CN114440776A - A method and system for automatic displacement measurement based on machine vision - Google Patents

Abstract

The invention discloses a displacement automatic measurement method and system based on machine vision, and the method comprises the following steps: calibrating a positioning point group of the target pattern; identifying and extracting a positioning point group of the target pattern to obtain image surface coordinates of each positioning point in the positioning point group; and establishing a two-dimensional coordinate system or a three-dimensional coordinate system of the object plane according to the image plane coordinates of the positioning points, and further transforming the image plane coordinates of the image to be detected into object plane coordinates. The invention has self-calibration function, can automatically construct the quantitative conversion relation from the image plane coordinate to the true object plane coordinate at any camera shooting angle, and can track the target by adopting various machine vision methods, thereby realizing the automatic measurement of the machine vision displacement without human intervention.

Description

A method and system for automatic displacement measurement based on machine vision

technical field

The invention relates to the technical field of machine vision, in particular to a method and system for automatic displacement measurement based on machine vision.

Background technique

Displacement measurement is one of the most frequently involved measurement contents in academic and engineering fields. With the development and popularization of machine vision technology, academia has gradually tried to combine machine vision and displacement measurement to realize displacement measurement based on machine vision, thus serving engineering practice. Machine vision displacement measurement has many advantages. Compared with traditional methods, it has obvious advantages such as non-contact, low cost, high precision, and real-time measurement. Therefore, displacement measurement based on machine vision has long been a research hotspot in academia. At this stage, there are many practical engineering cases and commercial products based on machine vision measurement of displacement at home and abroad.

Although there are many reported application cases of machine vision measurement of displacement, this advanced measurement method is still far from being widely used. One of the important reasons is the problem of external parameter calibration. Camera calibration involves two types of problems, one of which is the calibration of the physical scale related to the camera itself, such as the calibration of the internal optical parameters of the camera, or the calibration of the pose and attitude between the cameras inside the binocular camera group, which can be collectively referred to as the camera or camera group internal parameter calibration. . There are mature solutions for this type of problem, such as Zhang Zhengyou's calibration method.

There is another kind of problem, that is, the calibration of the conversion relationship from pixel scale to real physical scale. When measuring displacement with machine vision, the obtained original displacement data results are only pixel scale results in pixels, while the results required in actual engineering are displacement results in real physical scale (such as meters, millimeters, etc. ). At this time, it is necessary to calibrate the conversion relationship from pixel scale to physical scale, so as to convert the pixel displacement result into a physical scale displacement result with physical meaning, which is called external parameter calibration. However, there is no unified solution for this kind of external parameter calibration problem, and the biggest disadvantage of each existing solution is that human intervention is required.

Therefore, the current machine vision displacement measurement technology cannot realize the real automatic displacement measurement, and the calibration process that requires human intervention greatly increases the application barriers and measurement difficulties of the machine vision displacement measurement technology. In the actual application process, the situation is complex and changeable. For example, in the engineering site, the professional ability of the experimenters varies greatly. Therefore, the calibration method that requires human intervention is likely to affect the measurement results and greatly reduce the reliability of the results. It can be seen that the realization of the self-calibration method is of great significance for the automatic measurement of machine vision displacement.

SUMMARY OF THE INVENTION

Aiming at the problems existing in the above-mentioned prior art, the present invention proposes an automatic displacement measurement method and system based on machine vision, so as to solve the problem that intervention is required in the existing external parameter calibration.

In order to solve the above-mentioned technical problems, the present invention is achieved through the following technical solutions:

According to a first aspect of the present invention, there is provided an automatic displacement measurement method based on machine vision, comprising:

S11: The positioning point group of the calibration target pattern;

S12: Identify and extract the positioning point group of the target pattern, and obtain the image plane coordinates of each positioning point in the positioning point group;

S13: Establish a two-dimensional coordinate system or a three-dimensional coordinate system of the object plane according to the image plane coordinates of the positioning points, and then transform the image plane coordinates of the image to be measured into object plane coordinates.

