A method for extrinsic calibration of visual sensors based on stereo targets

Abstract

The present invention discloses a method for calibrating external parameters of a visual sensor based on a stereo target. The method comprises the following steps: a mounting base is provided with at least two cylindrical protrusions, and a visual sensor is provided with at least two reference holes; the two protrusions are plugged together during installation; a housing coordinate system of the visual sensor is established using the reference holes or the cylindrical protrusions; the visual sensor is mounted on the calibration fixture in the laboratory, a two-dimensional image of the stereo target in the field of view is collected, and the conversion relationship between the camera coordinate system and the housing coordinate system is obtained by combining the two-dimensional image and the standard coordinates of the standard holes / spheres on the stereo target in the housing coordinate system; at a detection station, the visual sensor is mounted on the mounting base; a conversion relationship between the coordinate system of the object to be measured and the housing coordinate system is obtained using a standard instrument; and the conversion relationship between the coordinate system of the object to be measured and the camera coordinate system is obtained by combining the conversion relationship between the camera coordinate system and the housing coordinate system, thereby completing the external parameter calibration. The method can accurately obtain the desired value, is simple to operate, and shortens the time of the external parameter calibration process.

Description

Technical Field

[0001] The present invention relates to the field of sensor calibration, and in particular to a method for calibrating external parameters of a visual sensor based on a stereo target. Background Art

[0002] As we all know, the external parameter calibration process of visual sensors is a prerequisite for the normal use of sensors. The effectiveness of the calibration results is directly related to the accuracy of subsequent detection results. In industrial sites, in order to accurately calibrate the conversion relationship (external parameters) between the camera coordinate system and the coordinate system of the object to be measured, the following process is usually used: That is, the sensor housing coordinate system is used as a transfer to obtain the relationship between the camera coordinate system and the coordinate system of the object to be measured. To this end, it is necessary to obtain the conversion relationship between the sensor housing coordinate system and the camera coordinate system. Currently, the calibration process of the sensor housing coordinate system and the camera coordinate system mainly presents the following form:

[0003] First, using a planar target (such as a checkerboard target or a concentric circle target), the sensor needs to capture the planar target image and obtain the feature information on the target. Then, with the help of a photogrammetry system, the relationship between the camera coordinate system, the target coordinate system, and the external coordinate system is established. Through coordinate system transformation, the external parameter matrix is ​​obtained. This method has the following disadvantages:

[0004] 1) Complex setup environment and many errors: The solution requires multiple conversions from camera coordinates to target coordinates, then from the target coordinates to theodolite coordinates, and finally from theodolite coordinates to the housing coordinates. This results in many errors and reduces calibration accuracy.

[0005] 2) Strict requirements on target processing accuracy and high equipment cost.

[0006] 3) Time-consuming: It takes 30 minutes to calibrate the external parameters of a single sensor.

[0007] Second, using a three-dimensional target, which is currently only applicable to line structured light sensors. The line laser in the sensor projects a laser strip onto a feature on the target, and the extrinsic parameters of the line structured light sensor are solved based on the three-dimensional coordinates of the feature. For example, patent document CN 111256592 B proposes an extrinsic parameter calibration device and method for a structured light sensor. Its target includes multiple standard spheres, requiring that the centers of the standard spheres are not on the same spatial plane, and the horizontal distance between the centers of two adjacent standard spheres is greater than the sum of the radii of the two standard spheres, and the vertical distance is less than the sum of the radii of the two standard spheres. When in use, it is necessary to ensure that the light plane can cover all the standard spheres in the calibration device. Another patent document, CN 111256591 B, proposes a method for calibrating visual sensor extrinsic parameters based on a stereoscopic target. The target is equipped with multiple standard holes, the centers of which are not located on the same spatial plane. The horizontal distance between the centers of two adjacent standard holes is greater than the sum of the radii of the two standard holes, and the vertical distance is less than the sum of the radii of the two standard holes. During calibration, a projector is required to project structured light onto the calibration device, and the light plane must cover all the standard holes in the calibration device. This method has the following problems:

[0008] 1) Poor compatibility: Due to the inevitable differences in the installation positions of cameras and lasers, it is difficult to ensure that the light plane can always cover all standard balls / holes in the calibration device; therefore, the target is difficult to be compatible with different types of sensors.

