CN110458897B

CN110458897B - Multi-camera automatic calibration method and system and monitoring method and system

Info

Publication number: CN110458897B
Application number: CN201910746029.7A
Authority: CN
Inventors: 孙晓路; 刘亚樵
Original assignee: Beijing Jijia Technology Co ltd
Current assignee: Beijing Jijia Technology Co ltd
Priority date: 2019-08-13
Filing date: 2019-08-13
Publication date: 2020-12-01
Anticipated expiration: 2039-08-13
Also published as: CN110458897A

Abstract

The application belongs to the technical field of video monitoring, and particularly relates to a multi-camera automatic calibration method and system and a multi-camera cooperative monitoring method and system. The multi-camera automatic calibration method comprises the following steps: acquiring first image information of a designated area; constructing a visual map of the designated area according to the first image information; acquiring a plurality of second image information captured by a plurality of cameras to be calibrated, and performing feature point matching on the plurality of second image information and the visual map one by one; and acquiring the pose information and the preset parameter information of each camera to be calibrated one by one according to the matching result, and calibrating each camera to be calibrated one by one according to the pose information and the preset parameter information. The multi-camera automatic calibration method and system, and the monitoring method and system can overcome the defects of high complexity, low universality and the like of the existing camera calibration method, and realize a semi-automatic multi-camera self-calibration mode which can be completed by only operating a single camera.

Description

Multi-camera automatic calibration method and system and monitoring method and system

Technical Field

The application belongs to the technical field of video monitoring, and particularly relates to a multi-camera automatic calibration method and system and a multi-camera cooperative monitoring method and system.

Background

With the change and the increasing complexity of economic environment, political environment and social environment, the requirements of various industries on security and protection are continuously increased, and higher requirements on the applicability, flexibility and humanization of the security and protection technology are also provided. Video surveillance systems have also become the hottest research direction in the scientific research and industrial fields in recent years as the most direct and effective system integration scheme. In recent years, security solutions using cameras as input quickly occupy favorable positions in the market by virtue of excellent characteristics of high performance, high convenience, high stability and the like. The intelligent video monitoring (IVS) completes functions of pedestrian recognition, license plate recognition, human body detection and the like by analyzing video signals transmitted by the camera, makes further thinking and judgment according to recognized contents, takes action on the basis, and can really enable the video monitoring to replace human beings to observe the world through a series of processes.

Most of the mature IVSs in the market are completed based on a single-camera system or a non-cascaded camera group, the single camera provides a 2D picture, certain defects exist in the accuracy of information processing, and meanwhile, the problems of actual scenes such as object shielding and multi-target tracking cannot be well solved. Therefore, the multi-camera cooperation mode becomes a new development and research hotspot. The multi-camera cooperation, namely, the information of a plurality of camera ends is fused through the interaction of the information among a plurality of cameras, the task of the most suitable current environment is distributed for each camera, the total target of the system is completed in cooperation, the problems of narrow visual field and target shielding of a single camera can be solved, and other requirements in the field of placement can be met. The integration mode has better effect in the current scientific research of pedestrian detection, target tracking and pedestrian re-identification, and meanwhile, the integration mode slowly occupies an important position in an intelligent security system.

Although multiple cameras have great advantages in information processing and integration, a precise calibration process can enable a multiple-camera system to work normally, and generally, camera calibration mainly comprises a traditional camera calibration method, an active vision camera calibration method and a camera self-calibration method.

In the conventional calibration method, calibration is performed by using known scene structure information, which usually involves a complicated calibration process and high-precision known calibration information, and requires complicated operations by professionals. Moreover, each time the position of the camera set is changed, a recalibration operation is required.

The active visual camera calibration method is characterized in that motion information of some cameras is known, and a calibration result is obtained in a motion equation solving mode.

The camera self-calibration algorithm is calibrated by depending on the corresponding relation among a plurality of images, and the method only needs to establish the relation among the images so as to finish the camera calibration; however, this method belongs to non-linear calibration and is not highly robust.

Disclosure of Invention

In order to solve at least one of the above technical problems, the present application provides a multi-camera automatic calibration method and system, and a multi-camera cooperative monitoring method and system.

In a first aspect, the application discloses a method for automatically calibrating multiple cameras, comprising the following steps:

acquiring first image information of a designated area;

constructing a visual map of the designated area according to the first image information;

acquiring a plurality of pieces of second image information captured by a plurality of cameras to be calibrated, and performing feature point matching on the plurality of pieces of second image information and the visual map one by one;

and acquiring the pose information and the preset parameter information of each camera to be calibrated one by one according to the matching result, and calibrating each camera to be calibrated one by one according to the pose information and the preset parameter information.

