Image fusion processing method and system

Abstract

The invention provides an image fusion processing method and system, and relates to the technical field of data processing, and the method comprises the steps: obtaining a real-time video stream set through updating monitoring configuration extraction; mapping the real-time video stream set to a regional three-dimensional scene to execute image fusion processing to obtain regional panoramic real-time monitoring; and the real-time risk is monitored and positioned in the regional panorama in real time based on the risk identification cloud, and an early warning management and control instruction is generated based on the positioning result. The technical problems that in the prior art, the operation monitoring coverage is not comprehensive, and the operation risk identification timeliness and certainty are not enough, so that the construction operation risk of the monitoring area is high are solved. The technical effects that no-dead-corner monitoring and risk identification of the whole area of production operation are carried out, specific risk behaviors of different operation areas are adapted, and the safety monitoring and management level of the whole monitoring area is improved are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an image fusion processing method and system. Background Art

[0002] Existing surveillance technologies often face challenges and shortcomings. For example, a construction site surveillance system may have an unreasonable camera layout, resulting in ineffective video coverage of key construction areas, such as the edges of tall buildings or crane operating areas.

[0003] Furthermore, existing risk identification systems may rely on simple motion detection technology, which cannot accurately identify and distinguish between normal operations and potentially high-risk behaviors, such as workers not wearing hard hats or safety belts.

[0004] The shortcomings of these technologies lead to defects in the real-time performance of the monitoring system, which is unable to issue early warnings in a timely manner. The system also lacks certainty and is prone to false alarms or missed alarms, which in turn increases the risks at the construction site and makes it difficult to effectively ensure construction safety.

[0005] In summary, the existing technology has technical problems such as insufficient comprehensive coverage of operation monitoring and insufficient timeliness and certainty in identifying operation risks, which leads to high construction operation risks in the monitored area. Summary of the Invention

[0006] The present application provides an image fusion processing method and system for solving the technical problems in the prior art of insufficient comprehensive coverage of operation monitoring and insufficient timeliness and certainty in identifying operation risks, which lead to high risks of construction operations in the monitored area.

[0007] In view of the above problems, the present application provides an image fusion processing method and system.

[0008] The first aspect of the present application provides an image fusion processing method, which includes: interactively obtaining regional measurement information of a target monitoring area, and using the regional measurement information as modeling data to construct a regional three-dimensional scene; interactively obtaining an original monitoring configuration, and scheduling historical video images according to the original monitoring configuration; mapping the historical video images to the regional three-dimensional scene to perform image fusion processing to obtain a regional blind spot distribution; optimizing the original monitoring configuration according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, K is a positive integer; extracting a real-time video stream set based on the updated monitoring configuration; mapping the real-time video stream set to the regional three-dimensional scene to perform image fusion processing to obtain regional panoramic real-time monitoring; pre-building a risk identification cloud, locating real-time risks in the regional panoramic real-time monitoring based on the risk identification cloud, and generating early warning control instructions based on the positioning results.

[0009] The second aspect of the present application provides an image fusion processing system, which includes: a three-dimensional scene construction unit for interactively obtaining regional measurement information of a target monitoring area, and using the regional measurement information as modeling data to construct a regional three-dimensional scene; a historical information scheduling unit for interactively obtaining an original monitoring configuration, and scheduling historical video images according to the original monitoring configuration; a blind spot distribution analysis unit for mapping the historical video images to the regional three-dimensional scene to perform image fusion processing and obtain a regional blind spot distribution; a monitoring configuration update unit for optimizing the original monitoring configuration according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer; a video extraction execution unit for extracting a real-time video stream set based on the updated monitoring configuration; an image fusion processing unit for mapping the real-time video stream set to the regional three-dimensional scene to perform image fusion processing and obtain regional panoramic real-time monitoring; a risk identification and positioning unit for pre-building a risk identification cloud, locating real-time risks in the regional panoramic real-time monitoring based on the risk identification cloud, and generating early warning control instructions based on the positioning results.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] The method provided by the embodiment of the present application obtains the regional measurement information of the target monitoring area through interaction, and uses the regional measurement information as modeling data to construct a regional three-dimensional scene; interactively obtains the original monitoring configuration, and schedules historical video images according to the original monitoring configuration; maps the historical video images to the regional three-dimensional scene to perform image fusion processing to obtain the regional blind spot distribution; optimizes the original monitoring configuration according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, K is a positive integer; extracts a set of real-time video streams based on the updated monitoring configuration; maps the set of real-time video streams to the regional three-dimensional scene to perform image fusion processing to obtain panoramic real-time monitoring of the region; pre-builds a risk identification cloud, locates real-time risks in the panoramic real-time monitoring of the region based on the risk identification cloud, and generates early warning and control instructions based on the positioning results. The method achieves the technical effect of carrying out blind-angle monitoring and risk identification in the entire production operation area, adapting to specific risk behaviors in different operation areas, and improving the safety monitoring and management level of the entire monitoring area. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart of an image fusion processing method provided in this application;