Preferably, the S11 includes:

S111: demarcate the center of the target pattern to obtain a center positioning point;

S112: Demarcate the edge of the target pattern to obtain a plurality of edge positioning points, where the edge positioning points may be points close to the boundary of the target pattern, or may be points on the boundary of the target pattern.

Preferably, the S12 includes:

S121: Identify and extract the center positioning point of the target pattern, and obtain the image plane coordinates of the center positioning point;

S122: Identify and extract multiple edge positioning points of the target pattern to obtain image plane coordinates of the multiple edge positioning points.

Preferably, before the step S121, the method further includes: performing a binarization process on the target pattern.

Preferably, the center positioning point in the S11 is the center of the center positioning circle, and the edge positioning point is the center of the edge positioning circle; correspondingly,

The S121 includes:

S1211: Use a boundary extraction method to perform edge detection on the target pattern, and extract an edge topology table with edge level information;

S1212: By searching the edge level information of the edge topology table, lock the boundary of the center positioning circle;

S1213: Use least squares fitting to process the boundary of the center positioning circle, and obtain the center coordinates of the center positioning circle, which are the image plane coordinates of the center positioning point;

The S122 includes:

S1221: Use a boundary extraction method to perform ellipse detection on the target pattern, and lock the boundaries of the multiple edge positioning points;

S1222 : Process the boundaries of the plurality of edge positioning points by using least squares fitting to obtain the coordinates of the circle centers of the plurality of the edge positioning points, that is, the image plane coordinates of the plurality of the edge positioning points.

Preferably, the center positioning circles in S111 include at least two, at least two of the center positioning circles are concentric circles, and at least two of the center positioning circles have different radii; correspondingly,

The step S1213 is: using least squares fitting to process the boundary of the centering circle, obtaining the center coordinates of the centering circle, and performing averaging processing on the center coordinates of at least two or more of the centering circles, The image plane coordinates of the central positioning point are obtained.

Preferably, establishing a two-dimensional coordinate system of the object plane in the step S13, and then transforming the image plane image of the image to be measured into the object plane coordinates includes:

S1311: Use each locating point in the locating point group to solve the perspective transformation matrix under the two-dimensional coordinate system in which the image plane is mapped to the object plane;

S1312: Use the perspective transformation matrix to transform the image plane coordinates of the image to be measured into the two-dimensional coordinates of the object plane, and the transformation formula is:

Among them, x/s is the abscissa of the two-dimensional coordinate system of the object surface, y/s is the vertical coordinate of the two-dimensional coordinate system of the object surface, s is the scale coefficient, W is the perspective transformation matrix, and P is the homogeneous image surface point. Coordinate vector, P^ is the coordinate vector of the object surface point.

Preferably, the target pattern in S11 includes: a left-view target pattern obtained by one of the cameras of the binocular camera group and a right-view target pattern obtained by another camera;

The image plane coordinates of each positioning point in the positioning point group obtained in S12 respectively include: the image plane coordinates of the left view and the image plane coordinates of the right view;

The three-dimensional coordinate system of the object plane is established in the S13, and then the image plane image of the image to be measured is transformed into the object plane coordinates, including:

S1321: According to the internal parameter calibration result of the binocular camera group, reconstruct the image plane coordinates of the left view and the image plane coordinates of the right view of the positioning points to obtain the coordinates of the positioning points in the binocular camera group coordinate system The three-dimensional coordinate result within P _cam ;

S1322: Using the three-dimensional coordinate result P _cam of each positioning point in the binocular camera group coordinate system, the object plane equation of the plane where the target pattern is located is obtained by least squares fitting;

S1323: the x and y axes are built on the object surface, and the z axis is built in the normal direction of the object surface, and the three-dimensional coordinate of the center positioning point P _c in the positioning point group is used as the origin of the coordinate system, and the three-dimensional coordinate system of the object surface is established;

S1324: Set the matrix formed by the three-axis normalized coordinates of the three-dimensional coordinate system of the object plane in the binocular camera group coordinate system as R, and use P _c as a translation vector, and transform the to-be-measured The three-dimensional coordinate result P _cam of the image in the coordinate system of the binocular camera group is converted into the coordinate result P _world in the three-dimensional coordinate system of the object surface, and the conversion formula is:

为P_c的三维坐标。in,

is the three-dimensional coordinate of _Pc .