[0009] 2) Complex operation: Since the light plane needs to pass through each standard sphere / hole simultaneously, and to ensure the accuracy of the sphere / hole center coordinate solution, the laser bar needs to be projected at 1 / 4 to 1 / 3 of the standard sphere / hole, that is, as close to the sphere / hole center as possible. If it is projected at 1 / 6 to 1 / 5 of the standard sphere / hole, the sphere / hole center fitting result will be reduced. Therefore, during calibration, it is necessary to manipulate the sensor and place it at the specified position and angle to achieve a satisfactory result. The process is tedious and time-consuming (calibrating the external parameters of a single sensor takes about 20 minutes).

[0010] 3) Intermediate errors are introduced during the solution process: When calculating the transformation relationship from the camera coordinate system to the sensor's external coordinate system, the light plane is needed to obtain the three-dimensional coordinates of the sphere center / hole center. Introducing the light plane coordinates will introduce intermediate errors, affecting the calibration accuracy. Summary of the Invention

[0011] In response to the above problems, the present invention proposes a method for extrinsic parameter calibration of a visual sensor based on a stereo target. This method can quickly establish a shell coordinate system and accurately obtain the conversion relationship between the camera coordinate system and the coordinate system of the object to be measured. The stereo target is used in the calibration process of the shell coordinate system and the camera coordinate system, which reduces the dependence on the external environment in which the calibration is performed. The operation is simple and only requires the acquisition of a two-dimensional image of the stereo target, which shortens the time of the extrinsic parameter calibration process, reduces the introduction of errors, and has high accuracy in the calibration results.

[0012] The technical solution is as follows:

[0013] A method for calibrating external parameters of a visual sensor based on a stereo target. The visual sensor mounting base is provided with at least two cylindrical protrusions, and correspondingly, the visual sensor is provided with at least two reference holes. The two are inserted into each other when the visual sensor is installed. The reference holes or cylindrical protrusions are used to establish the outer shell coordinate system of the visual sensor.

[0014] A calibration tool is set up in the laboratory, wherein the calibration tool is provided with columnar protrusions having the same number and position as those on the mounting base;

[0015] The visual sensor is mounted on the calibration fixture to collect a two-dimensional image of the stereo target in the field of view, and the conversion relationship between the camera coordinate system and the outer shell coordinate system is obtained and stored by combining the two-dimensional image and the standard coordinates of the standard hole / sphere on the stereo target in the outer shell coordinate system;

[0016] The three-dimensional target is provided with at least four standard holes / balls, and the centers of the standard balls and the centers of the standard holes are not on the same spatial plane;

[0017] At the inspection station, the visual sensor is mounted on the mounting base, and the object to be measured is placed in a preset position. The standard instrument is used to obtain the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument, and the conversion relationship between the coordinate system of the housing and the coordinate system of the standard instrument; the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the housing is obtained;

[0018] Combined with the pre-stored conversion relationship between the camera coordinate system and the housing coordinate system, the conversion relationship between the coordinate system of the object to be measured and the camera coordinate system is obtained to complete the external parameter calibration.

[0019] Furthermore, the method for obtaining the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument using the standard instrument is as follows:

[0020] The standard instrument collects multiple features on the surface of the object to be measured, wherein the features are holes, points or external spheres;

[0021] The coordinates of each feature in the coordinate system of the object to be measured and in the coordinate system of the standard instrument are used to solve the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument.

[0022] Furthermore, the method for obtaining the conversion relationship between the shell coordinate system and the standard instrument coordinate system using the standard instrument is as follows:

[0023] Fixing multiple feature points or balls on the housing surface of the vision sensor;

[0024] The coordinates of the feature points or spheres are collected using a standard instrument, and the conversion relationship between the outer shell coordinate system of the vision sensor and the standard instrument coordinate system is solved using the coordinates of the feature points or spheres in the outer shell coordinate system and the coordinates in the standard instrument coordinate system.

