to be monitored can be obtained by scanning or photographing the entire area using mobile devices such as drones or vehicles. Data can be acquired through methods such as laser scanning, monocular or binocular imaging, and then reconstructed using technologies such as Simultaneous Localization and Mapping (SLAM) or Multi-View Stereo (MVS). Alternatively, a 3D model of the area to be monitored, such as a substation, from its design and construction phases, such as computer-aided design drawings, can be directly obtained.

[0061] The three-dimensional model can be expressed using point clouds, triangular meshes, voxels, etc. It can also be expressed using CAD methods such as constructive solid set representation and boundary representation, or using methods such as 3D Gaussian Splatting.

[0062] In one implementation, please refer to Figure 2 The process of obtaining a three-dimensional model of the area to be monitored includes steps S011 and S012.

[0063] Step S011: Obtain a scene image set of the area to be monitored, the scene image set including: multiple scene images of the area to be monitored.

[0064] In this embodiment, a camera with known internal parameters is used to capture images of the monitored area from different angles under bright and relatively stable lighting conditions, resulting in multiple scene images that cover the entire monitored area with some overlap between them. The camera's intrinsic parameter matrix can be obtained using methods such as Zhang Zhengyou's calibration method. A monocular or multi-view camera can be used; using a multi-view camera yields better image quality.

[0065] In other embodiments, multiple scene images can be acquired by means of aerial photography using drones, aerial photography using ground mobile devices (handheld devices or various vehicle-mounted devices), or a combination of aerial photography and aerial photography using ground mobile devices.

[0066] Step S012: Based on the scene image set, point cloud reconstruction is performed to obtain the three-dimensional model.

[0067] In this embodiment, based on the scene image set, computer vision methods, such as the structure-of-motion algorithm, are used to reconstruct sparse point clouds to obtain a 3D model of the area to be monitored. The specific processing steps include: (a) feature point detection and matching; (b) epipolar geometry construction; (c) camera pose and scene structure estimation; and (d) bundled adjustment and optimization of camera pose and scene.

[0068] It should be noted that a 3D model based on sparse point clouds can be used to determine the target to be monitored. The location, extent, and orientation in three-dimensional space. Specifically, the target to be monitored can be directly determined through human-computer interaction based on the 3D model. The location, extent, and orientation in three-dimensional space; alternatively, the target to be monitored can be determined first through human-computer interaction within the scene image. Within the area of ​​the scene image, the monitoring target is determined by the correspondence between the scene image and the 3D model. Position, extent, and orientation in three-dimensional space; image detection technology can also be used to automatically detect targets to be monitored in scene images. Then, the target to be monitored is obtained by utilizing the correspondence between the corresponding scene images and the 3D model. Location, extent, and orientation in three-dimensional space. For example, using the target to be monitored. An approximate observation surface is obtained by fitting a plane to the corresponding points on the sparse point cloud, and the target to be monitored is determined based on the pinhole imaging model. The range within the image from which the camera is to be deployed, i.e., the range within the image from which the camera is to be deployed as the viewpoint.

[0069] Furthermore, for 3D models of sparse point clouds, common computer vision methods can be employed, such as voxel-based multi-view stereo (MVS), depth map fusion-based MVS, feature point diffusion-based MVS, or convolutional neural network (CNN)-based MVS, to reconstruct dense point clouds from sparse point clouds. Various interpolation techniques can also be used to achieve this reconstruction. This embodiment employs a multi-view depth-based MVS algorithm to reconstruct dense point clouds from multi-view depth images. Using methods similar to those described above for sparse point clouds, 3D models based on dense point clouds can more accurately determine the target to be monitored. It can determine the position, surface normal vector, and approximate range in three-dimensional space, and more accurately determine the target to be monitored. The range in the image on the camera to be deployed.

[0070] Furthermore, common algorithms such as Poisson surface reconstruction and Radial Basis Function (RBF) can be used to reconstruct the surface of the 3D model using dense point clouds and triangular meshes or voxels, obtaining a realistic 3D model of the area to be monitored. This embodiment employs the classic Marching Cubes algorithm (MC) to obtain an irregular triangular mesh representation of the area to be monitored, and adds texture maps to the surface to obtain a realistic 3D model of the area. Based on the realistic 3D model and the fusion and mapping relationships of surface parameters, the target to be monitored can be determined. The location, surface normal vector, and approximate range in three-dimensional space can also be used to monitor the target. Verification and confirmation are performed, for example, by rendering a 2D image from any viewpoint, including the viewpoint of the camera to be deployed, before target detection, recognition, and localization are conducted. The reconstructed 3D model can further improve the accuracy of monitoring of the target. The accuracy of parameters such as position in three-dimensional space, including the accuracy of the center of the monitored surface, the size of the monitored surface, and the surface normal vector of the monitored surface, as well as the pixel representation of the monitored surface and the specific target area on the image of the camera to be deployed.