According to a second aspect of the present invention, an automatic displacement measurement system based on machine vision is provided, which includes: a positioning point group calibration module, an image plane coordinate acquisition module, and an object plane coordinate system acquisition module; wherein,

The positioning point group calibration module is used for calibrating the positioning point group of the target pattern;

The image plane coordinate acquisition module is used to identify and extract the positioning point group of the target pattern, and obtain the image plane coordinates of each positioning point in the positioning point group;

The object plane coordinate system acquisition module is configured to establish a two-dimensional or three-dimensional coordinate system of the object plane according to the image plane coordinates of the positioning points, and then transform the image plane coordinates of the image to be measured into object plane coordinates.

Preferably, the image plane coordinate acquisition module includes: an edge detection unit, a boundary locking unit for the center positioning circle, an image plane coordinate obtaining unit for the center positioning point, a boundary locking unit for the edge positioning circle, and the image plane coordinates for the edge positioning point. get unit; where,

The positioning point group in the positioning point group calibration module includes: a center positioning point and an edge positioning point, the center positioning point is the center of the center positioning circle, and the edge positioning point is the center of the edge positioning circle;

The edge detection unit is used to perform edge detection on the target pattern by using the boundary extraction method, and extract an edge topology table with edge level information;

The boundary locking unit of the center positioning circle is configured to lock the boundary of the center positioning circle by looking up the edge level information of the edge topology table;

The image plane coordinate obtaining unit of the center positioning point is used to process the boundary of the center positioning circle by using least squares fitting to obtain the center coordinates of the center positioning circle, which is the image plane of the center positioning point. coordinate;

The boundary locking unit of the edge positioning circle is configured to perform ellipse detection on the target pattern by using a boundary extraction method, and lock the boundaries of the plurality of edge positioning points;

The image plane coordinate obtaining unit of the edge positioning point is used to process the boundaries of a plurality of the edge positioning points by using least squares fitting, and obtain the center coordinates of the plurality of edge positioning points, that is, a plurality of the edge positioning points. The image plane coordinates of the edge anchor point.

Compared with the prior art, the present invention has the following advantages:

The method and system for automatic displacement measurement based on machine vision provided by the present invention can calibrate the target pattern and obtain the coordinate system of the object plane according to the calibrated target pattern, and then the image plane coordinates of the image to be measured can be transformed into the object plane coordinates, that is, The automatic external parameter calibration is realized without manual access, which avoids the influence of the calibration method of human intervention on the measurement results.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

1 is a flowchart of an automatic displacement measurement method based on machine vision according to an embodiment of the present invention;

2 is a schematic diagram of a positioning point group of a self-calibrating target pattern according to a preferred embodiment of the present invention;

3 is a self-calibration result in a two-dimensional displacement measurement scenario based on machine vision according to a preferred embodiment of the present invention;

FIG. 4 is a self-calibration result in a scene of three-dimensional displacement measurement based on machine vision according to a preferred embodiment of the present invention.

Detailed ways

The embodiments of the present invention are described in detail below. This embodiment is implemented on the premise of the technical solution of the present invention, and provides a detailed implementation manner and a specific operation process, but the protection scope of the present invention is not limited to the following implementation. example.

The terms "first", "second", "third", "fourth", etc. (if present) in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to Describe a particular order or sequence. It is to be understood that the data so used may be interchanged under appropriate circumstances such that the embodiments of the invention described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having" and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those expressly listed Rather, those steps or units may include other steps or units not expressly listed or inherent to these processes, methods, products or devices.