According to at least one embodiment of the present application, the first image information includes video frame data and an initial camera pose corresponding to each image frame, wherein

Constructing a visual map of the designated area from the first image information, comprising:

sequentially overlapping the 3D point clouds of all the image frames according to the acquisition time line of the video frame data to generate a preliminary 3D model of the designated area;

extracting 2D feature points in each image frame according to the acquisition time line of the video frame data, and performing feature point matching between all the image frames to generate 3D structures of all the image frames;

mapping the 3D structure onto the preliminary 3D model, thereby generating the visual map.

According to at least one embodiment of the present application, the method for automatically calibrating multiple cameras further includes:

transmitting the obtained visual map in real time; and/or;

and acquiring information of the camera to be calibrated after calibration is completed, synchronizing the information to the visual map, and transmitting the synchronized visual map in real time.

According to at least one embodiment of the present application, performing feature point matching on a plurality of pieces of second image information with the visual map one by one includes:

and matching the 2D characteristic points in the second image information with the characteristic points of the corresponding point cloud in the 3D visual map.

In a second aspect, the present application discloses an automatic calibration system for multiple cameras, comprising:

the initial camera is used for acquiring first image information of the designated area;

the visual map building module is used for building a visual map of the designated area according to the first image information;

the characteristic point matching module is used for acquiring a plurality of pieces of second image information captured by a plurality of cameras to be calibrated and performing characteristic point matching on the plurality of pieces of second image information and the visual map one by one;

and the calibration module is used for acquiring the pose information and the preset parameter information of each camera to be calibrated one by one according to the matching result and calibrating each camera to be calibrated one by one according to the pose information and the preset parameter information.

According to at least one embodiment of the present application, the initial camera is an RGBD camera and is provided with a visual SLAM function, wherein

The first image information includes video frame data and an initial camera pose corresponding to each image frame.

According to at least one embodiment of the present application, the visual map building module comprises:

the preliminary 3D model building unit is used for sequentially superposing all the 3D point clouds of the image frames according to the acquisition time line of the video frame data to generate a preliminary 3D model of the designated area;

the 3D structure generating unit is used for extracting 2D feature points in each image frame according to the acquisition time line of the video frame data, matching the feature points among all the image frames and generating 3D structures of all the image frames; and

a visual map generation unit for mapping the 3D structure onto the preliminary 3D model, thereby generating the visual map.

According to at least one embodiment of the present application, the visual map building module is further configured to obtain information of the camera to be calibrated after calibration is completed, and synchronize the information to the visual map; wherein

The automatic calibration system of many cameras still includes:

and the visualization unit is used for displaying the visual map in real time according to the visual map information transmitted by the visual map construction module in real time.

In a third aspect, the present application discloses a multi-camera cooperative monitoring method, which includes the following steps:

calibrating each camera in the designated area by the multi-camera automatic calibration method in any one of the first aspect;

capturing a plurality of third image information of the designated area through the calibrated plurality of cameras;

carrying out target detection processing and/or multi-target tracking processing and/or pedestrian re-identification processing according to the third image information;

and dynamically displaying the processing result in a 3D form in the visual map obtained in the multi-camera automatic calibration step.

In a fourth aspect, the present application discloses a multi-camera cooperative monitoring system, comprising:

a multi-camera calibration system, which employs the multi-camera calibration system of any one of the second aspects and is configured to calibrate each camera in a specified area;

a plurality of cameras for capturing a plurality of third image information of the designated area after calibration is completed;

the monitoring processing module is used for carrying out target detection processing and/or multi-target tracking processing and/or pedestrian re-identification processing according to the third image information;

and the display module is used for dynamically displaying the processing result in a 3D form in a visual map, wherein the visual map is generated by the multi-camera calibration system when each camera in the designated area is calibrated.

The application has at least the following beneficial technical effects:

according to the multi-camera automatic calibration method and system and the multi-camera cooperative monitoring method and system, the 3D model of the designated area is generated, and the calibration of the camera to be calibrated is carried out according to the existing model, so that the defects of high complexity, low universality and the like of the existing camera calibration method are overcome, and a semi-automatic multi-camera self-calibration mode which can be completed only by operating a single camera is realized; meanwhile, a series of intelligent analysis processing processes can be better realized through a calibrated multi-camera cooperation mode, and 3D analysis information is given.