[0013] Figure 2 A schematic diagram of the process of constructing a regional three-dimensional scene in an image fusion processing method provided in this application;

[0014] Figure 3 This is a structural diagram of an image fusion processing system provided by this application.

[0015] Explanation of the accompanying symbols: three-dimensional scene construction unit 1, historical information scheduling unit 2, blind spot distribution analysis unit 3, monitoring configuration update unit 4, video extraction execution unit 5, image fusion processing unit 6, risk identification and positioning unit 7. DETAILED DESCRIPTION

[0016] This application provides an image fusion processing method and system to address the technical issues in existing technologies, such as insufficient comprehensive coverage of operational monitoring and insufficient timeliness and certainty in identifying operational risks, which lead to high construction operational risks in the monitored area. This method achieves comprehensive monitoring and risk identification across the entire production area, adapts to specific risk behaviors in different operational areas, and improves the safety monitoring and management level of the entire monitored area.

[0017] Below, the technical solutions of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the accompanying drawings.

[0018] Example 1

[0019] like Figure 1 As shown, the present application provides an image fusion processing method, the method comprising:

[0020] A100: Interactively obtain regional measurement information of the target monitoring area, and use the regional measurement information as modeling data to construct a regional three-dimensional scene.

[0021] In one embodiment, Figure 2 As shown, interactively obtaining regional measurement information of the target monitoring area, and using the regional measurement information as modeling data to construct a regional three-dimensional scene, the method step A100 provided in this application also includes:

[0022] Interactively obtain regional boundary input information, and select the target monitoring area based on the regional boundary input information; execute geographic information call with the target monitoring area as a constraint to obtain the regional measurement information; execute three-dimensional model creation based on the regional measurement information to obtain a regional three-dimensional model; import the regional three-dimensional model into the panoramic video monitoring system to generate the regional three-dimensional scene.

[0023] Specifically, in this embodiment, an intuitive user interface is designed to allow the user to draw or input the boundary of the target monitoring area through a graphical tool. The user selects the area boundary through the interface tool, which can be manually drawn, uploaded map marks or entered geographic coordinates to obtain the area boundary input information.

[0024] The system automatically or semi-automatically identifies and selects the target monitoring area based on the boundary information input by the user, and then calls relevant geographic information data from the integrated GIS system based on the target monitoring area as a constraint to obtain detailed measurement information characterizing the target monitoring area, such as terrain data and regional measurement information of building structures.

[0025] Furthermore, using terrain data and building structure information, a preliminary model of the terrain and buildings is constructed in three-dimensional space, and texture and details are added to the three-dimensional model to improve the realism and practicality of the model. The created three-dimensional model is then verified to ensure consistency with the actual monitored area, and adjustments are made as needed to optimize the model to ensure its efficient rendering and interaction in the panoramic video surveillance system, thereby obtaining a three-dimensional model of the area.

[0026] The optimized 3D model is imported into the panoramic video surveillance system and integrated with the system's existing video streams and monitoring functions to generate a 3D scene of the area to ensure that the scene can be synchronized with the real-time video stream and other monitoring data.

[0027] This embodiment pre-constructs the three-dimensional scene of the area, achieving the technical effect of providing a modeling basis for subsequent scene fusion of surveillance videos.

[0028] A200: Interactively obtain an original monitoring configuration, and schedule historical video images according to the original monitoring configuration.

