WO2019109226A1 - Binocular camera calibration method and device - Google Patents

Abstract

The present invention provides a binocular camera calibration method and device. The actual distance D between a laser distance sensor and a specified target is measured; a conversion matrix R of rotation movement of a binocular camera and a scaling value t of the distance between two cameras are calculated; a depth value d between the laser distance sensor and an infrared light spot is calculated; the actual distance T between the two cameras is calculated according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras.

Description

Binocular camera calibration method and device Technical field

The invention relates to the technical field of camera calibration, in particular to a binocular camera calibration method and device.

Background technique

The binocular camera is gradually used in various sweeping robots due to its stable and powerful slam (instant positioning and map construction).

At present, binocular cameras generally require human calibration to obtain external parameters before use. In order to save manpower, the default calibration parameters are generally used before leaving the factory, but because of its complicated structure, it is difficult to install. If there is a large offset between the two camera installation positions, the default calibration parameters cannot be applied to the actual use, which is easy to affect the subsequent positioning. With navigation.

Summary of invention

technical problem

The main object of the present invention is to provide a binocular camera calibration method and apparatus for calibrating a binocular camera.

Problem solution

Technical solution

The invention provides a binocular camera calibration method, which comprises the following steps:

The terminal measures the actual distance D between the laser ranging sensor and the specified target through a laser ranging sensor located at the same depth plane as the binocular camera;

The image information of the specified target is respectively acquired by the binocular camera, and the conversion matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras are calculated according to the collected image information of the specified target;

Acquiring the infrared spot image information of the laser ranging sensor on the specified target by the binocular camera, and calculating the depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information;

The actual distance T between the two cameras is calculated from the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras.

Further, the step of calculating a conversion matrix R of the rotational movement of the binocular camera and a scaling value t of the distance between the two cameras according to the collected image information of the specified target includes:

Performing feature extraction on the collected image information of the specified target, and extracting matching feature points in the images acquired by the two cameras;

According to the matched feature points, the conversion matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras are calculated.

Further, extracting the number of matching feature points in the images acquired by the two cameras is eight, and according to the matched feature points, calculating a conversion matrix R of the rotation movement of the binocular camera and a scaling value of the distance between the two cameras The steps include:

According to the matched eight feature points, the normalized eight-point method is used to calculate the transformation matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras.

Further, after the step of separately acquiring the image information of the specified target by the binocular camera, the method further includes:

Perform image denoising on the acquired image information of the specified target.

Further, the step of calculating the depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information includes:

According to the matching relationship of the infrared spot in the image acquired by the two cameras, the depth value d between the laser ranging sensor and the infrared spot is calculated by a trigonometry method.

Further, the step of calculating the actual distance T between the two cameras according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras includes:

Calculating a scale s according to the actual distance D and the depth value d;

The actual distance T between the two cameras is calculated based on the scale s and the scaling value t of the distance between the two cameras.

[Correction according to Rule 26 23.02.2018]
Further, the calculation formula of the scale s is s=D/d; the calculation formula of the actual distance T is T=s×t.

Further, the laser ranging sensor is located at a midpoint of two camera connections in the binocular camera; the terminal is a binocular vision cleaning robot.

The invention also provides a binocular camera calibration device, comprising:

a measuring unit, configured to measure an actual distance D between the laser ranging sensor and the specified target by a laser ranging sensor located at the same depth plane as the binocular camera;

The first calculating unit is configured to separately collect image information of the specified target through the binocular camera, and calculate a conversion matrix R of the rotation movement of the binocular camera and a scaling value of the distance between the two cameras according to the collected image information of the specified target t;

a second calculating unit, configured to separately collect, by the binocular camera, the infrared spot image information irradiated by the laser ranging sensor on the specified target, and calculate the distance between the laser ranging sensor and the infrared spot according to the matching relationship in the image information Depth value d;

And a third calculating unit, configured to calculate an actual distance T between the two cameras according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras.

Further, the first calculating unit includes:

Extracting a sub-unit for performing feature extraction on the collected image information of the specified target, and extracting matching feature points in the images acquired by the two cameras;

The first calculating subunit is configured to calculate a conversion matrix R of the rotation movement of the binocular camera and a scaling value t of the distance between the two cameras according to the matched feature points.

Further, the first calculating subunit is specifically configured to:

According to the matched feature points, the transformation matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras are calculated by the normalized eight-point method.

