CN209231985U - A kind of camera parameters automatic calibration device based on virtual Binocular Vision Principle - Google Patents

Abstract

The utility model proposes an automatic camera parameter calibration device based on the principle of virtual binocular vision. The device includes a pan-tilt rotating bracket, a camera, an external trigger circuit, a PC, a calibration board with a black and white checkerboard pattern, and an angle setting on both sides. The two plane mirrors are set behind the swivel bracket of the pan/tilt; the calibration plate is placed at the swivel bracket of the pan/tilt; the camera is set in front of the swivel bracket of the pan/tilt and the camera faces the two plane mirrors; The board image includes two left and right virtual images formed by the calibration board in the two plane mirrors; when the calibration board rotates an angle, the external trigger circuit controls the camera to take pictures of the calibration board; the PC processes the calibration board image captured by the camera to obtain and calibrate The camera's internal parameters, distortion parameters and external parameters in the world coordinate system; this product solves the traditional tedious process of manually finding the corner points of the calibration board, simplifies the experimental process, and is easy to operate.

Description

An automatic camera parameter calibration device based on the principle of virtual binocular vision

technical field

The utility model relates to the technical field of image processing, in particular to an automatic camera parameter calibration device based on the principle of virtual binocular vision.

Background technique

The purpose of camera parameter calibration is to determine the camera's position, attribute parameters and establish an imaging geometric model in order to determine the correspondence between the object point in the space coordinate system and its image point on the image plane. The accuracy of camera calibration results directly affects the accuracy of 3D reconstruction, so camera calibration plays a very important role in the process of 3D testing.

Zhang Zhengyou's method is a calibration method widely used in the field of computer vision. It uses black and white checkerboards as calibration objects and places them arbitrarily in front of the camera. The camera is determined by processing multiple calibration board images collected from different angles. It has the advantages of high calibration accuracy, but in the actual operation process, there are problems such as manual rotation of the calibration plate, more manual intervention in the image processing process, heavy workload, and inconvenient processing and analysis of related parameters. Based on this, Xu Hui et al. proposed the patent "A Quick Calibration Device and Calibration Method for Fisheye Camera", which automatically collects calibration plate images from different viewing angles by controlling the horizontal rotation and vertical rotation of the fisheye camera, but does not involve automatic processing of calibration plate images and Automatic calculation process of camera parameters. Wang Pingping proposed a patent "a 3D input method based on OpenCV camera calibration". By shooting calibration boards at different positions, OpenCV is used to automatically process the image, but only the internal parameters of the camera are calculated, which is suitable for single camera in monocular measurement. calibration.

The automatic calibration method of camera parameters based on the principle of virtual binocular vision proposed by the utility model is based on the principle of binocular stereo vision and the calibration method of Zhang Zhengyou. Only one camera and two plane mirrors can be used to realize the three-dimensional reconstruction of binocular vision. It solves the problems of poor synchronization and high cost in traditional binocular vision technology. At the same time, the external trigger circuit is used to control the automatic rotation of the calibration plate, and the camera is controlled to take time-lapse shots, which reduces the uncertainty caused by manual intervention. In terms of image processing, the utility model utilizes the correlation function processing in the OpenCV computer vision library to obtain the internal parameters of the camera, the distortion parameters and the external parameters corresponding to the world coordinate system, which solves the traditional cumbersome process of manually finding the corner points of the calibration board. It simplifies the experimental process, is easy to operate, and lays the foundation for the three-dimensional reconstruction of the measured object.

Contents of the invention

The utility model proposes an automatic camera parameter calibration device based on the principle of virtual binocular vision, which solves the traditional cumbersome process of manually finding the corner points of the calibration board, simplifies the experimental process, and is easy to operate.

The utility model adopts the following technical solutions.