In one embodiment, the present invention provides an automatic displacement measurement method based on machine vision, please refer to FIG. 1 , which includes:

S11: The positioning point group of the calibration target pattern;

S12: Identify and extract the positioning point group of the target pattern, and obtain the image plane coordinates of each positioning point in the positioning point group;

S13: Establish a two-dimensional coordinate system or a three-dimensional coordinate system of the object plane according to the image plane coordinates of each positioning point, and then transform the image plane coordinates of the image to be measured into object plane coordinates.

The above-mentioned embodiments of the present invention do not require manual access, thereby avoiding the influence of the calibration method of human intervention on the measurement results.

Preferably, in one embodiment, S11 may further include:

S111: Calibrate the center of the target pattern to obtain the center positioning point;

S112: Calibrate the edge of the target pattern to obtain a plurality of edge positioning points.

Preferably, in one embodiment, S12 includes:

S121: Identify and extract the center positioning point of the target pattern, and obtain the image plane coordinates of the center positioning point;

S122: Identify and extract multiple edge positioning points of the target pattern, and obtain image plane coordinates of the multiple edge positioning points.

Preferably, in an embodiment, before S121, further includes: performing a binarization process on the target pattern, which can make the boundary of the target pattern more prominent, and the subsequent extraction of the center positioning point and the edge positioning point is more accurate.

Preferably, in an embodiment, the center positioning point in S11 is the center of the center positioning circle, and the edge positioning point is the center of the edge positioning circle; correspondingly,

S121 includes:

S1211: Use the boundary extraction method to perform edge detection on the target pattern, and extract the edge topology table with edge level information;

Wherein, the edge level information may be: the center positioning circle is the first level, and the edge positioning circle is the second level; or other level division forms may also be used;

S1212: Lock the boundary of the center positioning circle by searching the edge level information of the edge topology table;

S1213: Use least squares fitting to process the boundary of the center positioning circle to obtain the center coordinates of the center positioning circle, which are the image plane coordinates of the center positioning point;

S122 includes:

S1221: Use the boundary extraction method to perform ellipse detection on the target pattern, and lock the boundaries of multiple edge positioning points;

The target pattern itself is a circle, and the camera is facing the shooting target, the target will appear circular in the camera screen; but when the camera is obliquely shooting the target, the target pattern will appear in the camera screen. and ellipse, so using ellipse detection can make the detection structure more accurate;

S1222: Use least squares fitting to process the boundaries of the multiple edge positioning points to obtain the coordinates of the circle centers of the multiple edge positioning points, that is, the image plane coordinates of the multiple edge positioning points.

Preferably, in an embodiment, the center positioning circles in S111 include at least two, the at least two center positioning circles are concentric circles, and the radii of the at least two center positioning circles are different. Please refer to FIG. 2 , the center positioning circle takes three concentric circles as an example. In different embodiments, the number of center positioning circles may also be two, or more than three.

Correspondingly, S1213 is: adopting least squares fitting to process the boundary of the center positioning circle, obtaining the center coordinates of the center positioning circle, and performing averaging processing on the center coordinates of at least two or more center positioning circles to obtain the center positioning point's coordinates. Image coordinates.

In the two-dimensional displacement measurement scenario, the quantization transformation relationship from the image plane coordinates to the object plane coordinates in S13 is a homography matrix of the image plane coordinate system transformed to a two-dimensional object plane coordinate system in space.

In one embodiment, establishing a two-dimensional coordinate system of the object plane in S13, and then transforming the image plane image of the image to be measured into object plane coordinates includes:

S1311: Use each positioning point in the positioning point group to solve the perspective transformation matrix W in the two-dimensional coordinate system in which the image plane is mapped to the object plane;

S1312: Use the perspective transformation matrix to transform the image plane coordinates of the image to be measured into the two-dimensional coordinates of the object plane, and the transformation formula is:

Among them, x is the abscissa of the two-dimensional coordinate system of the object plane, y is the ordinate of the two-dimensional coordinate system of the object plane, s is the scale coefficient, P is the image plane point, and P^ is the object plane point.