Further, it also includes:

The denoising unit is configured to perform denoising processing on the collected image information of the specified target.

Further, the second calculating unit is specifically configured to:

According to the matching relationship of the infrared spot in the image acquired by the two cameras, the depth value d between the laser ranging sensor and the infrared spot is calculated by a trigonometry method.

Further, the third calculating unit includes:

a second calculating subunit, configured to calculate a scale s according to the actual distance D and the depth value d;

And a third calculating subunit, configured to calculate an actual distance T between the two cameras according to the scale s and a scaling value t of the distance between the two cameras.

[Correction according to Rule 26 23.02.2018]
Further, the calculation formula of the scale s is s=D/d; the calculation formula of the actual distance T is T=s×t.

Further, the laser ranging sensor is located at a midpoint of two camera connections in the binocular camera; the terminal is a binocular vision cleaning robot.

Advantageous effects of the invention

Beneficial effect

The binocular camera calibration method and device provided by the invention have the following beneficial effects:

The binocular camera calibration method and device provided by the invention, the terminal measures the actual distance D between the laser ranging sensor and the specified target through a laser ranging sensor located at the same depth plane as the binocular camera; and respectively collects the specified target through the binocular camera Image information, and according to the acquired image information of the specified target, calculate a conversion matrix R of the rotation movement of the binocular camera and a scaling value t of the distance between the two cameras; respectively, the laser ranging sensor is irradiated by the binocular camera Specifying infrared spot image information on the target, and calculating a depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information; according to the actual distance D, the depth value d, and between the two cameras The zoom value t of the distance calculates the actual distance T between the two cameras; this is used to calibrate the binocular camera.

Brief description of the drawing

DRAWINGS

1 is a schematic diagram showing the steps of a binocular camera calibration method according to an embodiment of the present invention;

2 is a schematic diagram of specific steps of step S2 in an embodiment of the present invention;

3 is a schematic diagram of specific steps of step S4 in an embodiment of the present invention;

4 is a block diagram showing the structure of a binocular camera calibration apparatus according to an embodiment of the present invention;

Figure 5 is a block diagram showing the structure of a first computing unit in an embodiment of the present invention;

6 is a block diagram showing the structure of a third computing unit in an embodiment of the present invention.

The implementation, functional features, and advantages of the present invention will be further described in conjunction with the embodiments.

BEST MODE FOR CARRYING OUT THE INVENTION

BEST MODE FOR CARRYING OUT THE INVENTION

It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The singular forms "a", "the", "the" It is to be understood that the phrase "comprise" or "an" or "an" Other features, integers, steps, operations, elements, units, modules, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element. Further, "connected" or "coupled" as used herein may include either a wireless connection or a wireless coupling. The phrase "and/or" used herein includes all or any one and all combinations of one or more of the associated listed.

Those skilled in the art will appreciate that all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs, unless otherwise defined. It should also be understood that terms such as those defined in a general dictionary should be understood to have meaning consistent with the meaning in the context of the prior art, and will not be idealized or overly formal unless specifically defined. To explain.

1 is a schematic diagram showing the steps of a binocular camera calibration method according to an embodiment of the present invention.

In an embodiment of the invention, a binocular camera calibration method is provided, including the following steps:

Step S1, the terminal measures the actual distance D between the laser ranging sensor and the specified target through a laser ranging sensor located in the same depth plane as the binocular camera;

Step S2, separately acquiring image information of the specified target through the binocular camera, and calculating a conversion matrix R of the rotation movement of the binocular camera and the distance between the two cameras according to the collected image information of the specified target. Offset scaling value t;

Step S3, respectively, acquiring, by the binocular camera, the infrared spot image information irradiated by the laser ranging sensor on the specified target, and calculating the depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information. ;

In step S4, the actual distance T between the two cameras is calculated according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras.

In this embodiment, the binocular camera can be applied to a terminal device that requires imaging, such as a mobile phone, a cleaning robot, and the like. The terminal in this embodiment is described by taking a binocular vision cleaning robot as an example.

The two cameras included in the binocular camera have different external parameters depending on the installed terminal equipment. Therefore, when the binocular camera is mounted on the terminal (for example, a binocular vision cleaning robot), it needs to be calibrated. The calibration is to calculate the external parameters of the binocular camera, and the external parameters include the actual between the binocular cameras. The distance T, and the transformation matrix R of the binocular camera rotating movement.