[0071] Step S02: Divide the orthographic projection of the three-dimensional model onto the horizontal plane into M×N grids.

[0072] In this embodiment, a two-dimensional plan view can be obtained by projecting the area to be monitored onto a reference horizontal ground. Specifically, in three-dimensional coordinates, only the two-dimensional coordinates of the three-dimensional model on the horizontal plane are retained, thus obtaining the projection of the three-dimensional model onto the horizontal plane and the corresponding two-dimensional plan view. In other embodiments, a two-dimensional plan view can also be obtained by taking a bird's-eye view of the area to be monitored using a drone, or by directly using a two-dimensional plan view output from an engineering design. The two-dimensional plan view should include the entire area to be monitored, and also include the surrounding area where cameras can be placed.

[0073] It is understood that the area to be monitored is an irregular area. To facilitate computer processing, this embodiment uses a rectangular area that can completely cover the irregular area as the area boundary. A planar coordinate system is established using the two mutually perpendicular sides of the rectangular area as the X-axis and Y-axis, respectively. The rectangular area is then divided into M×N uniform grids, and the position of each grid can be determined using coordinates. It means that among them The row coordinates of the grid center. The column coordinates are the center of the grid.

[0074] It's understandable that setting the grid size should follow a core principle: when placing cameras at different locations within the same grid, the monitoring effect and cost should be roughly similar. In actual installation, cameras may be installed at the center of the grid or other locations within the grid, depending on the site conditions. In practical applications, a grid size of 0.5m x 0.5m to 2m x 2m is generally recommended. If the grid is too small, it will significantly increase the complexity of subsequent calculations; if the grid is too large, environmental differences at different locations within the same grid may lead to significant deviations in monitoring effect and cost, affecting the accuracy of the final camera placement.

[0075] Step S03: Determine the grid based on the scene information of the area to be monitored. Scene engineering cost ,in, The x-coordinate of the grid is represented. The vertical coordinate of the grid is represented. This indicates the installation height of the camera.

[0076] In this embodiment, the cost of scene engineering Used to measure the grid Whether a location is suitable for camera placement can be manually set. In this embodiment, a value not less than 0 is used to indicate that the location is within the grid. Inside, the installation height is The cost of scene engineering required to install cameras at the location A higher scene engineering cost indicates a less suitable grid for placement; an extremely large value (e.g., 100,000) signifies that the grid cannot be used for placement. Camera installation height. Different installation heights result in different project costs. Generally, as the camera installation height increases, the pole height also needs to increase, thus increasing the pole cost. For ease of calculation, the camera installation height is quantified within a selectable range. For example, if the maximum pole height is 6 meters, the possible camera installation height range is 2 meters to 6 meters. Using 1 meter as a quantification unit, the following can be taken: =2, 3, 4, 5, 6, a total of 5 values. Specifically... The value range is related to the grid position (x, y), and the quantization interval can be set as needed, such as a value between 0.5 meters and 1.5 meters. It should be noted that scene engineering costs... It has nothing to do with the camera model.

[0077] Furthermore, grid Corresponding scenario engineering costs Alternatively, maintenance costs can be considered, and other value selection methods can be adopted. For example, grid. If there are high-voltage facilities such as transformers in the grid or its vicinity, making it unsuitable for deployment, then the scene engineering cost will be high. For extremely large values ​​far exceeding normal costs, such as 100,000; if you want to use a grid... By mounting cameras on existing poles, the need for pole erection can be eliminated, resulting in lower installation costs. Within the possible height range, the mesh can be... Corresponding scenario engineering costs For smaller values, such as subtracting the cost of the poles; areas near internal walkways where installation and maintenance are easy can also be gridded. Corresponding scenario engineering costs The smaller the value. Grids outside the scene that are unsuitable for placement. You can exclude points by setting all their placement costs to an extremely large value or by using a mesh with a mask.

[0078] For the grid along the highway If a camera is installed, the image may shake due to ground vibrations when large vehicles pass by, or the camera housing may easily accumulate dust. Therefore, it is advisable to appropriately increase the corresponding scene engineering costs. Road vibration causing image jitter is actually a monitoring cost, but since it is only related to the location of the monitoring point (except for cases where the use of a long focal length due to distance exacerbates the image jitter), it is included in the scene engineering cost. Easy to calculate.

