The current system of panoramic cameras is mainly the following:
Camera splicing method: Uses multiple camera splicing to achieve a wide range of monitoring. The difficulty of this method lies in the image matching and image splicing of each camera's output image. In addition, synchronous triggering and synchronization output of each camera are also particularly important, and the requirements for software and hardware for real-time output of multi-image fusion at the later stage are also very high. The effect is higher in cost. It is generally suitable for specific applications where cost insensitivity and high resolution are required.
Single camera, through the rotation of the splicing method: This method is less real-time, and requires a rotating mechanism, high cost, only applicable to occasions where real-time requirements are not high.
Single-lens + single camera for panoramic imaging, which is currently widely used panoramic imaging. According to different imaging principles, a single lens is divided into a fish-eye panoramic lens and a reflection-refractive-type panoramic lens. In the era of analog cameras, the pixels constrained by the camera were not high. When single-lens and single-camera imaging were used to achieve panoramic imaging, the surveillance scene occupies very small pixels and the resolution is low, so the usability is not always strong. With the development of high-definition cameras, CCTV cameras can monitor up to 10 million pixels at this stage. The application scope and prospects of single-lens and single-camera imaging are very good. The following mainly introduces the principle and differences of these two lenses.
A, fisheye panoramic lens principle: different from the conventional optical system design, according to the principle of similar imaging, fisheye lens using large curvature negative lens artificially introduce a lot of barrel distortion to achieve wide-angle imaging, but also to avoid the edge due to The sharp increase in the illumination angle of the field angle problem.
Imaging features:
Can use a lens to achieve a field of view of 180° or more, and there is no blind spot in the field of view;
The radial direction (outward from the center of the imaging circle) contains the height information of the scene, and the tangential direction (the tangent of the concentric circle at the center of the imaging circle) contains the horizontal information of the object;
In the center of the fish-eye image, the scene in the smaller field of view occupies a large number of pixels. The outer-circle area of ​​the image has many scene scenes but it occupies less pixels. Therefore, one problem that the fish-eye image faces is how to increase the marginal vision area to occupy pixels. Improve the resolution of edge scenes; however, limited by the principle of imaging, improving the resolution of edge scenes will greatly increase the design difficulty and processing difficulty of fisheye lenses.
The fisheye lens design adopts the nonlinear similar imaging principle, and artificially introduces a large number of barrel-shaped distortions. Therefore, when the fisheye image is unfolded, only the fitting method can be used, and the correlation between the graphline expansion accuracy and the algorithm is relatively large. And it is difficult to completely correct the unfolded image distortion.
The design and installation of the fish-eye camera is convenient, and the panoramic camera can be realized by replacing the conventional lens with a suitable fish-eye lens using a conventional camera.
B. Reflection-Refraction Panoramic Lens Principle: Aspherical (spherical) mirrors (groups) are used to collect a wide range of field angles, and are then incident on a CCD/CMOS sensor after matching a regular relay lens. According to the mirror surface design, global (4Ï€) imaging can be realized theoretically.
Imaging features:
Can use the mirror (aspherical or spherical, one or a group) to achieve the horizontal 360°, pitching 90° above the circular field of view imaging; according to the imaging principle, there will be a blind zone in the direction of the camera installation, can improve the single-mirror The method of double mirrors and the reduction of the blind zone according to the specific structural design, but due to the presence of components that block the visual field on the structure, the blind zone cannot be completely eliminated;
The 360° field of view angle in the horizontal direction, the pitch field angle can be changed according to the requirements of the mirror surface design, in theory, can achieve -90 ° ~ 90 ° pitch angle of view;
Similar to the fisheye lens, the radial direction (outward from the center of the imaging circle) contains the height information of the scene, and the tangential direction (the tangent of the concentric circle in the center of the imaging circle) contains the horizontal information of the object;
Without increasing the design difficulty and cost, the design of the aspherical reflector (group) can greatly improve the pixel and resolution of the edge scene; it is very suitable for panoramic monitoring of the annular area;
The principle of imaging is similar to the single-viewpoint imaging of a camera. The mirror is generally designed with a quadratic surface. When performing panoramic expansion, the spatial coordinates of the imaged scene can be restored very accurately based on the curved surface shape and the geometric projection relationship, and the unfolded image has no distortion; Its high-precision reduction of the scene coordinates of the scene, can be linked to an external high-speed dome camera to observe the details; the use of reflection-refraction panoramic imaging camera was first applied to robot navigation and other machine vision field;
Lens adjustment requirements are high;
The structure of the camera is diversified;
Aspheric mirrors currently have high processing costs.