In this embodiment, before the calibration, the laser ranging sensor and the installed binocular camera are disposed in the same depth plane (the same vertical plane), and preferably, the laser ranging sensor is located in the binocular camera. The midpoint of the connection between the two cameras. The laser ranging sensor can measure the actual distance D from a specified target, which refers to a texture-rich target, such as a mural on a wall, a photo frame, etc., and a texture-rich designated target facilitates identification of feature points. The laser ranging sensor emits infrared light to form an infrared spot on the specified target, and the two cameras respectively acquire image information of the infrared spot on the specified target, and according to the image information, the infrared spot is in the two frames. The matching relationship calculates a depth value d between the laser ranging sensor and the infrared spot, which is a scale value representing how many scales there are between the infrared spot and the laser ranging sensor. It can be understood that, according to the actual distance D and the depth value d described above, the scale s, that is, the distance value of each scale can be calculated.

Further, the image information of the specified target is separately acquired by the two cameras, and the conversion matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras are calculated according to the acquired image information of the specified target, and the scaling The value t is not an actual distance value, which represents how many scales there are between the two cameras. Since the above measurements are all measured in the same state (ie, keeping the laser ranging sensor and the position of the double-sided camera unchanged), the scale s is the same, and the difference between the two cameras can be calculated according to the above-mentioned scale s and the scale value t. Actual distance T.

In summary, the conversion matrix R of the rotational movement of the double-sided camera and the actual distance T between the two cameras are calculated, and the external reference calibration of the double-sided camera is completed. Accurate access to external parameters is simple and convenient, with less computational complexity and no human calibration.

Referring to FIG. 2, in an embodiment, the step S2 of calculating the conversion matrix R of the rotation of the binocular camera and the scaling value t of the distance between the two cameras according to the acquired image information of the specified target includes:

Step S21: performing feature extraction on the collected image information of the specified target, and extracting matching feature points in the images acquired by the two cameras;

Step S22, calculating a conversion matrix R of the rotation movement of the binocular camera and a scaling value t of the distance between the two cameras according to the matched feature points.

In this embodiment, two cameras perform image acquisition on a specified target, perform feature extraction on the image information, extract the same feature points in the two frames of images for matching, and then use normalized eight points according to the matched eight feature points. The method calculates a conversion matrix R of the rotational movement of the binocular camera and a scaling value t of the distance between the two cameras. The scaling value t is not an actual distance value, which represents how many scales there are between the two cameras.

In this embodiment, after the step of separately acquiring the image information of the specified target by the binocular camera, the method further includes:

Perform image denoising on the acquired image information of the specified target. Specifically, denoising processing may be performed using Gaussian filtering, median filtering, or morphological operations or the like.

In another embodiment, the step S2 of calculating the depth value d between the laser ranging sensor and the infrared light spot according to the matching relationship in the image information includes:

According to the matching relationship of the infrared spot in the image acquired by the two cameras, the depth value d between the laser ranging sensor and the infrared spot is calculated by a trigonometry method.

In this embodiment, two cameras respectively acquire an image of one frame of infrared light spots, perform matching according to the same feature points of the infrared light spots in the image, and calculate a distance between the laser distance measuring sensor and the infrared light spot by using a trigonometry method. The depth value d; the depth value d is a scale value representing how many scales there are between the infrared spot and the laser ranging sensor.

Referring to FIG. 3, in the above embodiment, the image is based on the actual distance D, the depth value d, and two images. The step S4 of calculating the actual distance T between the two cameras by the scaling value t of the distance between the heads includes:

Step S41, calculating a scale s according to the actual distance D and the depth value d;

Step S42, calculating the actual distance T between the two cameras according to the scale s and the scaling value t of the distance between the two cameras.

In this embodiment, the actual distance between the specified target and the laser ranging sensor is D, and the depth value between the laser ranging sensor and the infrared spot is d, so that the scale s in the current measurement state can be calculated. . The scale s is the scale value of the distance between the two cameras, so the actual distance T between the two cameras can be calculated according to the scale s and the zoom value t of the distance between the two cameras.

[Correction according to Rule 26 23.02.2018]
Specifically, the calculation formula of the scale s is s=D/d; the calculation formula of the actual distance T is T=s×t.