[0025] Furthermore, the method of establishing the housing coordinate system of the vision sensor using the reference hole or cylindrical protrusion is as follows:

[0026] Use standard instruments to obtain the three-dimensional coordinates of each reference hole / cylindrical protrusion respectively, take the three-dimensional coordinate point of one of the reference holes / cylindrical protrusions as the coordinate origin, and take the line connecting the coordinate origin and any other reference hole / cylindrical protrusion as a coordinate axis; the direction perpendicular to the plane where the reference hole / cylindrical protrusion is located is another coordinate axis, and construct the third coordinate axis through the coordinate origin along the direction perpendicular to the first two coordinate axes to complete the establishment of the shell coordinate system.

[0027] Preferably, the standard instrument is a three-coordinate machine, a laser tracker, an imager, a total station, a theodolite or an articulated arm measuring machine.

[0028] Furthermore, the method for solving the transformation relationship between the camera coordinate system and the shell coordinate system is:

[0029] According to the spatial position relationship between each standard hole / ball on the three-dimensional target, an ID number is set for each standard hole / ball in advance, and the pixel coordinates and standard coordinates corresponding to the same ID number are found. Based on the PNP principle, the conversion relationship between the camera coordinate system and the shell coordinate system is solved;

[0030] The standard coordinates are the three-dimensional coordinates of the geometric center of each standard hole / sphere in the shell coordinate system obtained using a standard instrument.

[0031] In order to facilitate the distinction between the various standard holes / balls, preferably, the processing size of one or more standard holes / balls of the three-dimensional target is different from that of other standard holes / balls, or the processing sizes of the various standard holes / balls are different from each other.

[0032] Furthermore, the three-dimensional target is provided with a plurality of planes of different heights, and the plurality of planes are evenly distributed within the depth of field of the camera; and a single plane is provided with one or more standard holes / balls;

[0033] The standard holes are circular holes or regular polygonal holes; when in use, the upper surface of each standard hole is perpendicular to the optical axis of the camera;

[0034] Multiple planes are distributed in a stepped manner, gradually rising or falling, or first rising and then falling, or first falling and then rising, or randomly distributed in terms of height.

[0035] Preferably, the three-dimensional target is provided with one or more inclined surfaces, and the surface of the inclined surface is flat or curved; each standard ball on the three-dimensional target is evenly distributed within the depth of field of the camera;

[0036] The angle between the inclined plane and the horizontal plane is 20° to 60°;

[0037] When a plurality of standard balls are provided on a single inclined surface, the standard balls are staggered in distribution.

[0038] This method has the following characteristics:

[0039] ① The method of the present invention provides a reference hole on the housing surface of the vision sensor and matches it with a special calibration fixture with a cylindrical protrusion. During installation, the cylindrical protrusion is respectively embedded in the reference hole; the housing coordinate system is established using the reference hole or the cylindrical protrusion. This special design can quickly establish the sensor housing coordinate system and realize batch calibration of different vision sensors:

[0040] When there are many visual sensors that need to be calibrated, you only need to fix the visual sensors to be calibrated to the cylindrical protrusions to quickly establish the sensor housing coordinate system. Then, combined with the three-dimensional target, you can get the relationship between the camera coordinate system and the housing coordinate system. Combined with the standard instrument, you can get the relationship between the housing coordinate system and the coordinate system of the object to be measured, and finally deduce the conversion relationship between the camera coordinate system and the coordinate system of the object to be measured.

[0041] The whole process is efficient and accurate, and is suitable for rapid calibration of sensors.

[0042] ② Using a three-dimensional target with a simple and ingenious shape, the target can evenly distribute multiple standard holes / balls within the effective depth of field of the camera through the design of the target step blocks / slant surfaces. Compared with a plane target, the coordinates of the feature points (hole center, sphere center coordinates) change in the three coordinate axis directions, making the calibration results more accurate.