An automatic calibration device for camera parameters based on the principle of virtual binocular vision. The calibration method uses an automatic calibration device to calibrate the camera parameters. The calibration device includes a pan-tilt rotating bracket, a camera, an external trigger circuit, a PC, A calibration plate with a checkerboard pattern and two plane mirrors arranged at an angle; the external trigger circuit is connected to the PC, the pan-tilt rotating bracket and the camera; the two plane mirrors are arranged behind the pan-tilt rotating bracket; The camera is set at the front of the swivel bracket of the pan-tilt and the camera faces the two plane mirrors; when the camera takes pictures of the calibration board, the image of the calibration board it captures includes the left and right virtual images formed by the calibration board in the two plane mirrors; The external trigger circuit controls the pan-tilt rotation bracket to rotate the calibration board at multiple angles; when the calibration board rotates an angle, the external trigger circuit controls the camera to take pictures of the calibration board; the PC is connected to the camera and uses the OpenCV computer vision library to shoot the camera. The image of the calibration board is processed to obtain and calibrate the internal parameters of the camera, the distortion parameters and the external parameters in the world coordinate system.

Both the calibration board and the camera are arranged on the bisector of the angle between the two plane mirrors.

The number of squares on the calibration board is above a threshold.

The calibration method includes an image acquisition link and an image processing link; the image acquisition link includes the following steps;

A1. First place the plane mirror with a certain angle on both sides behind the calibration board, the camera is placed in front of the calibration board, and the calibration board is fixed on the pan/tilt rotating bracket. The midline positions of the camera, calibration board and plane mirror should ensure the images of the left and right mirrors It can be clear and centered as much as possible, so that the calibration plate can still be completely imaged in the mirror when it is rotated;

A2. When shooting, first place the calibration board vertically, take an image as the world coordinate system, and then control the external trigger circuit through the PC to automatically rotate the pan/tilt rotation bracket to different angles, and trigger the camera through the external trigger circuit to align the calibration board The virtual image in the plane mirror is photographed;

A3. Every time the gimbal rotates an angle, the camera shoots an image of the calibration board. During the shooting process of the camera, the mouse cursor on the PC side will display the position of the current coordinate point in real time. By defining a coordinate range, record the rotation process of the calibration board The coordinate range of the calibration plate changes, and all the captured images are uniformly stored in the image processing folder.

The image processing link regards the left and right virtual images of the calibration board in the captured plane mirrors as the calibration board images taken by the left and right virtual cameras in the two plane mirrors; process and process the multiple calibration board images obtained by rotating at different angles Calculate to obtain the internal parameters and distortion parameters of the camera; process the calibration plate image in the world coordinate system to obtain the external parameters of the camera and the relative pose parameters of the left virtual camera and the right virtual camera;

In the image processing link, in order to avoid the mutual influence of the left and right images when detecting the corner points of the black and white checkerboard pattern of the calibration board, the left and right mirror images are processed separately;

First, process the left mirror image, read in multiple images of the calibration board in sequence, and set the number and size of the corner points of the calibration board; according to the recorded coordinate range changes during the rotation of the calibration board, use the Rect command representing the rectangular area to specify The coordinates of the upper left corner of the rectangle and the length and width of the rectangle, that is, use the recorded changes in the coordinate range of the calibration plate rotation to define a rectangular area to intercept the region of interest, so that the calibration plate on the left mirror in this area is in the center and can be Accommodate the rotation range of the calibration plate;