In the three-dimensional displacement measurement scenario, the quantitative transformation relationship from the image plane coordinates to the object plane coordinates in S13 is the linear transformation relationship from the binocular camera group coordinate system to a three-dimensional object plane coordinate system in space. The transformation relationship can be determined by the corresponding It consists of translation matrix and rotation matrix.

In one embodiment, the target pattern in S11 includes: a left-view target pattern obtained by one of the cameras of the binocular camera group and a right-view target pattern obtained by another camera;

The image plane coordinates of each positioning point in the positioning point group obtained in S12 respectively include: the image plane coordinates of the left view and the image plane coordinates of the right view;

In S13, the three-dimensional coordinate system of the object plane is established, and then the image plane image of the image to be measured is transformed into the object plane coordinates, including:

S1321: According to the internal parameter calibration result of the binocular camera group, reconstruct the image plane coordinates of the left view and the image plane coordinates of the right view of each positioning point, and obtain the three-dimensional coordinate result P of each positioning point in the binocular camera group coordinate system _cam ;

S1322: Use the three-dimensional coordinate result P _cam of each positioning point in the binocular camera group coordinate system to solve the object plane equation of the plane where the target pattern is located;

S1323: The x and y axes are built on the object surface, and the z axis is built in the normal direction of the object surface, and the three-dimensional coordinate of the center positioning point P _c in the positioning point group is used as the origin of the coordinate system, and the three-dimensional coordinate system of the object surface is established;

S1324: Set the matrix formed by the three-axis normalized coordinates of the three-dimensional coordinate system of the object surface in the binocular camera group coordinate system as R, and use P _c as the translation vector to convert the image to be measured in the binocular camera group through coordinate transformation. The three-dimensional coordinate result P _cam in the coordinate system is converted into the coordinate result P _world in the three-dimensional coordinate system of the object surface, and the conversion formula is:

为P_c的三维坐标。in,

is the three-dimensional coordinate of _Pc .

The self-calibration method for the unique measurement based on machine vision in the above embodiment is described below with a specific example, but the target pattern, the number of each positioning point and each parameter are not limited to the following specific example.

Please refer to FIG. 2 , the target pattern in this example is a square, the center point positioning circle is three, and the edge positioning circle is four. In different embodiments, four edge positioning circles are not necessarily used as an example, nor are they necessarily the four positions in the figure, as long as the plane on which the target pattern is located can be located.

S121 includes:

S1211: Use the boundary extraction method to perform edge detection on the target pattern, and extract the edge topology table with edge level information; taking Figure 2 as an example, the edge detection obtains the edge shapes of three concentric circles and four corner points; the edge level The information can be: the innermost circle edge in the concentric circle is the first level, the secondary inner concentric circle edge is the second level, and the outermost concentric circle edge and the four corner edges are the third level;

S1212: By searching the edge level information of the edge topology table, lock the boundaries of the three center positioning circles;

S1213: adopt least squares fitting to process the boundaries of the three center positioning circles, obtain three elliptical circle center coordinates, and take the mean value as the image plane coordinates P _c of the center positioning point;

S122 includes:

S1221: Use the boundary extraction method to perform ellipse detection on the target pattern, and filter out all ellipse boundaries, because the target detection in the real situation is actually to detect the real target in a whole real shot image, then the ellipse detection will detect All ellipse feature boundaries in the real image shooting scene are obtained, so four corner points should be screened out. The specific screening method is: find the four ellipse boundaries closest to the center positioning point, that is, the ellipse of the four edge positioning points. boundary;

S1222: Process the elliptical boundaries of the four edge positioning points by using least squares fitting, and obtain the circle center coordinates of the four edge positioning points, which are respectively: P ₁ , P ₂ , P ₃ , and P ₄ .