[0043] This method is not only applicable to the calibration of various types of structured light sensors, but also to monocular, binocular vision sensors and surface structured light vision sensors. The calibration process only requires the participation of the camera, and no other components (laser projectors, projectors) are required throughout the process. It is easy to operate and has a fast solution speed. The calibration of the external parameters of a single sensor takes about 5 minutes. The advantages are more significant during the calibration process of online structured light vision sensors or surface structured light sensors.

[0044] ③ In the specific solution process, standard equipment is used to directly obtain the relationship between the camera coordinate system and the external coordinate system. There is no need for the target coordinate system or light plane to participate in the solution, which reduces the intermediate error and makes the calibration result more accurate. The extrinsic parameter calibration result is evaluated by the mean of the back-projection error. The extrinsic parameter matrix solved by this method has a solution error of less than 0.15 / pixel. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of a three-dimensional target structure in a specific embodiment;

[0046] Figure 2 This is a schematic diagram of the three-dimensional target structure of the second form in a specific embodiment;

[0047] Figure 3 Schematic diagram of the three-dimensional target structure in a specific embodiment;

[0048] Figure 4 The camera captures an image of a three-dimensional target in a specific embodiment;

[0049] Figure 5 Schematic diagram of the four-dimensional target structure in a specific embodiment;

[0050] Figure 6 Schematic diagram of the target structure of form five in a specific embodiment;

[0051] Figure 7 Schematic diagram of the structure of the calibration tool or installation base;

[0052] Figure 8 Schematic diagram of the visual sensor housing structure. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] A method for extrinsic calibration of visual sensors based on stereo targets, such as Figure 7 As shown, the mounting base of the visual sensor is provided with at least two columnar protrusions, corresponding to Figure 8 As shown, the vision sensor is provided with at least two reference holes; the two are plugged together when the vision sensor is installed; the reference holes or the cylindrical protrusions are used to establish the housing coordinate system of the vision sensor;

[0055] A calibration fixture is set up in the laboratory. The calibration fixture is provided with columnar protrusions with the same number and position as those on the mounting base (in specific implementation, the calibration fixture can be exactly the same as the mounting base);

[0056] The visual sensor is installed on the calibration fixture to collect the two-dimensional image of the stereo target in the field of view (such as Figure 4As shown), combining the two-dimensional image and the standard coordinates of the standard hole / sphere on the three-dimensional target in the shell coordinate system, obtaining the conversion relationship between the camera coordinate system and the shell coordinate system, and storing it;

[0057] There are at least four standard holes / balls on the three-dimensional target, and the centers of the standard balls and the centers of the standard holes are not on the same spatial plane;

[0058] The above process is performed in the laboratory and can be batch-calibrated for different vision sensors. Simply machining a reference hole in the sensor housing and mounting it on a calibration fixture quickly establishes the housing coordinate system. By capturing a 2D image of the stereo target, the same calculation method is used to quickly determine the conversion relationship between the camera coordinate system and the housing coordinate system. This enables batch calibration, making the entire process fast and efficient.

[0059] When the vision sensor leaves the factory, the conversion relationship between the camera coordinate system and the housing coordinate system is known.

[0060] The following steps describe the use of the vision sensor after it leaves the factory:

[0061] At the inspection station, the visual sensor is mounted on the mounting base, and the object to be measured is placed in a preset position. The standard instrument is used to obtain the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument, and the conversion relationship between the coordinate system of the housing and the coordinate system of the standard instrument; the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the housing is obtained;

[0062] Combined with the pre-stored conversion relationship between the camera coordinate system and the housing coordinate system, the conversion relationship between the coordinate system of the object to be measured and the camera coordinate system is obtained to complete the external parameter calibration.

[0063] Among them, the standard instruments are three-coordinate machines, laser trackers, imagers, total stations, theodolites or articulated arm measuring machines.

[0064] The method for obtaining the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument using the standard instrument is as follows:

[0065] The standard instrument collects multiple features on the surface of the object to be measured, which are holes, points or peripheral balls;

[0066] The coordinates of each feature in the coordinate system of the object to be measured and in the coordinate system of the standard instrument are used to solve the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument.