[0029] In one embodiment, the original monitoring configuration is interactively obtained, and historical video images are scheduled according to the original monitoring configuration. Step A200 of the method provided in this application further includes:

[0030] The original monitoring configuration includes multiple configuration parameters of multiple original cameras, wherein each configuration parameter includes a camera position parameter, a field of view parameter, and a device model parameter; a plurality of original video images are obtained by scheduling retrieval in a monitoring storage module according to the multiple configuration parameters; and acquisition time alignment processing is performed on the multiple original video images to obtain multiple regional video images, and the multiple regional video images constitute the historical video image.

[0031] Specifically, in this embodiment, the original monitoring configuration of the original monitoring camera deployment and the deployed monitoring camera parameters of the target monitoring area is interactively obtained, and the original monitoring configuration includes multiple configuration parameters of multiple original cameras.

[0032] The configuration parameters of each original camera include camera position parameters, field of view parameters, and device model parameters. The camera position parameters include specific latitude and longitude coordinates, altitude, orientation, etc. The field of view parameters include horizontal and vertical field of view angles, and may also include focal length information. The device model parameters include technical specifications such as the camera's brand, model, and resolution.

[0033] According to the multiple configuration parameters, a monitoring storage module schedules retrieval to obtain multiple original video images, and the multiple original video images are acquired by the multiple original cameras at historical times.

[0034] Time alignment processing is performed on the retrieved multiple original video images to ensure that all images correspond to the same time point or time period, thereby obtaining the multiple regional video images for subsequent analysis, and the multiple regional video images constitute the historical video images.

[0035] A300: Map the historical video image to the regional three-dimensional scene to perform image fusion processing to obtain the regional blind spot distribution.

[0036] In one embodiment, the historical video image is mapped to the regional three-dimensional scene and image fusion processing is performed to obtain the regional blind spot distribution. Step A300 of the method provided in this application further includes:

[0037] A plurality of original field of view models are constructed according to the plurality of field of view range parameters of the plurality of original cameras; the plurality of original field of view models are spliced in the regional three-dimensional scene according to the plurality of camera position parameters of the plurality of original cameras to obtain a plurality of field of view non-overlapping areas; the plurality of regional video images are spliced in the regional three-dimensional scene according to the plurality of camera position parameters of the plurality of original cameras to obtain a plurality of video non-overlapping areas; the intersection of the plurality of field of view non-overlapping areas and the plurality of video non-overlapping areas is calculated to obtain K monitoring blind spot parameters, and the K monitoring blind spot parameters constitute the regional blind spot distribution.

[0038] Specifically, in this embodiment, the field of view parameters of each original camera (such as horizontal and vertical field of view angles) are used to construct an original field of view model for each original camera in the virtual three-dimensional scene of the area. These models represent the monitoring area that each original camera can cover in the target monitoring area.

[0039] Furthermore, each original field of view model is combined with the camera position parameters (such as latitude and longitude coordinates, height, and orientation) of the corresponding original camera to ensure that the position of each original field of view model in the regional three-dimensional scene is accurate.

[0040] On this basis, in the regional 3D scene, the multiple original field of view models are spatially spliced to identify non-overlapping areas between the fields of view, thereby obtaining the multiple non-overlapping fields of view areas. It should be understood that the non-overlapping fields of view areas are monitoring blind spots, that is, areas not covered by the fields of view of multiple cameras.

[0041] Furthermore, the multiple camera position parameters of the multiple original cameras are used as multiple geographic spatial coordinates, and the multiple regional video images are spliced in the regional three-dimensional scene with reference to the multiple geographic spatial coordinates, and areas that are not covered in all video images are identified, and these areas are the non-overlapping areas of the multiple videos.

[0042] The multiple non-overlapping areas of visual field and the multiple non-overlapping areas of video are compared, and their intersection is calculated to accurately find the areas that are blind spots in both physical space and video surveillance, and obtain K monitoring blind spot parameters.

[0043] The K monitoring blind area parameters are combined to construct the regional blind area distribution of the entire target monitoring area, and the regional blind area distribution clearly shows the location, size and range of all monitoring blind areas.

[0044] This embodiment obtains the blind spot distribution of the area based on two-dimensional analysis of video analysis and field of view analysis, providing a reliable reference for subsequent iteration and optimization of the monitoring system, such as adjusting the position of existing cameras or adding new cameras to cover blind spots.

[0045] A400: Optimize the original monitoring configuration according to the blind spot distribution of the area to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer.