[0079] Step S04: Obtain the target to be monitored. and the target to be monitored Corresponding monitoring requirements.

[0080] For substations, the monitored objects such as transformers, disconnectors, and oil level gauges are each defined as a monitoring target. These are respectively designated as targets to be monitored. Target to be monitored Target to be monitored Wait, note that for different targets to be monitored, The values ​​are different for each target to be monitored. The area to be monitored is determined by the optimal viewing direction (angle) and the area range. The optimal viewing direction can be determined by the target in three-dimensional coordinates. The points on the corresponding plane are determined, such as by least-squares fitting of a plane, and the normal direction of the plane is the optimal observation direction; the target to be monitored in three-dimensional space. The area encompassed by the projections of points on the plane along the normal direction onto the plane can be considered the target area. If an object to be monitored, such as a transformer, needs to be monitored from different directions, i.e., different sides, then each side is considered a target to be monitored. That is, it is treated as multiple targets to be monitored; if one target to be monitored Different regions of an object may have different resolution requirements. For example, if both global monitoring and local magnification are needed to observe certain areas, the object may be treated as multiple targets to be monitored, depending on the resolution requirements and the location of the target. Therefore, one object to be monitored may correspond to multiple targets, each with a different optimal observation direction and area.

[0081] If the target to be monitored has been obtained The image of the target to be monitored Given a template image, image detection technology can be used to detect the target to be monitored in the scene image. Then, based on the 3D model of sparse point cloud, the target to be monitored is obtained by utilizing the correspondence between the corresponding scene image and the 3D model. The point cloud and surface are represented in three-dimensional space to obtain their position and the optimal monitoring direction (the normal direction of the surface to be monitored).

[0082] If a realistic 3D model of the area to be monitored is available, such as a 3D Gaussian Splatting (3DGS) model, a Neural Radiance Field (NeRF) model, or a 3D model with textured surfaces, a set of 2D scene images from different viewpoints can be generated. Then, image detection techniques can be used to determine the target to be monitored within this set of generated scene images. Then utilize the detected targets to be monitored The correspondence between the target and the 3D model is used to obtain the target to be monitored. The location and optimal monitoring direction (surface normal direction) in three-dimensional space are determined. Based on a realistic 3D model of the area to be monitored and a human-computer interaction interface, the monitored target can also be easily manually confirmed or adjusted. Position in three-dimensional space.

[0083] Step S05: Obtain the pre-selected camera Corresponding camera cost , in, express One of the different models of pre-selected cameras, , .

[0084] In this embodiment, based on the project's requirements for monitoring equipment, its functions and performance, and cost considerations, the following was selected: Different models of pre-selected cameras were selected, and the cost of each pre-selected camera was determined. Different pre-selected cameras have different functions, performance, and structures, such as whether they have variable focal length, lens focal length or focal length range, resolution, sensitivity, protection level, whether they have a pan / tilt unit, near-infrared illumination, automatic day / night switching, etc., and their structures include bullet type, dome type, hemispherical type, multi-camera combination, etc. Different pre-selected cameras have different costs, such as using the factory price or purchase price as the pre-selected camera. The cost.

[0085] Step S06: Based on the three-dimensional model and the target to be monitored... The corresponding monitoring requirements are determined in the grid. The pre-selected camera is installed inside. Installation height At that time, the grid For the target to be monitored The cost of surveillance .

[0086] In this embodiment, for each target to be monitored, pre-selected cameras are installed within each grid. And the installation height is The cost of surveillance ,in, express One type of pre-selected camera. Surveillance costs. Considering practical use, if the image is highly distorted, has low clarity, or is obstructed, the monitoring cost will be significant. It's big.

[0087] It should be noted that this only applies to targets that may constitute a surveillance relationship. and grid Calculate the cost of surveillance For example, there is a situation where the target to be monitored... With grid When pairing up, due to the target to be monitored Normal direction and the mesh paired with it Pre-selected cameras deployed inside If the direction (the direction in which the two are connected) deviates too much or the distance is too great, the two cannot form a monitoring relationship; therefore, there is no need to calculate the mesh. Treating surveillance targets The cost of surveillance Simply assign a very large value (such as 100000). This can be achieved using a two-dimensional plan view and pre-selected cameras. By calculating the installation height, the monitoring cost can be roughly estimated by assuming the angle between the camera's direction and the normal direction of the monitored surface, and the distance between the monitored surface and the camera lens (i.e., the object distance). For example, it can be determined using the following formula: .in Target to be monitored Center to Grid Installation height is Distance at location; Target to be monitored Normal direction and camera direction (preselected camera) Observe the surveillance target The angle between the optical axis in the opposite direction of the time (that is, the angle between the center of the monitored surface and the center of the camera lens); This is a preset constant that determines the importance of surveillance costs when deploying surveillance points. Of course, this method does not take into account factors such as camera obstruction.