[0067] The method for obtaining the conversion relationship between the shell coordinate system and the standard instrument coordinate system using a standard instrument is as follows:

[0068] Fixing multiple feature points or balls on the housing surface of the vision sensor;

[0069] The coordinates of the feature points or spheres are collected using a standard instrument, and the conversion relationship between the outer shell coordinate system of the vision sensor and the standard instrument coordinate system is solved using the coordinates of the feature points or spheres in the outer shell coordinate system and the coordinates in the standard instrument coordinate system.

[0070] Specifically, the method of establishing the housing coordinate system of the vision sensor using the reference hole or the cylindrical protrusion is as follows:

[0071] Use standard instruments to obtain the three-dimensional coordinates of each reference hole / cylindrical protrusion respectively, take the three-dimensional coordinate point of one of the reference holes / cylindrical protrusions as the coordinate origin, and take the line connecting the coordinate origin and any other reference hole / cylindrical protrusion as a coordinate axis; the direction perpendicular to the plane where the reference hole / cylindrical protrusion is located is another coordinate axis, and construct the third coordinate axis through the coordinate origin along the direction perpendicular to the first two coordinate axes to complete the establishment of the shell coordinate system.

[0072] More specifically, the method for solving the transformation relationship between the camera coordinate system and the shell coordinate system is:

[0073] According to the spatial position relationship between each standard hole / ball on the three-dimensional target, an ID number is set for each standard hole / ball in advance, and the pixel coordinates and standard coordinates corresponding to the same ID number are found. Based on the PNP principle, the conversion relationship between the camera coordinate system and the shell coordinate system is solved;

[0074] The standard coordinates are the three-dimensional coordinates of the geometric center of each standard hole / sphere in the shell coordinate system obtained using a standard instrument.

[0075] In order to facilitate the distinction between the various standard holes / balls, as a preferred embodiment, the processing size of one or more standard holes / balls of the three-dimensional target is different from that of other standard holes / balls, or the processing sizes of the standard holes / balls are different from each other.

[0076] The shape of the three-dimensional target can be set as follows:

[0077] like Figures 1 to 3 As shown, the three-dimensional target is provided with multiple planes of different heights, and the multiple planes are evenly distributed within the depth of field of the camera; one or more standard holes / balls are provided on a single plane;

[0078] The standard holes are circular holes or regular polygonal holes; when in use, the upper surface of each standard hole is perpendicular to the optical axis of the camera;

[0079] Multiple planes are distributed in a stepped manner, gradually rising or falling, or first rising and then falling, or first falling and then rising, or randomly distributed in terms of height.

[0080] like Figures 5-6As shown, the three-dimensional target is provided with one or more inclined surfaces, and the surface of the inclined surface is flat or curved; each standard ball on the three-dimensional target is evenly distributed within the depth of field of the camera;

[0081] The angle between the inclined plane and the horizontal plane is 20° to 60°;

[0082] When a plurality of standard balls are provided on a single inclined surface, the standard balls are staggered in distribution.

[0083] When used, the three-dimensional targets can be used alone or in combination.

[0084] The following is an example of extrinsic parameter calibration of a visual sensor used in an automobile body-in-white inspection station:

[0085] In this embodiment, the standard instrument is a laser tracker, and the stereo target is as follows: Figure 3 As shown, there are 5 planes with different heights, and multiple planes are evenly distributed within the depth of field of the camera; multiple standard holes are set on a single plane.

[0086] A method for extrinsic calibration of visual sensors based on stereo targets, such as Figure 7 As shown, there are two columnar protrusions 1 on the mounting base of the visual sensor, corresponding to Figure 8 As shown, the visual sensor is also provided with two reference holes 2; the two are plugged together when the visual sensor is installed;

[0087] Use the reference hole or cylindrical protrusion to establish the housing coordinate system of the vision sensor; the method is as follows:

[0088] Take the three-dimensional coordinate point of one of the reference holes / cylindrical protrusions as the coordinate origin, and the line connecting the coordinate origin and the other reference hole / cylindrical protrusion as a coordinate axis; the direction perpendicular to the plane where the reference hole / cylindrical protrusion is located is another coordinate axis, and the third coordinate axis is constructed through the coordinate origin along the direction perpendicular to the first two coordinate axes to complete the establishment of the shell coordinate system.