[0046] In one embodiment, the original monitoring configuration is optimized according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer. Step A400 of the method provided in this application further includes:

[0047] K updated position parameters are calculated based on the K monitoring blind spot parameters; field of view matching is performed based on the K updated position parameters and the K monitoring blind spot parameters to obtain K updated field of view range configurations; K updated device models are obtained based on the comparison of the K updated field of view range configurations; the K monitoring cameras are obtained by scheduling based on the K updated device models, and the K monitoring cameras are arranged according to the K updated position parameters; the multiple original cameras and the K monitoring cameras constitute the updated monitoring configuration.

[0048] Specifically, in this embodiment, K monitoring blind spot parameters are analyzed in detail to understand the size, shape and position of each blind spot. Based on the blind spot parameters, geometric and spatial optimization algorithms are used to calculate the newly deployed camera positions. The new positions should be able to cover the existing blind spots. For each newly calculated camera position, a field of view matching analysis is performed to ensure that the camera's field of view can overlap and cover the blind spot. According to the updated position parameters, the field of view configuration is adjusted to maximize the coverage of the monitoring area and reduce overlap. According to the updated field of view configuration, the available camera models on the market are compared, and K updated device models that meet the field of view, resolution and environmental requirements are selected.

[0049] Based on the K updated device models, K surveillance cameras are dispatched and obtained from suppliers or inventory. According to the calculated K updated location parameters, K surveillance cameras are deployed on site to ensure that they are correctly installed and pointed at the predetermined K blind spots.

[0050] The multiple original cameras and the K surveillance cameras constitute the updated surveillance configuration, and the updated surveillance configuration can achieve full coverage video surveillance of the target surveillance area.

[0051] A500: Extract and obtain a real-time video stream set based on the updated monitoring configuration.

[0052] A600: Map the real-time video stream set to the three-dimensional scene of the area to perform image fusion processing to obtain panoramic real-time monitoring of the area.

[0053] Specifically, in this embodiment, the real-time monitoring video call is performed based on the multiple original cameras and K monitoring cameras in the updated monitoring configuration to obtain the real-time video stream set.

[0054] Using geographic information system (GIS) data and three-dimensional modeling technology, several real-time video streams in the real-time video stream set are mapped to corresponding positions in the three-dimensional scene of the area, so that the several real-time video streams are spliced according to their actual positions and directions to form a continuous monitoring picture. At the same time, fusion technology needs to be applied in the overlapping area of the images to smoothly connect different images, reduce the seam effect, and obtain panoramic real-time monitoring of the area.

[0055] A700: Pre-build a risk identification cloud, based on which the real-time risks in the area are monitored and located in a panoramic real-time manner, and early warning and control instructions are generated based on the location results.

[0056] In one embodiment, the risk identification cloud is pre-built, and the method step A700 provided in this application further includes:

[0057] The three-dimensional scene of the region is divided according to the risk pattern to obtain M risk identification regions, wherein the M risk identification regions have M risk behavior pattern sets; M risk identification networks are constructed and generated according to the M risk behavior pattern sets; M fusion image transmission channels are configured in the three-dimensional scene of the region according to the M risk identification regions, and the output ends of the M fusion image transmission channels are connected to the input ends of the M risk identification networks; the input ends of the M risk identification networks are connected to the risk identification cloud to complete the construction of the risk identification cloud.

[0058] In one embodiment, based on the risk identification, the cloud performs panoramic real-time monitoring and locates real-time risks in the area, and generates early warning and control instructions based on the location results. The method step A700 provided in this application also includes:

[0059] The area is divided into the M risk identification areas for panoramic real-time monitoring, and the division results are sent to the M risk identification networks via the M fusion image transmission channels to perform risk positioning, thereby obtaining M regional risk positioning results; the risk identification cloud integrates the M regional risk positioning results to generate the early warning control instructions.

[0060] Specifically, in this embodiment, the target monitoring area is subdivided into multiple operation areas with specific risk characteristics to obtain M risk identification areas, wherein the M risk identification areas have M risk behavior pattern sets, and each risk behavior pattern set contains one or more groups of risk behavior pattern images of one or more risk behavior patterns prohibited in each risk identification area.