When obtaining the internal parameters of the camera and the distortion parameters, first perform image grayscale processing, and then use the findChessboardCorners function in the OpenCV vision library to initially detect the corner point information on the calibration board, and automatically retrieve the pixel coordinates of the inner corner points of the calibration board. From left to right, from top to bottom, if there are some pictures that cannot accurately find the corner points, you need to reset the scope of the area of interest or delete these pictures; in order to make the extracted corner point coordinates To be more precise, use the find4QuadCornerSubpix function to accurately extract sub-pixel corner points and record the extracted corner point coordinate information. In this process, more than ten images of the calibration plate should be rotated to ensure accuracy. If the number of calibrated images is less than the specified If the specified number is required, the region of interest needs to be redefined; take the first corner point in the upper left corner of the calibration board as the origin, and obtain the three-dimensional space point coordinates of the inner corner point, that is, the point coordinates in the camera coordinate system, so that the camera coordinate system The xy plane is located at the plane position of the calibration board, that is, the z-axis of the camera coordinate system is zero, and the coordinates of the three-dimensional space points are initialized one by one according to the size of each square of the calibration board in order from left to right and from top to bottom; in order to increase The accuracy of the calibration results requires taking more than ten images of the calibration board to solve multiple equations; continuous calibration processes multiple rotating images of the calibration board to obtain internal parameters with higher precision; use the calibrateCamera function to calculate, you can get Internal parameters and distortion parameters of the left virtual camera;

When obtaining the external parameters of the left virtual camera in the world coordinate system, read in the obtained internal parameters of the left virtual camera, distortion parameters and the captured single image in the world coordinate system, intercept the left region of interest, and use The cvFindExtrinsicCameraParams2 function processes the image in the world coordinate system according to the above calibration method, and obtains the external parameters of the left virtual camera in the world coordinate system;

The image processing process of the right mirror is the same as that of the left mirror, and the internal parameters of the right virtual camera and the external parameters in the world coordinate system are obtained;

After the calibration is completed, the calculated corner image coordinates are re-projected to the calibration board, and compared with the actual known image coordinates, the calibration error of each image and the overall average error are calculated; On the basis of the external parameters of the virtual camera, use the StereoCalibrate function to calculate the relative rotation matrix and translation vector of the left and right virtual cameras, so as to obtain the relative position relationship of the two virtual cameras.

When it comes to the step of intercepting the left and right mirror images for separate processing, since the coordinate position will change when intercepting the region of interest, the length and width values of the intercepted region of interest should be superimposed when extracting the coordinate values of the corner points.

The calibration plate rotates under the drive of the pan-tilt rotating bracket, and the pan-tilt rotating bracket can be controlled by a PC or a remote controller.

The external trigger circuit can control the camera to perform delayed shooting, and the delayed time can be customized.

The image processing method based on OpenCV proposed by the utility model saves the process of manually looking for corner points and calculation, and only needs to use VC++ as the programming development environment to realize image processing through programming, which is conducive to the further development of the later three-dimensional reconstruction work .

The utility model can realize the three-dimensional reconstruction of binocular vision only by using one camera and two plane mirrors, which solves the problems of poor synchronization and high cost in the traditional binocular vision technology; at the same time, the utility model uses an external trigger circuit to control the calibration board. Automatically rotate and control the time-lapse shooting of the camera, reducing the uncertainty caused by manual intervention.

Description of drawings

Below in conjunction with accompanying drawing and specific embodiment the utility model is described in further detail:

Accompanying drawing 1 is the schematic flow sheet of the utility model;

Accompanying drawing 2 is the schematic diagram of the used device of the utility model;

Accompanying drawing 3 is the schematic flow chart of the utility model in image processing;

Accompanying drawing 4 is the checkerboard pattern schematic diagram of calibration plate;

In the figure: 1-PC; 2-external trigger circuit; 3-camera; 4-pan/tilt rotating bracket; 5-calibration board; 6-left plane mirror; 7-right plane mirror; .

Detailed ways

As shown in Figures 1-4, a camera parameter automatic calibration device based on the principle of virtual binocular vision, the calibration method uses an automatic calibration device to calibrate the camera parameters, and the calibration device includes a pan/tilt rotating bracket 4, a camera 3 , external trigger circuit 2, PC1, the calibration plate 5 that is painted with black and white checkerboard pattern and the plane mirror 6 and 7 that two sides form an included angle and arrange; Described external trigger circuit links to each other with PC, pan tilt rotating support and video camera; Described two plane mirrors Be located at the rear of the pan-tilt rotating bracket; Described calibration plate is placed at the pan-tilt rotating bracket place; Video camera is located at the front of the pan-tilt rotating bracket and photographing aspect towards two plane mirrors; The left and right virtual images of the calibration board formed in the two plane mirrors; the PC controls the pan-tilt rotation bracket through the external trigger circuit to make the calibration board rotate at multiple angles; when the calibration board rotates an angle, the external trigger circuit controls the camera to take pictures of the calibration board; The above PC is connected to the camera and the calibration board image taken by the camera is processed with the OpenCV computer vision library to obtain and calibrate the camera's internal parameters, distortion parameters and external parameters in the world coordinate system.