Camera splicing method: Uses multiple camera splicing to achieve a wide range of monitoring. The difficulty of this method lies in the image matching and image splicing of each camera's output image. In addition, synchronous triggering and synchronization output of each camera are also particularly important, and the requirements for software and hardware for real-time output of multi-image fusion at the later stage are also very high. The effect is higher in cost. It is generally suitable for specific applications where cost insensitivity and high resolution are required.
Single camera, through the rotation of the splicing method: This method is less real-time, and requires a rotating mechanism, high cost, only applicable to occasions where real-time requirements are not high.
Single-lens + single camera for panoramic imaging, which is currently widely used panoramic imaging. According to different imaging principles, a single lens is divided into a fish-eye panoramic lens and a reflection-refractive-type panoramic lens. In the era of analog cameras, the pixels constrained by the camera were not high. When single-lens and single-camera imaging were used to achieve panoramic imaging, the surveillance scene occupies very small pixels and the resolution is low, so the usability is not always strong. With the development of high-definition cameras, CCTV cameras can monitor up to 10 million pixels at this stage. The application scope and prospects of single-lens and single-camera imaging are very good. The following mainly introduces the principle and differences of these two lenses.
A, fisheye panoramic lens principle: different from the conventional optical system design, according to the principle of similar imaging, fisheye lens using large curvature negative lens artificially introduce a lot of barrel distortion to achieve wide-angle imaging, but also to avoid the edge due to The sharp increase in the illumination angle of the field angle problem.
Imaging features:
Can use a lens to achieve a field of view of 180° or more, and there is no blind spot in the field of view;
The radial direction (outward from the center of the imaging circle) contains the height information of the scene, and the tangential direction (the tangent of the concentric circle at the center of the imaging circle) contains the horizontal information of the object;
In the center of the fish-eye image, the scene in the smaller field of view occupies a large number of pixels. The outer-circle area of ​​the image has many scene scenes but it occupies less pixels. Therefore, one problem that the fish-eye image faces is how to increase the marginal vision area to occupy pixels. Improve the resolution of edge scenes; however, limited by the principle of imaging, improving the resolution of edge scenes will greatly increase the design difficulty and processing difficulty of fisheye lenses.
The fisheye lens design adopts the nonlinear similar imaging principle, and artificially introduces a large number of barrel-shaped distortions. Therefore, when the fisheye image is unfolded, only the fitting method can be used, and the correlation between the graphline expansion accuracy and the algorithm is relatively large. And it is difficult to completely correct the unfolded image distortion.
The design and installation of the fish-eye camera is convenient, and the panoramic camera can be realized by replacing the conventional lens with a suitable fish-eye lens using a conventional camera.
B. Reflection-Refraction Panoramic Lens Principle: Aspherical (spherical) mirrors (groups) are used to collect a wide range of field angles, and are then incident on a CCD/CMOS sensor after matching a regular relay lens. According to the mirror surface design, global (4Ï€) imaging can be realized theoretically.
Imaging features:
Can use the mirror (aspherical or spherical, one or a group) to achieve the horizontal 360°, pitching 90° above the circular field of view imaging; according to the imaging principle, there will be a blind zone in the direction of the camera installation, can improve the single-mirror The method of double mirrors and the reduction of the blind zone according to the specific structural design, but due to the presence of components that block the visual field on the structure, the blind zone cannot be completely eliminated;
The 360° field of view angle in the horizontal direction, the pitch field angle can be changed according to the requirements of the mirror surface design, in theory, can achieve -90 ° ~ 90 ° pitch angle of view;
Similar to the fisheye lens, the radial direction (outward from the center of the imaging circle) contains the height information of the scene, and the tangential direction (the tangent of the concentric circle in the center of the imaging circle) contains the horizontal information of the object;
Without increasing the design difficulty and cost, the design of the aspherical reflector (group) can greatly improve the pixel and resolution of the edge scene; it is very suitable for panoramic monitoring of the annular area;
The principle of imaging is similar to the single-viewpoint imaging of a camera. The mirror is generally designed with a quadratic surface. When performing panoramic expansion, the spatial coordinates of the imaged scene can be restored very accurately based on the curved surface shape and the geometric projection relationship, and the unfolded image has no distortion; Its high-precision reduction of the scene coordinates of the scene, can be linked to an external high-speed dome camera to observe the details; the use of reflection-refraction panoramic imaging camera was first applied to robot navigation and other machine vision field;
Lens adjustment requirements are high;
The structure of the camera is diversified;
Aspheric mirrors currently have high processing costs.
SICHUAN TELOS NEW MATERIAL TECHNOLOGY CO., LTD , https://www.kenlarcutters.com