[0088] For installation height Engineering cost Grids assigned extremely large values This means that the grid where points need to be placed needs to be excluded. No need to calculate the corresponding monitoring cost Or simply assign it a specific value, such as a value not less than 0. Because at this time... and It is irrelevant and can be simply remembered as .

[0089] In one embodiment, the target to be monitored The corresponding monitoring requirements include: minimum number of pixels in the horizontal direction. and the smallest pixel in the vertical direction Please see Figure 3 Step S06 includes: steps S061 to S069.

[0090] Step S061, according to the target to be monitored and its position in the 3D model and the pre-selected camera The parameters are used to determine the pre-selected camera. The target to be monitored can be captured by camera. The image is used to determine the target to be monitored. The maximum possible horizontal pixels of the image and the maximum possible vertical pixels , , , and According to the pre-selected camera The parameters are obtained.

[0091] In the grid In the middle, the installation height is Use pre-selected cameras at the location. Treating surveillance targets Pre-select cameras during surveillance. The target to be monitored was captured by the camera. The image needs to be large enough, with the number of pixels in both the horizontal and vertical directions not less than the minimum number of pixels in the horizontal direction. and the smallest pixel in the vertical direction Otherwise, pre-select camera The target to be monitored was captured by the camera. The image is too small and lacks clarity. It should be noted that the pre-selected camera described here... The target to be monitored was captured by the camera. The image is not from a pre-selected camera. Actually installed on the grid In the middle, the installation height is Instead of using images captured at that location, it assumes a pre-selected camera is installed at that location. Based on the 3D model, imaging formula, and pre-selected camera The image range is obtained by calculating the parameters.

[0092] Set the target to be monitored Maximum allowed width of the image and maximum allowable height , and According to the pre-selected camera The parameters are obtained, including the pre-selected camera. The parameters include: pre-selected cameras The focal length range, sensor imaging area size, and sensor resolution. It can be understood that if the target to be monitored... If the image is too large, pre-select the camera. Preset position deviations, etc., can cause the monitored target to... The image extends beyond the screen border, or the pre-selected camera... Unable to capture the complete target under surveillance. Therefore, the target to be monitored Images should not be too large and need to be limited to a certain range. For example, allow 10% of the image width on the left and right sides, and 10% of the image height on the top and bottom sides, or the maximum allowable width. and maximum allowable height Set each camera to a pre-selected camera Capture 80% of the width and height of the image (Note: pre-selected camera). The captured images can always be scaled to fill the entire monitor screen or a window on the monitor screen, so only the acquisition end needs to be considered, not the display end. (If the camera resolution is 1920...) 1080, =1920 0.8 = 1536 =1080 0.8 = 864. The specific value can be set according to project requirements.

[0093] Pre-selected cameras Determine the maximum permissible focal length within the lens range. Using pre-selected cameras Calculate the positional relationship between the camera coordinate system and the 3D model to monitor the target. Horizontal pixels in the image and vertical pixel count Requirements , .

[0094] For the identified target to be monitored Based on stereo analytical geometry, it is easy to determine the position perpendicular to the pre-selected camera. The optical axis direction passes through the target to be monitored The size of the projection on the central plane is... and Indicates. Pre-selected cameras. The larger the focal length of the lens, the more likely the monitored target will be in the acquired image. The larger the image, the better. Given the object distance and the target to be monitored... Perpendicular to the pre-selected camera The actual size of the projection of the plane along the optical axis and and pre-selected cameras The width and height of the target surface (the effective area of ​​the photosensitive surface of the image sensor) can be determined based on a pinhole imaging or thin lens imaging model. This model, based on similar triangles and sensor parameters, allows for pixel-level determination of the target area. and Converting the image size to millimeters, and then calculating the appropriate values ​​for each. The image distance value is obtained because the object distance is much larger than the image distance in video surveillance applications. The image distance is approximated by the focal length, thus yielding the maximum possible focal length in both the horizontal and vertical directions. and Combined with pre-selected cameras The maximum permissible focal length is the minimum of the three actual usable focal lengths. .

[0095] Step S062, based on the minimum number of pixels in the horizontal direction The minimum pixel in the vertical direction The maximum possible horizontal pixels and the maximum possible vertical pixel Determine the sharpness parameters .