Both the calibration board and the camera are arranged on the bisector of the angle between the two plane mirrors.

The number of squares on the calibration board is above a threshold.

The calibration method includes an image acquisition link and an image processing link; the image acquisition link includes the following steps;

A1. First place the plane mirror with a certain angle on both sides behind the calibration board, the camera is placed in front of the calibration board, and the calibration board is fixed on the pan/tilt rotating bracket. The midline positions of the camera, calibration board and plane mirror should ensure the images of the left and right mirrors It can be clear and centered as much as possible, so that the calibration plate can still be completely imaged in the mirror when it is rotated;

A2. When shooting, first place the calibration board vertically, take an image as the world coordinate system, and then control the external trigger circuit through the PC to automatically rotate the pan/tilt rotation bracket to different angles, and trigger the camera through the external trigger circuit to align the calibration board The virtual image in the plane mirror is photographed;

A3. Every time the gimbal rotates an angle, the camera shoots an image of the calibration board. During the shooting process of the camera, the mouse cursor on the PC side will display the position of the current coordinate point in real time. By defining a coordinate range, record the rotation process of the calibration board The coordinate range of the calibration plate changes, and all the captured images are uniformly stored in the image processing folder.

The image processing link regards the left and right virtual images of the calibration board in the captured plane mirrors as the calibration board images taken by the left and right virtual cameras in the two plane mirrors; process and process the multiple calibration board images obtained by rotating at different angles Calculate to obtain the internal parameters and distortion parameters of the camera; by processing the calibration plate image under the world coordinate system, obtain the external parameters of the camera and the relative pose parameters of the left virtual camera 8 and the right virtual camera 9;

In the image processing link, in order to avoid the mutual influence of the left and right images when detecting the corner points of the black and white checkerboard pattern of the calibration board, the left and right mirror images are processed separately;

First, process the left mirror image, read in multiple images of the calibration board in sequence, and set the number and size of the corner points of the calibration board; according to the recorded coordinate range changes during the rotation of the calibration board, use the Rect command representing the rectangular area to specify The coordinates of the upper left corner of the rectangle and the length and width of the rectangle, that is, use the recorded changes in the coordinate range of the calibration plate rotation to define a rectangular area to intercept the region of interest, so that the calibration plate on the left mirror in this area is in the center and can be Accommodate the rotation range of the calibration plate;

When obtaining the internal parameters of the camera and the distortion parameters, first perform image grayscale processing, and then use the findChessboardCorners function in the OpenCV vision library to initially detect the corner point information on the calibration board, and automatically retrieve the pixel coordinates of the inner corner points of the calibration board. From left to right, from top to bottom, if there are some pictures that cannot accurately find the corner points, you need to reset the scope of the area of interest or delete these pictures; in order to make the extracted corner point coordinates To be more precise, use the find4QuadCornerSubpix function to accurately extract sub-pixel corner points and record the extracted corner point coordinate information. In this process, more than ten images of the calibration plate should be rotated to ensure accuracy. If the number of calibrated images is less than the specified If the specified number is required, the region of interest needs to be redefined; take the first corner point in the upper left corner of the calibration board as the origin, and obtain the three-dimensional space point coordinates of the inner corner point, that is, the point coordinates in the camera coordinate system, so that the camera coordinate system The xy plane is located at the plane position of the calibration board, that is, the z-axis of the camera coordinate system is zero, and the coordinates of the three-dimensional space points are initialized one by one according to the size of each square of the calibration board in order from left to right and from top to bottom; in order to increase The accuracy of the calibration results requires shooting more than ten images of the calibration board to solve multiple equations; continuous calibration processes multiple images of the calibration board to obtain internal parameters with higher precision; use the calibrateCamera function to calculate, you can get Internal parameters and distortion parameters of the left virtual camera;