In the two-dimensional displacement measurement scenario, S13 specifically includes:

S1311: Use P _c and five positioning points of P ₁ , P ₂ , P ₃ , and P ₄ to solve the perspective transformation matrix W that maps the image plane to the object plane;

S1312: For the pixel coordinate results of the image to be tested obtained in each test, the perspective transformation matrix W is used to transform the image plane coordinates into object plane coordinates, and the perspective transformation is implemented according to the following formula:

Among them, x is the abscissa of the two-dimensional coordinate system of the object plane, y is the ordinate of the two-dimensional coordinate system of the object plane, s is the scale coefficient, P is the image plane point, and P^ is the object plane point.

Figure 3 shows an object plane coordinate system obtained by self-calibration using the self-calibration method of the above example in a two-dimensional displacement measurement scene, wherein the x and y axes are the two axes of the coordinate system.

In the three-dimensional displacement measurement scenario, S13 specifically includes:

S1321: Using the coordinates of P _c and five positioning points P ₁ , P ₂ , P ₃ , and P ₄ respectively identified in the left and right views of the binocular camera, and reconstructing five positioning points according to the internal parameter calibration results of the binocular camera group The three-dimensional coordinate result P _cam of the point in the coordinate system of the binocular camera group;

S1322: Use the three-dimensional coordinate result P _cam of the five positioning points in the binocular camera group coordinate system to solve the object plane equation of the plane where the target pattern is located;

S1323: The x and y axes are built on the object surface, and the z axis is built in the normal direction of the object surface, and the three-dimensional coordinate of the center positioning point P _c in the positioning point group is used as the origin of the coordinate system, and the three-dimensional coordinate system of the object surface is established;

S1324: Set the matrix formed by the three-axis normalized coordinates of the three-dimensional coordinate system of the object surface in the binocular camera group coordinate system as R, and use P _c as the translation vector to convert the image to be measured in the binocular camera group through coordinate transformation. The three-dimensional coordinate result P _cam in the coordinate system is converted into the coordinate result P _world in the three-dimensional coordinate system of the object surface, and the conversion formula is:

为P_c的三维坐标。in,

is the three-dimensional coordinate of _Pc .

Figure 4 shows the object plane coordinate system obtained by self-calibration in the three-dimensional displacement measurement scene by using the self-calibration method of the above example, wherein the x, y, and z axes are the three axes of the coordinate system.

In an embodiment, an automatic displacement measurement system based on machine vision is also provided, which includes: a positioning point group calibration module, an image plane coordinate acquisition module, and an object plane coordinate system acquisition module; wherein,

The positioning point group calibration module is used to calibrate the positioning point group of the target pattern;

The image plane coordinate acquisition module is used to identify and extract the positioning point group of the target pattern, and obtain the image plane coordinates of each positioning point in the positioning point group;

The object plane coordinate system acquisition module is used to establish a two-dimensional or three-dimensional coordinate system of the object plane according to the image plane coordinates of each positioning point, and then transform the image plane coordinates of the image to be measured into object plane coordinates.

In one embodiment, the locating point group calibration module includes: a center locating point calibrating unit and an edge locating point calibrating unit; wherein,

The midpoint locating point calibration unit is used to calibrate the center of the target pattern to obtain the center locating point;

The edge locating point calibration unit is used for calibrating the edge of the target pattern to obtain a plurality of edge locating points.

In one embodiment, the image plane coordinate acquisition module includes: an edge detection unit, a boundary locking unit for the center positioning circle, an image plane coordinate obtaining unit for the center positioning point, a boundary locking unit for the edge positioning circle, and an image plane coordinate obtaining unit for the edge positioning point. unit; of which,

The positioning point group in the positioning point group calibration module includes: a center positioning point and an edge positioning point, the center positioning point is the center of the center positioning circle, and the edge positioning point is the center of the edge positioning circle;

The edge detection unit is used to detect the edge of the target pattern by using the edge extraction method, and extract the edge topology table with edge level information;