In summary, the binocular camera calibration method provided in the embodiment of the present invention accurately calculates the actual distance T between the two cameras and the conversion matrix R of the rotation movement of the binocular camera, thereby completing the dual-sided camera. The external reference is calibrated, and it is simple and convenient, and the amount of calculation is small, and no human calibration is required.

Referring to FIG. 4, an embodiment of the present invention further provides a binocular camera calibration apparatus, including:

The measuring unit 10 is configured to measure the actual distance D between the laser ranging sensor and the specified target by a laser ranging sensor located at the same depth plane as the binocular camera;

The first calculating unit 20 is configured to separately collect image information of the specified target through the binocular camera, and calculate a conversion matrix R of the rotation movement of the binocular camera and a zoom of the distance between the two cameras according to the collected image information of the specified target. Value t;

The second calculating unit 30 is configured to separately collect infrared spot image information that is irradiated on the specified target by the laser ranging sensor through the binocular camera, and calculate the laser ranging sensor and the infrared spot according to the matching relationship in the image information. Depth value d;

The third calculating unit 40 is configured to calculate an actual distance T between the two cameras according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras.

In this embodiment, the binocular camera can be applied to a terminal device that requires imaging, such as a mobile phone, a scan. Ground robots, etc. The terminal in this embodiment is described by taking a binocular vision cleaning robot as an example.

The two cameras included in the binocular camera have different external parameters depending on the installed terminal equipment. Therefore, when the binocular camera is mounted on the terminal (for example, a binocular vision cleaning robot), it needs to be calibrated. The calibration is to calculate the external parameters of the binocular camera, and the external parameters include the actual between the binocular cameras. The distance T, and the transformation matrix R of the binocular camera rotating movement.

In this embodiment, before the calibration, the laser ranging sensor and the installed binocular camera are disposed in the same depth plane (the same vertical plane), and preferably, the laser ranging sensor is located in the binocular camera. The midpoint of the connection between the two cameras. The measuring unit 10 can measure the actual distance D from the specified target by the laser ranging sensor, and the specified target refers to a texture-rich target, such as a mural on a wall, a photo frame, etc., and the texture-rich designated target facilitates the feature point. Identification. The laser ranging sensor emits infrared light to form an infrared spot on the specified target, and the two cameras respectively collect infrared spot image information on the specified target; the second calculating unit 30 according to the image information, the infrared spot is in two The matching relationship in the frame image calculates the depth value d between the laser ranging sensor and the infrared spot, which is a scale value representing how many scales there are between the infrared spot and the laser ranging sensor. It can be understood that, according to the actual distance D and the depth value d described above, the scale s, that is, the distance value of each scale can be calculated.

Further, the first calculating unit 20 separately collects the image information of the specified target through the two cameras, and calculates the conversion matrix R of the rotation movement of the binocular camera and the scaling of the distance between the two cameras according to the collected image information of the specified target. The value t, which is not an actual distance value, represents how many scales there are between the two cameras. Since the above measurements are all measured in the same state (ie, keeping the laser ranging sensor and the position of the double-sided camera unchanged), the scale s is the same. Finally, the third calculating unit 40 can calculate the actual distance T between the two cameras according to the above-mentioned scale s and the zoom value t.

In summary, the conversion matrix R of the rotational movement of the double-sided camera and the actual distance T between the two cameras are calculated, and the external reference calibration of the double-sided camera is completed. Accurate access to external parameters is simple and convenient, with less computational complexity and no human calibration.

Referring to FIG. 5, in an embodiment, the first calculating unit 20 includes:

The extracting subunit 201 is configured to perform feature extraction on the collected image information of the specified target, and extract two Matching feature points in the image captured by the camera;

The first calculating sub-unit 202 is configured to calculate, according to the matched feature points, a conversion matrix R of the rotation movement of the binocular camera and a scaling value t of the distance between the two cameras.

In this embodiment, image acquisition is performed on a specified target by two cameras, and the extraction subunit 201 extracts feature information of the image information, extracts the same feature points in the two frames of images for matching, and extracts matching features in the images acquired by the two cameras. The number of points is eight, and the first calculating sub-unit 202 calculates the conversion matrix R of the rotation of the binocular camera and the scaling value t of the distance between the two cameras by using the normalized eight-point method according to the matched feature points. The scaling value t is not an actual distance value, which represents how many scales there are between the two cameras.