[0061] M risk identification networks are constructed and generated based on the M risk behavior pattern sets. Since the construction methods of the M risk identification networks are consistent, this embodiment takes the construction process of the first risk identification network of the first risk identification area as an example and elaborates on the technical solution in detail.

[0062] The first risk identification area has multiple prohibited risk behaviors. The first risk behavior pattern set among the M risk behavior pattern sets includes multiple sets of historical risk behavior images of the multiple prohibited risk behaviors. A standard risk identification model is then constructed based on a back-propagation neural network. Conventional back-propagation neural network training methods are employed, using the multiple sets of historical risk behavior images and the multiple prohibited risk behaviors as training data to train the standard risk identification model and construct a first risk identification network. The video stream of the first risk identification area is then transmitted to the first risk identification network, thereby accurately and rapidly identifying the real-time risk behavior dynamics of the first risk identification area. Similarly, M risk identification networks are constructed and generated based on the M risk behavior pattern sets.

[0063] An adaptive fusion image transmission channel is configured for each operation area to ensure that the real-time video stream can be transmitted for different risk identification networks. Specifically, M fusion image transmission channels are configured in the three-dimensional scene of the area according to the M risk identification areas.

[0064] Connect the output ends of the M fused image transmission channels to the input ends of the M risk identification networks; connect the input ends of the M risk identification networks to the risk identification cloud to complete the construction of the risk identification cloud, so as to integrate the risk identification networks of each operating area with the risk identification cloud and realize centralized management and analysis of data.

[0065] The area is divided into the M risk identification areas for panoramic real-time monitoring, and the division results are sent to the M risk identification networks via the M fusion image transmission channels to perform risk positioning, so as to utilize the risk identification network to analyze the real-time video stream of each operation area, locate specific risk behaviors, and obtain M regional risk positioning results.

[0066] The risk identification cloud integrates the M regional risk positioning results to generate the early warning control instructions, which are M differentiated control instructions to adapt to the risk mitigation needs of different regions. The M differentiated control instructions are automatically distributed to relevant personnel or systems, and a manual intervention interface is provided to deal with complex situations.

[0067] This embodiment achieves the technical effect of conducting no-blind-angle monitoring and risk identification in the entire production operation area, adapting to specific risk behaviors in different operation areas, and improving the safety monitoring and management level of the entire monitoring area.

[0068] Example 2

[0069] Based on the same inventive concept as the image fusion processing method in the above embodiment, Figure 3 As shown, the present application provides an image fusion processing system, wherein the system includes:

[0070] The three-dimensional scene construction unit 1 is used to interactively obtain regional measurement information of a target monitoring area, and use the regional measurement information as modeling data to construct a regional three-dimensional scene.

[0071] The historical information scheduling unit 2 is used to interactively obtain the original monitoring configuration and schedule historical video images according to the original monitoring configuration.

[0072] The blind spot distribution analysis unit 3 is used to map the historical video image to the regional three-dimensional scene to perform image fusion processing to obtain the regional blind spot distribution.

[0073] The monitoring configuration updating unit 4 is configured to optimize the original monitoring configuration according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer.

[0074] The video extraction execution unit 5 is configured to extract and obtain a real-time video stream set based on the updated monitoring configuration.

[0075] The image fusion processing unit 6 is used to map the real-time video stream set to the regional three-dimensional scene to perform image fusion processing to obtain panoramic real-time monitoring of the region.

[0076] The risk identification and positioning unit 7 is used to pre-build a risk identification cloud, monitor and locate real-time risks in the area in real time based on the risk identification cloud, and generate early warning control instructions based on the positioning results.

[0077] In one embodiment, the three-dimensional scene construction unit 1 further includes:

[0078] Interactively obtain regional boundary input information, and select the target monitoring area based on the regional boundary input information; execute geographic information call with the target monitoring area as a constraint to obtain the regional measurement information; execute three-dimensional model creation based on the regional measurement information to obtain a regional three-dimensional model; import the regional three-dimensional model into the panoramic video monitoring system to generate the regional three-dimensional scene.

[0079] In one embodiment, the historical information scheduling unit 2 further includes:

[0080] The original monitoring configuration includes multiple configuration parameters of multiple original cameras, wherein each configuration parameter includes a camera position parameter, a field of view parameter, and a device model parameter; a plurality of original video images are obtained by scheduling retrieval in a monitoring storage module according to the multiple configuration parameters; and acquisition time alignment processing is performed on the multiple original video images to obtain multiple regional video images, and the multiple regional video images constitute the historical video image.