The boundary locking unit of the center positioning circle is used to lock the boundary of the center positioning circle by looking up the edge level information of the edge topology table;

The image plane coordinate obtaining unit of the center positioning point is used to process the boundary of the center positioning circle by using least squares fitting to obtain the center coordinates of the center positioning circle, which are the image plane coordinates of the center positioning point;

The boundary locking unit of the edge positioning circle is used to perform ellipse detection on the target pattern by using the boundary extraction method, and lock the boundaries of the plurality of edge positioning points;

The image plane coordinate obtaining unit of the edge positioning point is used to process the boundaries of the multiple edge positioning points by using least squares fitting, and obtain the center coordinates of the multiple edge positioning points, that is, the multiple edge positioning points. The image plane coordinates of the point.

In one embodiment, the image plane coordinate acquisition module further includes: a binarization processing unit for performing binarization processing on the target pattern, which can make the boundary of the target pattern more prominent and the subsequent edge detection results more accurate.

For the technologies used by the above-mentioned modules, reference may be made to the description of the target self-calibration method based on machine vision displacement measurement, which will not be repeated here.

The method and system in the above embodiments of the present invention have a self-calibration function, and can automatically construct a quantitative conversion relationship from the coordinates of the image plane to the coordinates of the real object under any camera shooting angle, and can use various machine vision methods to track the target. , so as to realize the automatic measurement of machine vision displacement without human intervention.

It should be noted that the steps in the method provided by the present invention can be implemented by using the corresponding modules, devices, units, etc. in the system, and those skilled in the art can refer to the technical solutions of the system to implement the method. The step flow, that is, the embodiments in the system can be understood as a preferred example for implementing the method, and details are not described here.

Those skilled in the art know that, in addition to implementing the system provided by the present invention and its respective devices in the form of pure computer-readable program codes, the system provided by the present invention and its respective devices can be made by logic gates, Switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers are used to achieve the same function. Therefore, the system and its various devices provided by the present invention can be regarded as a kind of hardware components, and the devices for realizing various functions included in the system can also be regarded as structures in the hardware components; The means for implementing various functions can be regarded as either a software module implementing a method or a structure within a hardware component.

In the description of this specification, reference to the terms "an embodiment", "an example", "a specific implementation process", "an example", etc. refers to the specific features described in conjunction with the embodiment or example, A structure, material, or feature is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Only preferred embodiments of the present invention are disclosed herein, and the present specification selects and specifically describes these embodiments to better explain the principles and practical applications of the present invention, rather than limiting the present invention. Any modifications and changes made by those skilled in the art within the scope of the description should fall within the protection scope of the present invention.

Claims (10)

1. A displacement automatic measurement method based on machine vision is characterized by comprising the following steps:

s11: calibrating a positioning point group of the target pattern;

s12: identifying and extracting a positioning point group of the target pattern to obtain image plane coordinates of each positioning point in the positioning point group;

s13: and establishing a two-dimensional coordinate system or a three-dimensional coordinate system of the object plane according to the image plane coordinates of the positioning points, and further transforming the image plane coordinates of the image to be detected into object plane coordinates.

2. The machine-vision-based automatic displacement measurement method of claim 1, wherein the S11 includes:

s111: calibrating the center of the target pattern to obtain a center positioning point;

s112: and calibrating the edge of the target pattern to obtain a plurality of edge positioning points.

3. The machine-vision-based automatic displacement measurement method of claim 2, wherein the S12 includes:

s121: identifying and extracting a central positioning point of the target pattern to obtain an image surface coordinate of the central positioning point;

s122: and identifying and extracting a plurality of edge positioning points of the target pattern to obtain image surface coordinates of the plurality of edge positioning points.

4. The machine-vision-based automatic displacement measurement method of claim 3, wherein the step S121 is preceded by the step of: and carrying out binarization processing on the target pattern.