In another embodiment, the double-sided camera calibration device further includes:

The denoising unit is configured to perform denoising processing on the collected image information of the specified target. Specifically, denoising processing may be performed using Gaussian filtering, median filtering, or morphological operations or the like.

In another embodiment, the second calculating unit 30 is specifically configured to:

According to the matching relationship of the infrared spot in the image acquired by the two cameras, the depth value d between the laser ranging sensor and the infrared spot is calculated by a trigonometry method.

In this embodiment, the two cameras respectively acquire an image of one frame of infrared light spots, and the image is matched according to the same feature point of the infrared light spot, and the second calculating unit 30 calculates the laser distance measuring sensor and the infrared light by using the triangulation method. The depth value d between the points; the depth value d is a scale value representing how many scales there are between the infrared spot and the laser ranging sensor.

Referring to FIG. 6, in the above embodiment, the third calculating unit 40 includes:

a second calculating subunit 401, configured to calculate a scale s according to the actual distance D and the depth value d;

The third calculating subunit 402 is configured to calculate an actual distance T between the two cameras according to the scale s and the scaling value t of the distance between the two cameras.

In this embodiment, the actual distance between the specified target and the laser ranging sensor is D, and the depth value between the laser ranging sensor and the infrared spot is d, so the second calculating sub-unit 401 can calculate the current The scale s in the measurement state. The scale s is a scale value of the distance between the two cameras, so according to the scale s and the zoom value t of the distance between the two cameras, the third calculation sub-unit 402 can calculate the actual between the two cameras. Distance T.

[Correction according to Rule 26 23.02.2018]
Specifically, the calculation formula of the scale s is s=D/d; the calculation formula of the actual distance T is T=s×t.

In summary, the binocular camera calibration method and device provided in the embodiment of the present invention, the terminal measures the actual distance D between the laser ranging sensor and the specified target through a laser ranging sensor located at the same depth plane as the binocular camera; The binocular camera separately collects infrared spot image information irradiated by the laser ranging sensor on the specified target, and calculates a depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information; The camera separately collects image information of the specified target, and calculates a conversion matrix R of the rotation movement of the binocular camera and a scaling value t of the distance between the two cameras according to the collected image information of the specified target; according to the actual distance D, the depth The value d and the scaling value t of the distance between the two cameras calculate the actual distance T between the two cameras; thereby calibrating the binocular camera; accurately calculating the actual distance T between the two cameras and the binocular The rotation matrix of the camera rotates to obtain the external parameters accurately, and it is simple and convenient, and the amount of calculation is small, no need for artificial Calibration.

Those skilled in the art will appreciate that each block of the block diagrams and/or block diagrams and/or flow diagrams and combinations of blocks in the block diagrams and/or block diagrams and/or flow diagrams can be implemented by computer program instructions. . Those skilled in the art will appreciate that these computer program instructions can be implemented by a general purpose computer, a professional computer, or a processor of other programmable data processing methods, such that the processor is executed by a computer or other programmable data processing method. The blocks of the disclosed structure and/or block diagrams and/or flow diagrams or blocks specified in the various blocks.

Those skilled in the art can understand that the steps, measures, and solutions in the various operations, methods, and processes that have been discussed in the present invention may be alternated, changed, combined, or deleted. Further, other steps, measures, and schemes of the various operations, methods, and processes that have been discussed in the present invention may be alternated, modified, rearranged, decomposed, combined, or deleted. Further, the steps, measures, and solutions in the prior art having various operations, methods, and processes disclosed in the present invention may also be alternated, changed, rearranged, decomposed, combined, or deleted.

The above description is only a preferred embodiment of the present invention, and thus does not limit the scope of the patent of the present invention. The equivalent structure or equivalent flow transformation made by the present specification and the contents of the drawings, or directly or indirectly applied to other related technical fields, are all included in the scope of patent protection of the present invention.