Drawings

FIG. 1 is a flow chart of a preferred embodiment of the multi-camera auto-calibration method of the present application;

FIG. 2 is a flowchart of a preferred embodiment of a visual map construction method in the multi-camera automatic calibration method of the present application;

FIG. 3 is a block diagram of a preferred embodiment of the multi-camera auto-calibration system of the present application;

FIG. 4 is a block diagram of a preferred embodiment of a visual map building block in the multi-camera auto-calibration system of the present application;

FIG. 5 is a flow chart of a preferred embodiment of the multi-camera cooperative monitoring method of the present application;

fig. 6 is a block diagram of a preferred embodiment of the multi-camera cooperative monitoring system of the present application.

Detailed Description

In order to make the implementation objects, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described in more detail below with reference to the drawings in the embodiments of the present application. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are a subset of the embodiments in the present application and not all embodiments in the present application. The embodiments described below with reference to the drawings are exemplary and intended to be used for explaining the present application and should not be construed as limiting the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application. Embodiments of the present application will be described in detail below with reference to the accompanying drawings.

Before formally describing each method and system of the present application, the full-text english and full-text chinese explanations are performed on some special terms and special english abbreviations in the related industries, which are specifically as follows:

1)SLAM

SLAM (simultaneous localization and mapping), immediate localization and mapping. The problem can be described as: if a robot is placed at an unknown position in an unknown environment and there is a way to draw a complete map of the environment while the robot is moving, the complete map (a continuous map) refers to every corner where a room can enter without being obstructed. SLAM requires real-time.

2)Visual SLAM

Visual SLAM, SLAM technology based on image matching algorithms.

3)SFM

Sfm (structure from motion), a motion restoration structure, is a technique that obtains camera parameters by analyzing an image sequence and performs three-dimensional reconstruction. SFM does not require real-time and can be processed off-line.

4)3D map

And the SLAM or the SFM projects the characteristic points in the two-position camera picture to a three-dimensional space to generate a 3D point cloud map. These 3D point clouds are key information for camera localization.

5) Camera calibration

In image measurement processes and machine vision applications, in order to determine the correlation between the three-dimensional geometric position of a certain point on the surface of an object in space and the corresponding point in the image, a geometric model of camera imaging must be established, and the parameters of the geometric model are the parameters of the camera. Under most conditions, the parameters must be obtained through experiments and calculation, and the process of solving the parameters is called camera calibration (or video camera calibration).

6) Intelligent video monitoring

The intelligent video monitoring (IVS intelligent video surveillance) is a full-automatic all-weather real-time monitoring intelligent system which adopts image processing, mode recognition and computer vision technology, filters useless or interference information of video pictures by means of strong data processing capacity of a computer, automatically recognizes different objects, analyzes and extracts key useful information in a video source, quickly and accurately positions an accident site, judges abnormal conditions in a monitoring picture, sends out an alarm or triggers other actions in a fastest and optimal mode, and effectively performs early warning, in-situ processing and timely evidence obtaining after the accident.

The multi-camera automatic calibration method and system, and the multi-camera cooperative monitoring method and system of the present application will be further described in detail with reference to fig. 1 to 6.

As can be seen from the discussion in the background art, in the intelligent monitoring system, the multi-camera mode can bring about a great improvement in the security effect; however, how to complete the multi-camera cooperation and how to complete the calibration become a key difficulty in the task, so that the method for completing the accurate and fast semi-automatic camera calibration by using the thought of SLAM/SFM is the key point of the application.

In a first aspect, as shown in fig. 1, the present application discloses a method for automatically calibrating multiple cameras, which includes the following steps:

step S11, first image information of the designated area is acquired.

The designated area can be selected as various suitable scene areas according to the needs of the user, such as areas in a living room, a front desk of an office area, a classroom of a school and the like in a user room.

Further, in this step S11, the capturing of the first image information of the designated area may be performed by various suitable scan acquisition devices; in this embodiment, the scanning and collecting device is defined as an initial camera, and it should be noted that the initial camera may exist independently of a subsequent camera to be calibrated, and of course, in some cases, the camera to be calibrated may also be used as the initial camera.