[0096] It's understandable that lower resolution leads to higher monitoring costs, while higher resolution leads to lower monitoring costs. This can be achieved by calculating resolution parameters. In order to measure the clarity.

[0097] In one embodiment, based on the minimum number of pixels in the horizontal direction The minimum pixel in the vertical direction The maximum possible horizontal pixels and the maximum possible vertical pixel Determine the sharpness parameters The formula is:

[0098] ,

[0099] min is the smaller of the two.

[0100] Step S063, according to the resolution parameter Determine the cost index for clarity .

[0101] Pre-selected cameras Use the maximum permissible focal length At that time, the target to be monitored is obtained. Pre-selected cameras The maximum possible horizontal pixels in the captured image and the maximum possible vertical pixel The sharpness parameter can then be calculated using the formula above. . like This indicates that the minimum resolution requirement for observation is met. The smaller the value, the clearer the observation.

[0102] In one embodiment, the step of basing the resolution parameter on... Determine the cost index for clarity The formula is:

[0103]

[0104] in, It is a logarithmic function. and It is a pre-set scaling factor constant. It is a pre-set threshold. .

[0105] because The resolution does not meet the monitoring requirements. A larger value should be chosen that significantly increases the cost of surveillance; hour, The smaller the value, the higher the clarity and the lower the cost; a linear or logarithmic increasing function can be used.

[0106] It should be noted that, among them It is a logarithmic function. The sum is a pre-defined scaling factor constant, such as taking... and , It is a pre-set threshold, such as 0.5. Values ​​less than this threshold... The time has little impact on the clarity of the surveillance, so the clarity cost index is set to a fixed value. Other non-decreasing functions can be selected as needed.

[0107] Step S064: Based on the three-dimensional model, determine the target to be monitored. The normal to the pre-selected camera, and the pre-selected camera. The optical axis, based on the angle between the normal and the optical axis. Determine perspective distortion parameters .

[0108] As described in step S01, the target to be monitored can be easily calculated based on the three-dimensional model and stereo analytical geometry. The normal line, and from the target to be monitored Center to pre-selected camera The optical axis of the optical center (the so-called optical axis refers to the hypothetical pre-selected camera) Monitor the target to be monitored The optical axis at that time (the opposite direction of the optical axis is the camera direction), that is, the target to be monitored. To pre-selected cameras The direction of the center of the lens. Therefore, in three-dimensional space, it is easy to calculate the angle between the two. And based on the included angle The cosine function value is determined as the perspective distortion parameter. .

[0109] Another method is to calculate the targets to be monitored separately. normals and pre-selected cameras The optical axis of the pre-selected camera The larger of the projected angle values ​​between the horizontal and vertical directions of the imaging target surface is used to calculate the distortion. Assuming pre-selected cameras The imaging coordinate system is Oxyz, where O is the origin, and x, y, and z represent the horizontal, vertical, and depth directions, respectively. This allows for the calculation of the target being monitored. Normal direction and pre-selected camera The angle between the projections onto the xOz and yOz planes, respectively.

[0110] In one embodiment, the perspective distortion parameters are determined based on the angle between the normal and the optical axis. The formula is: ,in, This represents the perspective distortion parameter. This represents the angle between the normal and the optical axis.

[0111] The cosine of the angle between the normal and the optical axis is determined as the perspective distortion parameter.

[0112] Step S065: Determine the perspective distortion cost index based on the perspective distortion parameters. .

[0113] In one embodiment, the step of determining the perspective distortion cost index based on the perspective distortion parameters is... The formula is: .

[0114] It should be noted that the pre-selected cameras Acquired target to be monitored The image can also be affected by perspective distortion, which can impact the monitoring effect. (The aforementioned angle...) The larger the value, the greater the perspective distortion, and therefore the greater the cost of surveillance. The smaller the value, the lower the perspective distortion cost index. It is a non-increasing function. If perspective distortion is disregarded, then it can be taken as... Let it be a constant, such as taking This embodiment takes ,here ,exist Then it cannot be monitored. A very large value should be selected, such as 100000. Other non-increasing functions can also be chosen, such as , wait.

[0115] Step S066: Based on the three-dimensional model, determine the target to be monitored. With the pre-selected camera actual distance .

[0116] It's understandable that if pre-selected cameras... To the target to be monitored actual distance For cameras that are far away, pre-select cameras. Lenses requiring long focal lengths, i.e., large magnification, will result in blurred images due to the limited optical resolution of the lenses. Therefore, the cost of surveillance is also limited by the actual distance. The impact.