Claims (16)

A binocular camera calibration method, comprising the steps of: The terminal measures the actual distance D between the laser ranging sensor and the specified target through a laser ranging sensor located at the same depth plane as the binocular camera; The image information of the specified target is respectively acquired by the binocular camera, and the conversion matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras are calculated according to the collected image information of the specified target; Acquiring the infrared spot image information of the laser ranging sensor on the specified target by the binocular camera, and calculating the depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information; The actual distance T between the two cameras is calculated from the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras. The binocular camera calibration method according to claim 1, wherein the conversion matrix R of the rotation movement of the binocular camera and the scaling value of the distance between the two cameras are calculated according to the acquired image information of the specified target. Steps include: Performing feature extraction on the collected image information of the specified target, and extracting matching feature points in the images acquired by the two cameras; According to the matched feature points, the conversion matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras are calculated. The binocular camera calibration method according to claim 2, wherein the number of matching feature points in the images acquired by the two cameras is eight, and the rotation of the binocular camera is calculated according to the matched feature points. The step of converting the matrix R and the scaling value t of the distance between the two cameras includes: According to the matched eight feature points, the normalized eight-point method is used to calculate the transformation matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras. The binocular camera calibration method according to claim 1, wherein the step of separately acquiring the image information of the specified target by the binocular camera further comprises: Perform image denoising on the acquired image information of the specified target. The binocular camera calibration method according to claim 1, wherein the step of calculating a depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information comprises: According to the matching relationship of the infrared spot in the image acquired by the two cameras, the depth value d between the laser ranging sensor and the infrared spot is calculated by a trigonometry method. The binocular camera calibration method according to claim 1, wherein the actual distance between the two cameras is calculated according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras. The steps of T include: Calculating a scale s according to the actual distance D and the depth value d; The actual distance T between the two cameras is calculated based on the scale s and the scaling value t of the distance between the two cameras. [Correction according to Rule 26 23.02.2018]
The binocular camera calibration method according to claim 6, wherein the calculation formula of the scale s is
s=D/d
The calculation formula of the actual distance T is
T=s×t
. The binocular camera calibration method according to claim 1, wherein the laser ranging sensor is located at a midpoint of two camera connections in the binocular camera; The terminal is a binocular vision cleaning robot. A binocular camera calibration device, comprising: a measuring unit, configured to measure an actual distance D between the laser ranging sensor and the specified target by a laser ranging sensor located at the same depth plane as the binocular camera; The first calculating unit is configured to separately collect image information of the specified target through the binocular camera, and calculate a conversion matrix R of the rotation movement of the binocular camera and a scaling value of the distance between the two cameras according to the collected image information of the specified target t; a second calculating unit for separately collecting laser ranging sensor photos through a binocular camera Infrared spot image information incident on the specified target, and calculating a depth value d between the laser ranging sensor and the infrared spot according to the matching relationship in the image information; And a third calculating unit, configured to calculate an actual distance T between the two cameras according to the actual distance D, the depth value d, and the scaling value t of the distance between the two cameras. The binocular camera calibration apparatus according to claim 9, wherein the first calculation unit comprises: Extracting a sub-unit for performing feature extraction on the collected image information of the specified target, and extracting matching feature points in the images acquired by the two cameras; The first calculating subunit is configured to calculate a conversion matrix R of the rotation movement of the binocular camera and a scaling value t of the distance between the two cameras according to the matched feature points. The binocular camera calibration device according to claim 10, wherein the number of matching feature points in the images acquired by the two cameras is eight, and the first calculation sub-unit is specifically configured to: According to the matched eight feature points, the normalized eight-point method is used to calculate the transformation matrix R of the rotation movement of the binocular camera and the scaling value t of the distance between the two cameras. The binocular camera calibration device according to claim 9, further comprising: The denoising unit is configured to perform denoising processing on the collected image information of the specified target. The binocular camera calibration device according to claim 9, wherein the second calculation unit is specifically configured to: According to the matching relationship of the infrared spot in the image acquired by the two cameras, the depth value d between the laser ranging sensor and the infrared spot is calculated by a trigonometry method. The binocular camera calibration device according to claim 9, wherein the third calculating unit comprises: a second calculating subunit, configured to calculate a scale s according to the actual distance D and the depth value d; And a third calculating subunit, configured to calculate an actual distance T between the two cameras according to the scale s and a scaling value t of the distance between the two cameras. [Correction according to Rule 26 23.02.2018]
The binocular camera calibration apparatus according to claim 14, wherein the scale s is calculated as
s=D/d
The calculation formula of the actual distance T is
T=s×t
. The binocular camera calibration device according to claim 9, wherein the laser ranging sensor is located at a midpoint of a connection between two cameras of the binocular camera; The terminal is a binocular vision cleaning robot.

Images