Further, in this embodiment, the scan collecting device is preferably an RGBD (i.e. having a color + depth mode) camera, which may be a binocular camera, a tof (time of light) or a module composed of structured light and RGB, and the like; moreover, the scanning acquisition device has a Visual SLAM function. The RGBD camera has the advantages that the RGBD camera can generate 3D point cloud, and RGB texture corresponds to the point cloud, so that the accuracy of the model is improved. If only an RGB camera is used, the accuracy of the model and the density of the point cloud are greatly reduced.

Further, the shooting (or scanning) of a certain area by the initial camera is a continuous process, and a plurality of suitable shooting modes can be adopted; in the present embodiment, a user manual mode and a robot automatic mode are preferable.

The user manual mode is that a user holds the RGBD camera for scanning; meanwhile, a visualization device (a mobile phone, a computer, VR glasses and the like) can be arranged to display the currently established model in real time, so that a user is helped to complete the scanning process more conveniently.

The robot automatic modeling adopts corresponding cameras with autonomous moving capability, such as a household robot, an unmanned aerial vehicle and the like, and automatically completes whole-house scanning through a SLAM + autonomous navigation algorithm (scanning path planning).

Step S12, a visual map (3D map) of the designated area is constructed from the first image information.

The construction of the visual map may be performed in any suitable manner known at present. In this embodiment, as described above, the initial camera is an RGBD camera and has a Visual SLAM function, and when the initial camera is used for acquisition, the movement track of the initial camera and the real-time pose of the initial camera can be recorded; specifically, the acquired first image information includes video frame data and an initial camera pose corresponding to each image frame.

Based on this, as shown in fig. 2, in this embodiment, the step S12 may be specifically divided into:

and S121, sequentially overlapping all the 3D point clouds of the image frames according to the acquisition time line of the video frame data, and generating a preliminary 3D model of the designated area in real time through processing by a preset optimization and meshing method.

Step S122, extracting 2D feature points (feature points) in each image frame according to the acquisition timeline of the video frame data, and further preferably performing feature point matching between all 2D image frames by using an SFM technique to generate 3D structures of all image frames.

And step S123, mapping the 3D structure to the preliminary 3D model so as to generate a visual map. The selected corresponding algorithm can perform global optimization on all point clouds to ensure the optimal mapping effect.

The process from step S122 to step S123 corresponds to the process of texture mapping the 2D image onto the 3D point cloud, so that a textured whole house model can be generated. In addition, since the feature points on the 2D image are also mapped to the 3D points in the space, a visual map can be generated. The visual map is composed of 3D coordinates of feature points, 2D image features and corresponding camera poses and is used for subsequent camera calibration.

It should be further noted that the above-mentioned visual map generation process may be offline or online, and depending on the computing power of the server (processor), the server may be deployed in the cloud or in the local (such as a home PC, or even a robot, a mobile phone, or other devices with sufficient computing power). The application does not limit the specific SFM algorithm, the key point algorithm, and the mapping optimization algorithm.

And step S13, acquiring a plurality of second image information captured by a plurality of cameras to be calibrated, and performing feature point matching on the plurality of second image information and the visual map one by one.

Specifically, in this embodiment, a user can freely secure and protect a plurality of cameras to be calibrated, and after the indoor 3D visual map is obtained, matching (also called key point matching) is performed between the 2D feature point of each second image information obtained from each camera to be calibrated and the feature point of the corresponding point cloud in the 3D visual map.

And step S14, acquiring the pose information and the preset parameter information of each camera to be calibrated one by one according to the matching result, and calibrating each camera to be calibrated one by one according to the pose information and the preset parameter information.

After the key point matching is performed in step S14, the pose of the camera (i.e., the camera to be calibrated) corresponding to the captured feature picture (i.e., the second image) can be obtained, and then the parameters of the camera to be calibrated are adjusted, so that the calibration of the camera can be completed.

The above steps S13-S14 correspond to a camera self-calibration process, and the whole process is fully automatic. In addition, the position of the camera to be calibrated can be updated to the 3D visual map, and the field range of the camera is displayed in the 3D visual model.

Further, in the multi-camera automatic calibration method of the present application, after the visual map is constructed, the method may further include:

step S15, transmitting the obtained visual map in real time; and/or;

and acquiring information after the calibration of the camera to be calibrated is completed, synchronizing the information to the visual map, and then transmitting the synchronized visual map in real time.

It should be noted that the step S15 is not limited to which step it is necessarily followed; it may follow step S12, i.e. transmit the visual map obtained in real time; in addition, it may be that the visual map after the synchronization process is transmitted in real time immediately after step S14. Specifically, the visual map may be transmitted to a corresponding visualization device, such as a mobile phone, a computer, VR glasses, and the like.