[0081] In one embodiment, the blind spot distribution analysis unit 3 further includes:

[0082] A plurality of original field of view models are constructed according to the plurality of field of view range parameters of the plurality of original cameras; the plurality of original field of view models are spliced in the regional three-dimensional scene according to the plurality of camera position parameters of the plurality of original cameras to obtain a plurality of field of view non-overlapping areas; the plurality of regional video images are spliced in the regional three-dimensional scene according to the plurality of camera position parameters of the plurality of original cameras to obtain a plurality of video non-overlapping areas; the intersection of the plurality of field of view non-overlapping areas and the plurality of video non-overlapping areas is calculated to obtain K monitoring blind spot parameters, and the K monitoring blind spot parameters constitute the regional blind spot distribution.

[0083] In one embodiment, the monitoring configuration updating unit 4 further includes:

[0084] Calculate and obtain K updated position parameters according to the K monitoring blind area parameters;

[0085] Performing field of view matching according to the K updated position parameters and the K monitoring blind zone parameters to obtain K updated field of view range configurations;

[0086] Obtain K updated device models according to the K updated field of view configuration comparison;

[0087] Obtaining the K surveillance cameras according to the K updated device model scheduling, and deploying the K surveillance cameras according to the K updated location parameters;

[0088] The multiple original cameras and the K monitoring cameras constitute the updated monitoring configuration.

[0089] In one embodiment, the risk identification and positioning unit 7 further includes:

[0090] The three-dimensional scene of the region is divided according to the risk pattern to obtain M risk identification regions, wherein the M risk identification regions have M risk behavior pattern sets; M risk identification networks are constructed and generated according to the M risk behavior pattern sets; M fusion image transmission channels are configured in the three-dimensional scene of the region according to the M risk identification regions, and the output ends of the M fusion image transmission channels are connected to the input ends of the M risk identification networks; the input ends of the M risk identification networks are connected to the risk identification cloud to complete the construction of the risk identification cloud.

[0091] In one embodiment, the risk identification and positioning unit 7 further includes:

[0092] The area is divided into the M risk identification areas for panoramic real-time monitoring, and the division results are sent to the M risk identification networks via the M fusion image transmission channels to perform risk positioning, thereby obtaining M regional risk positioning results; the risk identification cloud integrates the M regional risk positioning results to generate the early warning control instructions.

[0093] Any of the methods or steps described above may be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs may be recognized by various types of computer processors to implement any of the methods or steps described above.

[0094] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention shall fall within the scope of patent protection of the present invention.

Claims

1. A method for image fusion processing, characterized in that: The method comprises: Interactively obtain regional measurement information of the target monitoring area, and use the regional measurement information as modeling data to construct a regional three-dimensional scene; interactively obtaining an original monitoring configuration, and scheduling historical video images according to the original monitoring configuration; Mapping the historical video image to the regional three-dimensional scene to perform image fusion processing to obtain regional blind spot distribution; Optimizing the original monitoring configuration according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer; Extract and obtain a real-time video stream set based on the updated monitoring configuration; Mapping the real-time video stream set to the regional three-dimensional scene to perform image fusion processing to obtain panoramic real-time monitoring of the region; A risk identification cloud is pre-built, and real-time risks are located in the area through panoramic real-time monitoring based on the risk identification cloud, and early warning control instructions are generated based on the positioning results.

2. The image fusion processing method according to claim 1, characterized in that: Interactively obtaining regional measurement information of the target monitoring area, and using the regional measurement information as modeling data to construct a regional three-dimensional scene, the method further comprising: interactively obtaining area boundary input information, and selecting the target monitoring area according to the area boundary input information; Executing geographic information call with the target monitoring area as a constraint to obtain the regional measurement information; Execute three-dimensional model creation based on the regional measurement information to obtain a regional three-dimensional model; The three-dimensional model of the area is imported into a panoramic video surveillance system to generate a three-dimensional scene of the area.