[0117] Step S067: Obtain the pre-selected camera For the target to be monitored Maximum surveillance distance and minimum monitoring distance .

[0118] Maximum surveillance distance and minimum monitoring distance The threshold is determined based on the camera lens.

[0119] Step S068, based on the actual distance Maximum monitoring distance and minimum monitoring distance Determine the distance sharpness index .

[0120] Distance Clarity Index Used to measure pre-selected cameras The target to be monitored was captured by the camera. The clarity of the image.

[0121] In one embodiment, the step of basing the actual distance Maximum monitoring distance and minimum monitoring distance Determine the distance sharpness index The formula is:

[0122] 。

[0123] It should be noted that, It is a pre-selected camera based on 3D model calculation. To the target to be monitored distance, It is the distance sharpness index. The time is set to a constant value. When it is a decreasing function, Time-increasing function, and The threshold is determined based on the camera lens.

[0124] exist The time is set to the optimal value for best clarity. ;when At that time, that is, the target to be monitored With pre-selected cameras The distance is too close, the target to be monitored If the image size is too large, it may partially exceed the sensor target area, affecting the monitoring effect and thus increasing the monitoring cost. Therefore, at this time... Distance The decreasing function, take ;when At that time, the target to be monitored With pre-selected cameras If the distance is too far, the lens's optical resolution and atmospheric turbulence will affect image clarity, and wind or road vibrations will affect the pre-selected camera. Shaking is amplified and affects image sharpness, therefore... It is an increasing function. .

[0125] Step S069, according to the resolution cost index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location For the target to be monitored The cost of surveillance Alternatively, based on the aforementioned clarity cost index. The distance sharpness index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location At the time of the target to be monitored The cost of surveillance .

[0126] This application provides two different monitoring costs in its embodiments. For details on the calculation method, please refer to the following text.

[0127] In one embodiment, the step of adjusting the resolution cost index... and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location At that time, for the target to be monitored The cost of surveillance The formula is:

[0128] ;

[0129] in, It is a set constant.

[0130] Surveillance Costs With pre-selected cameras The target to be monitored was captured by the camera. The sharpness and perspective distortion are related; specifically, the higher the sharpness, the higher the monitoring cost. The smaller the size, the lower the resolution, and the higher the cost of surveillance. The larger the image, the greater the perspective distortion and the higher the cost of surveillance. The smaller the size, the less perspective distortion, and the lower the cost of surveillance. The larger it is. It should be noted that the clarity cost index... Perspective Distortion Cost Index is used to measure the level of sharpness. Used to measure the magnitude of perspective distortion.

[0131] In one embodiment, the step of adjusting the resolution cost index... The distance sharpness index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location At that time, for the target to be monitored The cost of surveillance The formula is:

[0132]

[0133] in, It is a set constant.

[0134] Surveillance Costs Also with pre-selected cameras and the target to be monitored Since the distance between them is related, a distance sharpness index can also be introduced in this embodiment. Calculate the cost of surveillance .

[0135] Step S07, determine in the grid The pre-selected camera is installed inside. And the installation height is At that time, the grid For the target to be monitored The cost of placement is calculated using the following formula:

[0136]

[0137] in, This represents the cost of the deployment. The grid is At installation height Install the pre-selected camera at the location. The set of targets to be monitored selected at that time. Indicates the set of targets to be monitored The target number in Represents the grid For the set of targets to be monitored The sum of the surveillance costs of each of the targets to be monitored, , A pre-defined positive integer representing the pre-selected camera. The maximum number of monitorable targets.

[0138] Since a single camera should not monitor too many targets, as this would hinder patrol and switching control, affecting monitoring effectiveness, and excessive scene switching would reduce the effectiveness of the recorded footage for post-event investigation, the maximum number of targets that each pre-selected camera can monitor is set to [number missing]. In practice, having too many targets monitored by the pre-selected cameras also increases the cost of use. This can be mitigated by adding a variable representing the number of targets. The cost function of the increasing function. If a pre-selected camera does not have a pan / tilt head and uses a fixed-focus lens, then it can be set... This means that only one target can be monitored.