According to the multi-camera automatic calibration method, cameras to be calibrated are supported to complete real-time dynamic calibration, calibration can be completed in real time corresponding to some cameras with movable holders or cameras (robots and the like) which move autonomously, and plane images of the cameras are directly projected onto a three-dimensional model. Therefore, in the using process, the camera to be calibrated can synchronize the calibrated information to the visual map so as to adapt to the change of the indoor environment.

In summary, according to the multi-camera automatic calibration method, the 3D model of the designated area is generated, and then the calibration of the camera to be calibrated is performed according to the existing model, so that the defects of high complexity, low universality and the like of the existing camera calibration method are overcome, and a semi-automatic multi-camera self-calibration mode which can be completed only by operating a single camera is realized.

In a second aspect, as shown in fig. 3, the present application discloses a multi-camera automatic calibration system, which may include an initial camera 21, a visual map building module 22, a feature point matching module 23, a calibration module 24, and so on.

An initial camera 21 for acquiring first image information of the designated area. The type, shooting mode, and the like of the initial camera 21 may refer to the statements about the scan acquisition device in the multi-camera automatic calibration method of the first aspect, and are not described herein again.

And the visual map building module 22 is used for building a visual map of the designated area according to the first image information.

Similarly, the initial camera is an RGBD camera and has a visual SLAM function, so that the first image information includes video frame data and an initial camera pose corresponding to each image frame; therefore, as shown in fig. 4, in this embodiment, the visual map building module 22 may further include a preliminary 3D model building unit 221, a 3D structure generating unit 222, and a visual map generating unit 223.

And the preliminary 3D model constructing unit 221 is configured to sequentially superimpose the 3D point clouds of all the image frames according to the acquisition timeline of the video frame data, and generate a preliminary 3D model of the designated area.

And the 3D structure generating unit 222 is configured to extract 2D feature points in each image frame according to the acquisition timeline of the video frame data, perform feature point matching between all the image frames, and generate a 3D structure of all the image frames.

A visual map generating unit 223 for mapping the 3D structure onto the preliminary 3D model, thereby generating a visual map.

The feature point matching module 23 is configured to obtain a plurality of pieces of second image information captured by a plurality of cameras to be calibrated, and perform feature point matching on the plurality of pieces of second image information one by one with the visual map.

And the calibration module 24 is configured to obtain the pose information and the predetermined parameter information of each camera to be calibrated one by one according to the matching result, and calibrate each camera to be calibrated one by one according to the pose information and the predetermined parameter information.

Further, the visual map building module 22 can also be configured to obtain information of the camera to be calibrated after calibration is completed, synchronize the information to the visual map, and finally transmit the synchronized visual map.

The multi-camera automatic calibration system further comprises a visualization unit 25, which is used for displaying the visual map in real time according to the visual map (including the visual map generated at the beginning and/or the visual map after subsequent synchronous processing) information transmitted by the visual map building module in real time. Likewise, the visualization unit 25 may be, for example, a mobile phone, a computer, VR glasses, etc.

In summary, the multi-camera automatic calibration system of the application performs calibration of the camera to be calibrated according to the existing model by generating the 3D model of the designated area, thereby overcoming the defects of high complexity, low universality and the like of the existing camera calibration method, and realizing a semi-automatic multi-camera self-calibration mode which can be completed only by operating a single camera.

Furthermore, after the calibration of a plurality of cameras is completed, a 3D multi-camera space model can be obtained, information of a plurality of cameras can be integrated, and better effects can be obtained on services such as target detection, multi-target tracking, pedestrian re-identification and the like.

To this end, in a third aspect, as shown in fig. 5, the present application discloses a multi-camera cooperative monitoring method, including the following steps:

step S31, calibrating each camera in the designated area by the multi-camera automatic calibration method according to any one of the first aspect.

And step S32, capturing a plurality of third image information of the designated area through the calibrated plurality of cameras.

And step S33, performing target detection processing and/or multi-target tracking processing and/or pedestrian re-identification processing according to the third image information.

And step S34, dynamically displaying the 3D form of the processing result in the visual map obtained in the multi-camera automatic calibration step.

In a fourth aspect, as shown in fig. 6, the present application discloses a multi-camera cooperative monitoring system, which includes a multi-camera calibration system 41, a plurality of cameras 42, a monitoring processing module 43, and a display module 44.