When obtaining the external parameters of the left virtual camera in the world coordinate system, read in the obtained internal parameters of the left virtual camera, distortion parameters and the captured single image in the world coordinate system, intercept the left region of interest, and use The cvFindExtrinsicCameraParams2 function processes the image in the world coordinate system according to the above calibration method, and obtains the external parameters of the left virtual camera in the world coordinate system;

The image processing process of the right mirror is the same as that of the left mirror, and the internal parameters of the right virtual camera and the external parameters in the world coordinate system are obtained;

After the calibration is completed, the calculated corner image coordinates are re-projected to the calibration board, and compared with the actual known image coordinates, the calibration error of each image and the overall average error are calculated; On the basis of the external parameters of the virtual camera, use the StereoCalibrate function to calculate the relative rotation matrix and translation vector of the left and right virtual cameras, so as to obtain the relative position relationship of the two virtual cameras.

When it comes to the step of intercepting the left and right mirror images for separate processing, since the coordinate position will change when intercepting the region of interest, the length and width values of the intercepted region of interest should be superimposed when extracting the coordinate values of the corner points.

The calibration plate rotates under the drive of the pan-tilt rotating bracket, and the pan-tilt rotating bracket can be controlled by a PC or a remote controller.

The external trigger circuit can control the camera to perform delayed shooting, and the delayed time can be customized.

This calibration method is also applicable to single-target calibration of video cameras or ordinary cameras and calibration of two cameras or ordinary cameras based on the principle of binocular vision.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present utility model and not to limit it; although the utility model has been described in detail with reference to the preferred embodiment, those of ordinary skill in the art should understand that: still The specific implementation of the utility model can be modified or some technical features can be equivalently replaced; without departing from the spirit of the technical solution of the utility model, all of them should be included in the scope of the technical solution claimed by the utility model.

Claims (4)

1. A camera parameter automatic calibration device based on the principle of virtual binocular vision, characterized in that: the calibration device adopts an automatic calibration device to calibrate the camera parameters, and the calibration device includes a pan-tilt rotating bracket, a video camera, an external trigger circuit , a calibration plate with a black and white checkerboard pattern and a plane mirror arranged at an included angle on both sides; the external trigger circuit is connected with the pan-tilt rotation bracket and the camera; the two plane mirrors are arranged behind the pan-tilt rotation bracket; the calibration plate is placed At the swivel bracket of the pan-tilt; the camera is set in front of the swivel bracket of the pan-tilt and the camera is facing the two plane mirrors; when the camera takes a picture of the calibration plate, the image of the calibration plate it captures includes the left and right virtual images formed by the calibration plate in the two plane mirrors; The swivel bracket of the pan/tilt rotates the calibration board at multiple angles; when the calibration board rotates an angle, the external trigger circuit controls the camera to take pictures of the calibration board; The calibration plate rotates under the drive of the pan-tilt rotating bracket, and the pan-tilt rotating bracket can be controlled by a remote controller. 2. A camera parameter automatic calibration device based on the principle of virtual binocular vision according to claim 1, characterized in that: the calibration board and the camera are all arranged on the angle bisector of the angle between the two plane mirrors. 3. A device for automatically calibrating camera parameters based on the principle of virtual binocular vision according to claim 1, characterized in that: the number of squares on the calibration board is above a threshold. 4. A device for automatically calibrating camera parameters based on the principle of virtual binocular vision according to claim 1, characterized in that: the external trigger circuit can control the camera to perform delayed shooting and the delayed time can be customized.