[0139] Due to the number of monitored targets and and It is irrelevant, only related to Related, the number of sums is Therefore, for a given installation height of... You need to choose the smallest one. As a monitoring target, this embodiment will consider the possible camera selections, their installation heights, and the corresponding monitoring costs for each grid. Joined one by one, from childhood to adulthood. When smaller, With the number of monitored targets The cost decreases with the increase of monitoring, because the cost of monitoring added later is reduced. It gradually grows larger, and after reaching a certain quantity... With the number of monitored targets The increase is due to the increase, making smallest Each target to be monitored is treated as a grid. Pre-selection of cameras And its installation height is The optimal set of targets to be monitored at that time Assume the next target to be added for monitoring is... At this time, there is

[0140]

[0141] We can obtain:

[0142]

[0143] Therefore, the target to be monitored Add to target set During the process, targets to be monitored are added one by one. Calculate the target set at time Corresponding deployment cost

[0144] ,

[0145] If the next target to be added to the monitoring list The cost of surveillance Then add the target to be monitored. Otherwise, no more targets to be monitored will be added. That is, we get:

[0146] .

[0147] Step S08, will Determined as the grid The minimum deployment cost.

[0148] For each grid, different pre-selected cameras and installation heights can result in different deployment costs. Therefore, choosing a location that minimizes the cost of deployment is crucial. Minimum pre-selected camera and installation height As the cost of placing points in the corresponding grid. ,Right now .

[0149] Step S09: Select the minimum placement cost Minimum grid As a camera placement grid, the pre-selected cameras are subsequently determined. and its installation height and including A set of targets to be monitored. .

[0150] Obtain the placement cost for each grid cell. Next, it is necessary to select the grid with the lowest cost for placement. Therefore, it is necessary to traverse all available grids and select the grid with the lowest placement cost as the placement grid, which determines the cost of obtaining that placement. Corresponding camera selection and its installation height and containing selected A set of targets to be monitored. .

[0151] Step S10: Remove the selected camera placement grid and its selected... For each of the targets to be monitored, repeat steps S07 to S09 until all the targets to be monitored are monitored.

[0152] Remove the selected mesh and its selected mesh in step S09 After identifying the target to be monitored, the cost of placing points on the remaining grid cells needs to be recalculated as the number of targets decreases. Then choose the location cost Minimum grid As a grid for camera placement, and to determine the installation height. Camera selection The removed mesh can reduce its engineering cost. Set it to an extremely large value, or use a masking method to exclude it during the calculation process.

[0153] Step S11: Based on the coordinate positions of all the camera grid points in the 3D model, the coordinate positions of each target to be monitored in each target set in the 3D model, and the camera and its installation height, determine the preset position of each target to be monitored by the camera.

[0154] Once the camera placement is completed, the grid for each placement point is determined. Corresponding pre-selected cameras Model and installation height and the set of targets to be monitored Then, by utilizing the coordinate relationships in three-dimensional space and using pre-selected cameras... By using the reference direction, the pre-selected camera can be calculated. Monitor the required azimuth and focal length for each target to be monitored, and set it as a preset position.

[0155] In practical engineering applications, all calculated preset positions are imported into the system configuration software, and the preset positions of each camera are automatically set one by one. However, due to possible calculation errors, the calculated preset positions may deviate from the actual positions. After installation, during system debugging, each preset position may be fine-tuned through manual correction or automatic correction methods based on target image registration.

[0156] The camera placement and preset position determination method provided in this application embodiment can determine the camera placement and preset position at low cost and high efficiency, effectively improving the design efficiency of video surveillance projects, reducing manual input, and significantly shortening the construction period.

[0157] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limitations; therefore, other examples of exemplary embodiments may have different values.

[0158] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0159] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for determining the location and preset position of a camera deployment point, characterized in that, The method includes: Step S01: Obtain a 3D model of the area to be monitored, wherein the area to be monitored includes: the target to be monitored. ,in, ; Step S02: Divide the orthographic projection of the three-dimensional model onto the horizontal plane into M×N grids; Step S03: Determine the grid based on the scene information of the area to be monitored. Scene engineering cost ,in, The x-coordinate of the grid is represented. The vertical coordinate of the grid is represented. Indicates the installation height of the camera; Step S04: Obtain the target to be monitored. and the target to be monitored Corresponding monitoring requirements; Step S05: Obtain the pre-selected camera Corresponding camera cost , in, express One of the different models of pre-selected cameras, , ; Step S06: Based on the three-dimensional model and the target to be monitored... The corresponding monitoring requirements are determined in the grid. The pre-selected camera is installed inside. Installation height At that time, the grid For the target to be monitored The cost of surveillance ; Step S07, determine in the grid The pre-selected camera is installed inside. And the installation height is At that time, the grid For the target to be monitored The cost of placement is calculated using the following formula: in, This represents the cost of the deployment. The grid is At installation height Install the pre-selected camera at the location. The set of targets to be monitored selected at that time. Indicates the set of targets to be monitored The target number in Represents the grid For the set of targets to be monitored The sum of the surveillance costs of each of the targets to be monitored, , A pre-defined positive integer representing the pre-selected camera. The maximum number of monitorable targets; Step S08, will Determined as the grid The minimum deployment cost; Step S09: Select the minimum placement cost Minimum grid As a camera placement grid, the pre-selected cameras are subsequently determined. and its installation height and including A set of targets to be monitored. ; Step S10: Remove the selected camera placement grid and its selected... For each of the targets to be monitored, repeat steps S07 to S09 until all of the targets to be monitored are monitored. Step S11: Based on the coordinate positions of all the camera grid points in the 3D model, the coordinate positions of each target to be monitored in each target set in the 3D model, and the camera and its installation height, determine the preset position of each target to be monitored by the camera.