3. The image fusion processing method according to claim 1, characterized in that: Interactively obtaining an original monitoring configuration, and scheduling historical video images according to the original monitoring configuration, the method further comprising: The original monitoring configuration includes multiple configuration parameters of multiple original cameras, wherein each configuration parameter includes a camera position parameter, a field of view parameter, and a device model parameter; Scheduling retrieval in the monitoring storage module to obtain multiple original video images according to the multiple configuration parameters; An acquisition time alignment process is performed on the multiple original video images to obtain multiple regional video images, and the multiple regional video images constitute the historical video image.

4. The image fusion processing method according to claim 3, characterized in that: Mapping the historical video image to the regional three-dimensional scene and performing image fusion processing to obtain regional blind spot distribution, the method further includes: Constructing a plurality of original field of view models according to a plurality of field of view parameters of the plurality of original cameras; splicing the multiple original field of view models in the regional three-dimensional scene according to multiple camera position parameters of the multiple original cameras to obtain multiple non-overlapping field of view areas; splicing the plurality of regional video images in the regional three-dimensional scene according to the plurality of camera position parameters of the plurality of original cameras to obtain a plurality of video non-overlapping regions; The intersection of the multiple field of view non-overlapping areas and the multiple video non-overlapping areas is calculated to obtain K monitoring blind area parameters, and the K monitoring blind area parameters constitute the regional blind area distribution.

5. The image fusion processing method according to claim 4, characterized in that: Optimizing the original monitoring configuration according to the regional blind spot distribution to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer. The method further includes: Calculate and obtain K updated position parameters according to the K monitoring blind area parameters; Performing field of view matching according to the K updated position parameters and the K monitoring blind zone parameters to obtain K updated field of view range configurations; Obtain K updated device models according to the K updated field of view configuration comparison; Obtaining the K surveillance cameras according to the K updated device model scheduling, and deploying the K surveillance cameras according to the K updated location parameters; The multiple original cameras and the K monitoring cameras constitute the updated monitoring configuration.

6. The image fusion processing method according to claim 1, characterized in that: Pre-building a risk identification cloud, the method further includes: Dividing the regional three-dimensional scene according to the risk pattern to obtain M risk identification areas, wherein the M risk identification areas have M risk behavior pattern sets; Constructing and generating M risk identification networks according to the M risk behavior pattern sets; According to the M risk identification areas, M fusion image transmission channels are configured in the regional three-dimensional scene, and the output ends of the M fusion image transmission channels are connected to the input ends of the M risk identification networks; The input ends of the M risk identification networks are connected to the risk identification cloud to complete the construction of the risk identification cloud.

7. The image fusion processing method according to claim 6, characterized in that: Based on the risk identification cloud, real-time monitoring and positioning of real-time risks in the area are performed in a panoramic real-time manner, and early warning and control instructions are generated based on the positioning results. The method further includes: Dividing the area into the M risk identification areas for panoramic real-time monitoring, and sending the division results to the M risk identification networks via the M fusion image transmission channels to perform risk positioning, thereby obtaining M regional risk positioning results; The risk identification cloud integrates the M regional risk positioning results and generates the early warning control instructions.

8. An image fusion processing system, characterized in that: A system for implementing an image fusion processing method according to any one of claims 1 to 7, comprising: A three-dimensional scene construction unit, configured to interactively obtain regional measurement information of a target monitoring area and construct a regional three-dimensional scene using the regional measurement information as modeling data; A historical information scheduling unit, configured to interactively obtain an original monitoring configuration and schedule historical video images according to the original monitoring configuration; A blind spot distribution analysis unit, configured to map the historical video image to the regional three-dimensional scene, perform image fusion processing, and obtain regional blind spot distribution; a monitoring configuration updating unit, configured to optimize the original monitoring configuration according to the blind spot distribution of the area to obtain an updated monitoring configuration, wherein the updated monitoring configuration includes K monitoring cameras, where K is a positive integer; A video extraction execution unit, configured to extract and obtain a real-time video stream set based on the updated monitoring configuration; An image fusion processing unit is used to map the real-time video stream set to the regional three-dimensional scene to perform image fusion processing to obtain a panoramic real-time monitoring of the region; The risk identification and positioning unit is used to pre-build a risk identification cloud, monitor and locate real-time risks in the area in real time based on the risk identification cloud, and generate early warning and control instructions based on the positioning results.

Images