[0089] The three-dimensional coordinates of the geometric center of each standard hole in the shell coordinate system obtained by using a laser tracker are recorded as standard coordinates.

[0090] A calibration fixture is set up in the laboratory, and the calibration fixture is provided with columnar protrusions with the same number and position as those on the mounting base;

[0091] The visual sensor is installed on the calibration fixture to collect the two-dimensional image of the stereo target in the field of view (such as Figure 4 As shown), combining the two-dimensional image and the standard coordinates of the standard hole on the three-dimensional target in the shell coordinate system, obtaining the conversion relationship between the camera coordinate system and the shell coordinate system, and storing it;

[0092] Among them, the method for solving the transformation relationship between the camera coordinate system and the shell coordinate system is:

[0093] According to the spatial position relationship between the standard holes on the three-dimensional target, an ID number is set for each standard hole / ball in advance, and the pixel coordinates and standard coordinates corresponding to the same ID number are found. Based on the PNP principle, the conversion relationship between the camera coordinate system and the shell coordinate system is solved.

[0094] At the inspection station, the vision sensor is mounted on a mounting base, and the body-in-white is placed in a preset position. The mounting base can be fixed around the body-in-white or fixed to the end of a robot, and the robot is placed around the body-in-white.

[0095] Use standard instruments to obtain the conversion relationship between the body-in-white coordinate system and the standard instrument coordinate system:

[0096] Standard instruments capture multiple holes on the body-in-white surface;

[0097] The conversion relationship between the body-in-white coordinate system and the standard instrument coordinate system is solved using the coordinates of each hole in the body-in-white coordinate system and the coordinate system of the standard instrument.

[0098] Use standard instruments to obtain the conversion relationship between the shell coordinate system and the standard instrument coordinate system:

[0099] A plurality of tracker target balls are fixed on the housing surface of the vision sensor;

[0100] The coordinates of the target sphere are collected by a standard instrument, and the conversion relationship between the outer shell coordinate system of the vision sensor and the standard instrument coordinate system is solved using the coordinates of the target sphere in the outer shell coordinate system and the coordinates in the standard instrument coordinate system.

[0101] Combining the above two transformation relationships, the transformation relationship between the body-in-white coordinate system and the shell coordinate system is obtained;

[0102] Combined with the pre-stored conversion relationship between the camera coordinate system and the shell coordinate system, the conversion relationship between the body-in-white coordinate system and the camera coordinate system is obtained to complete the external parameter calibration.

[0103] The foregoing descriptions of specific exemplary embodiments of the present invention have been presented for purposes of illustration and description. The foregoing descriptions are not intended to be exhaustive or to limit the invention to the precise forms disclosed, and it is apparent that many variations and modifications are possible in light of the foregoing teachings. The exemplary embodiments have been chosen and described in order to explain the specific principles of the invention and their practical application, thereby enabling others skilled in the art to make and utilize the various exemplary embodiments of the invention and various alternatives and modifications thereof. The scope of the invention is intended to be defined by the appended claims and their equivalents.

Claims

1. A method for calibrating external parameters of a visual sensor based on a stereoscopic target. The visual sensor's mounting base is provided with at least two cylindrical protrusions, and correspondingly, the visual sensor is provided with at least two reference holes. The two protrusions are inserted into each other when the visual sensor is installed. The reference holes or cylindrical protrusions are used to establish the visual sensor's housing coordinate system. Its characteristics are: A calibration tool is set up in the laboratory, wherein the calibration tool is provided with columnar protrusions having the same number and position as those on the mounting base; The visual sensor is mounted on the calibration fixture to collect a two-dimensional image of the stereo target in the field of view, and the conversion relationship between the camera coordinate system and the outer shell coordinate system is obtained and stored by combining the two-dimensional image and the standard coordinates of the standard hole / sphere on the stereo target in the outer shell coordinate system; The three-dimensional target is provided with at least four standard holes / balls, and the centers of the standard balls and the centers of the standard holes are not on the same spatial plane; At the inspection station, the visual sensor is mounted on the mounting base, and the object to be measured is placed in a preset position. The standard instrument is used to obtain the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument, and the conversion relationship between the coordinate system of the housing and the coordinate system of the standard instrument; the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the housing is obtained; Combined with the pre-stored conversion relationship between the camera coordinate system and the housing coordinate system, the conversion relationship between the coordinate system of the object to be measured and the camera coordinate system is obtained to complete the external parameter calibration.

2. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1, characterized in that: The method for obtaining the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument using a standard instrument is as follows: The standard instrument collects multiple features on the surface of the object to be measured, wherein the features are holes, points or external spheres; The coordinates of each feature in the coordinate system of the object to be measured and in the coordinate system of the standard instrument are used to solve the conversion relationship between the coordinate system of the object to be measured and the coordinate system of the standard instrument.

3. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1, characterized in that: The method for obtaining the conversion relationship between the shell coordinate system and the standard instrument coordinate system using a standard instrument is as follows: Fixing multiple feature points or balls on the housing surface of the vision sensor; The coordinates of the feature points or spheres are collected using a standard instrument, and the conversion relationship between the outer shell coordinate system of the vision sensor and the standard instrument coordinate system is solved using the coordinates of the feature points or spheres in the outer shell coordinate system and the coordinates in the standard instrument coordinate system.

4. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1, wherein: The method for establishing the housing coordinate system of the vision sensor using reference holes or cylindrical protrusions is as follows: Use standard instruments to obtain the three-dimensional coordinates of each reference hole / cylindrical protrusion, use the three-dimensional coordinate point of one reference hole / cylindrical protrusion as the coordinate origin, and use the line connecting the coordinate origin and any other reference hole / cylindrical protrusion as a coordinate axis; The direction perpendicular to the plane where the reference hole / columnar protrusion is located is another coordinate axis. Through the coordinate origin, a third coordinate axis is constructed along the direction perpendicular to the first two coordinate axes to complete the establishment of the shell coordinate system.

5. The method for extrinsic calibration of a visual sensor based on a stereo target according to any one of claims 1 to 4, characterized in that: The standard instrument is a three-coordinate machine, a laser tracker, an imager, a total station, a theodolite or an articulated arm measuring machine.

6. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1, wherein: The method for solving the transformation relationship between the camera coordinate system and the shell coordinate system is: According to the spatial position relationship between each standard hole / ball on the three-dimensional target, an ID number is set for each standard hole / ball in advance, and the pixel coordinates and standard coordinates corresponding to the same ID number are found. Based on the PNP principle, the conversion relationship between the camera coordinate system and the shell coordinate system is solved; The standard coordinates are the three-dimensional coordinates of the geometric center of each standard hole / ball in the shell coordinate system obtained using a standard instrument.

7. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1 or 6, characterized in that: The processing size of one or more standard holes / balls of the three-dimensional target is different from that of other standard holes / balls, or the processing sizes of the standard holes / balls are different from each other.

8. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1, wherein: The three-dimensional target is provided with multiple planes of different heights, and the multiple planes are evenly distributed within the depth of field of the camera; one or more standard holes / balls are provided on a single plane; The standard holes are circular holes or regular polygonal holes; when in use, the upper surface of each standard hole is perpendicular to the optical axis of the camera; Multiple planes are distributed in a stepped manner, gradually rising or falling, or first rising and then falling, or first falling and then rising, or randomly distributed in terms of height.

9. The method for extrinsic calibration of a visual sensor based on a stereo target according to claim 1, wherein: The three-dimensional target is provided with one or more inclined surfaces, and the surface of the inclined surface is flat or curved; each standard ball on the three-dimensional target is evenly distributed within the depth of field of the camera; The angle between the inclined plane and the horizontal plane is 20° to 60°; When a plurality of standard balls are provided on a single inclined surface, the standard balls are staggered in distribution.

Images