A multi-camera calibration system 41, which is used for calibrating each camera 42 in the designated area, and adopts any one of the above-mentioned third aspects;

the plurality of cameras 42 are installed at the designated area predetermined positions for capturing the plurality of third image information of the designated area after calibration is completed.

And a monitoring processing module 43, configured to perform target detection processing and/or multi-target tracking processing and/or pedestrian re-identification processing according to the plurality of third image information. The specific processing method for target detection, multi-target tracking and pedestrian re-identification may adopt a corresponding processing method in currently known intelligent video monitoring, and details are not repeated here.

And a display module 44, configured to dynamically display the processing result in a 3D form in a visual map, where the visual map is generated when each camera in the designated area is calibrated by a multi-camera calibration system (i.e., the visual map described in step S12 in the first aspect). In addition, the display module 44 in the present embodiment may be the visualization unit and the apparatus described in the first and second aspects above.

In conclusion, the multi-camera cooperative monitoring method and system directly calculate the geometric relationship between the cameras according to the 3D model and the camera pose, and can integrate the 2D image information of a plurality of cameras into 3D spatial information, thereby realizing a 3D target detection tracking algorithm. The system can better process tasks such as object shielding, multi-target tracking and the like.

Furthermore, the multi-camera cooperative monitoring method and system upgrade 2D monitoring into a 3D monitoring scheme, all camera information is truly in one space model, all monitored dynamic information (people, pets, mobile equipment and the like) and changes of some environments (table and chair movement and the like) can be dynamically displayed in the 3D room model in a 3D form, and therefore all-around intelligent monitoring of the whole indoor scene is achieved.

The multi-camera automatic calibration method and system, and the multi-camera cooperative monitoring method and system of the present application will be further described below with an embodiment.

Specifically, a plurality of calibrated cameras are installed in a family scene, so that functions of omnidirectional pet motion tracking, electronic nursing, danger early warning and the like are realized, and the method comprises the following steps:

firstly, shooting a video in a home by using a camera, and finishing 3D modeling of a home scene according to the prompt of app at a mobile phone end; a built-in computer vision algorithm reconstructs a 3D model according to the acquired video sequence, and a visualized result can be seen at the mobile phone end after reconstruction.

Then, the four cameras are installed at the planned positions, and calibration is realized at the mobile phone end according to the operation of the calibration method; at this moment, the used camera is a multi-camera system which can work together after calibration and calibration.

And finally, intelligent applications such as pet detection, motion tracking and the like are added, so that more efficient and accurate pet motion track generation and analysis can be realized. Meanwhile, the functions of electronic nursing and danger early warning are added according to the result, and unattended operation of the pet can be realized in all directions.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A multi-camera automatic calibration method is characterized by comprising the following steps:

acquiring first image information of a designated area;

constructing a visual map of the designated area according to the first image information, wherein the visual map is a 3D map;

acquiring the pose information and the preset parameter information of each camera to be calibrated one by one according to the matching result, and calibrating each camera to be calibrated one by one according to the pose information and the preset parameter information;

the first image information comprises video frame data and an initial camera pose corresponding to each image frame, wherein

2. The method for automatic calibration of multiple cameras according to claim 1, further comprising:

transmitting the obtained visual map in real time; and/or;

3. The method for automatically calibrating multiple cameras according to claim 1, wherein the step of performing feature point matching on the plurality of pieces of second image information one by one with the visual map comprises the following steps:

4. The utility model provides an automatic calibration system of many cameras which characterized in that includes:

the visual map construction module is used for constructing a visual map of the designated area according to the first image information, wherein the visual map is a 3D map;

5. The multi-camera automatic calibration system of claim 4, wherein the initial camera is an RGBD camera and has visual SLAM functionality, wherein

6. The multi-camera automatic calibration system of claim 5, wherein the visual map building module comprises:

7. The system according to claim 5 or 6, wherein the visual map building module is further configured to obtain information of the camera to be calibrated after calibration is completed, and synchronize the information to the visual map; wherein

The automatic calibration system of many cameras still includes:

8. A multi-camera cooperative monitoring method is characterized by comprising the following steps:

calibrating each camera within a specified area by a multi-camera automatic calibration method as claimed in any one of claims 1-3;

9. A multi-camera collaborative monitoring system, comprising:

a multi-camera calibration system, which employs the multi-camera calibration system of any one of claims 4-7, for calibrating each camera within a specified area;