2. The method for determining camera placement locations and preset positions according to claim 1, characterized in that, The acquisition of the three-dimensional model of the area to be monitored includes: Acquire a scene image set of the area to be monitored, the scene image set including: multiple scene images of the area to be monitored; Based on the scene image set, point cloud reconstruction is performed to obtain the three-dimensional model.

3. The method for determining the camera placement location and preset position according to claim 1, characterized in that, The target to be monitored The corresponding monitoring requirements include: minimum number of pixels in the horizontal direction. and the smallest pixel in the vertical direction The three-dimensional model and the target to be monitored are used as the basis for this process. The corresponding monitoring requirements are determined in the grid. The pre-selected camera is installed inside. Installation height At that time, the grid For the target to be monitored The cost of surveillance ,include: According to the target to be monitored and its position in the 3D model and the pre-selected camera The parameters are used to determine the pre-selected camera. The target to be monitored can be captured by camera. The image is used to determine the target to be monitored. The maximum possible horizontal pixels of the image and the maximum possible vertical pixels , , , and According to the pre-selected camera The parameters are obtained; Based on the minimum number of pixels in the horizontal direction The minimum pixel in the vertical direction The maximum possible horizontal pixels and the maximum possible vertical pixel Determine the sharpness parameters ; According to the clarity parameter Determine the cost index for clarity. ; Based on the aforementioned 3D model, the target to be monitored is determined. The normal to the pre-selected camera, and the pre-selected camera. The optical axis, based on the angle between the normal and the optical axis. Determine perspective distortion parameters ; Determine the perspective distortion cost index based on the aforementioned perspective distortion parameters. ; Based on the aforementioned 3D model, the target to be monitored is determined. With the pre-selected camera actual distance ; Obtain the pre-selected camera For the target to be monitored Maximum surveillance distance and minimum monitoring distance ; Based on the actual distance Maximum monitoring distance and minimum monitoring distance Determine the distance sharpness index ; According to the clarity cost index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location For the target to be monitored The cost of surveillance ; Alternatively, based on the aforementioned clarity cost index The distance sharpness index and the aforementioned perspective distortion cost index Determine the grid At the installation height Install the pre-selected camera at the location At the time of the target to be monitored The cost of surveillance .

4. The method for determining the camera placement location and preset position according to claim 3, characterized in that, Based on the minimum number of pixels in the horizontal direction The minimum pixel in the vertical direction The maximum possible horizontal pixels and the maximum possible vertical pixel Determine the sharpness parameters The formula is: 。 5. The method for determining the camera placement location and preset position according to claim 4, characterized in that, According to the clarity parameter Determine the cost index for clarity. The formula is: in, It is a logarithmic function. and It is a pre-set scaling factor constant. It is a pre-set threshold.

6. The method for determining the camera placement location and preset position according to claim 3, characterized in that, The perspective distortion parameters are determined based on the angle between the normal and the optical axis. The formula is: ,in, This represents the perspective distortion parameter. This represents the angle between the normal and the optical axis.

7. The method for determining the camera placement location and preset position according to claim 4, characterized in that, The perspective distortion cost index is determined based on the perspective distortion parameters. The formula is: 。 8. The method for determining the camera placement location and preset position according to claim 3, characterized in that, The actual distance Maximum monitoring distance and minimum monitoring distance Determine the distance sharpness index The formula is: 。 9. The method for determining the location and preset position of a camera according to claim 3, characterized in that, According to the clarity cost index and the aforementioned perspective distortion cost index Determine the grid For the target to be monitored The cost of surveillance The formula is: ； in, It is a set constant.

10. The method for determining the location and preset position of a camera according to claim 3, characterized in that, According to the clarity cost index The distance sharpness index and the aforementioned perspective distortion cost index Determine the grid For the target to be monitored The cost of surveillance The formula is: in, It is a set constant.

Images