5. The machine-vision-based automatic displacement measurement method of claim 3, wherein the center positioning point in the S11 is a center of a center positioning circle, and the edge positioning points are centers of edge positioning circles; in a corresponding manner, the first and second electrodes are,

the S121 includes:

s1211: performing edge detection on the target pattern by adopting a boundary extraction method, and extracting an edge topology table with edge grade information;

s1212: locking the boundary of the central positioning circle by searching the edge grade information of the edge topology table;

s1213: processing the boundary of the central positioning circle by adopting least square fitting to obtain the center coordinates of the central positioning circle, namely the image plane coordinates of the central positioning point;

the S122 includes:

s1221: carrying out ellipse detection on the target pattern by adopting a boundary extraction method, and locking the boundaries of the plurality of edge positioning points;

s1222: and processing the boundaries of the edge positioning points by adopting least square fitting to obtain circle center coordinates of the edge positioning points, namely image surface coordinates of the edge positioning points.

6. The machine-vision-based automatic displacement measuring method according to claim 5, wherein the center positioning circles in S111 include at least two, at least two of the center positioning circles are concentric circles, and at least two of the center positioning circles have different radii; in a corresponding manner, the first and second electrodes are,

the S1213 is: and processing the boundary of the central positioning circle by adopting least square fitting to obtain the center coordinates of the central positioning circle, and averaging the center coordinates of at least two central positioning circles to obtain the image plane coordinates of the central positioning point.

7. The machine-vision-based automatic displacement measuring method according to any one of claims 1 to 6, wherein the establishing of the two-dimensional coordinate system of the object plane in S13, and the transforming of the image plane image of the image to be measured into the object plane coordinates comprises:

s1311: solving a perspective transformation matrix under a two-dimensional coordinate system of which the image surface is mapped to the object surface by utilizing each positioning point in the positioning point group;

s1312: and transforming the image surface coordinate of the image to be measured into a two-dimensional coordinate of the object surface by using the perspective transformation matrix, wherein the transformation formula is as follows:

wherein x/s is the abscissa of the two-dimensional coordinate system of the object plane, y/s is the ordinate of the two-dimensional coordinate system of the object plane, s is a proportionality coefficient, W is a perspective transformation matrix, P is an image plane point homogeneous coordinate vector, and P ^ is an object plane point coordinate vector.

8. The machine-vision-based automatic displacement measurement method of any one of claims 1 to 6, wherein the target pattern in S11 comprises: a left view target pattern obtained by one camera of the binocular camera set and a right view target pattern obtained by the other camera;

the image plane coordinates of each positioning point in the positioning point group obtained in S12 include: the image plane coordinates of the left view and the image plane coordinates of the right view;

in S13, establishing a three-dimensional coordinate system of the object plane, and further transforming the image plane image of the image to be measured into the object plane coordinates includes:

s1321: reconstructing the image plane coordinates of the left view and the image plane coordinates of the right view of each positioning point according to the internal reference calibration result of the binocular camera set to obtain a three-dimensional coordinate result P of each positioning point in a coordinate system of the binocular camera set_cam；

S1322: utilizing the three-dimensional coordinate result P of each positioning point in the coordinate system of the binocular camera set_camObtaining an object plane equation of a plane where the target pattern is located through least square fitting;

s1323: building x and y axes in an object plane, building a z axis in a normal direction of the object plane, and locating a central locating point P in the locating point group_cThe three-dimensional coordinates of the three-dimensional coordinate system are used as the origin of the coordinate system to establishA three-dimensional coordinate system of the object plane;

s1324: setting a matrix formed by three-axis normalized coordinates of the three-dimensional coordinate system of the object plane in the coordinate system of the binocular camera set as R, and taking P as the reference value_cAs a translation vector, converting the three-dimensional coordinate result P of the image to be detected in the coordinate system of the binocular camera set through coordinate transformation_camConversion into coordinate results P in a three-dimensional coordinate system of the object plane_worldThe conversion formula is:

wherein,

is P_cThree-dimensional coordinates of (a).

9. An automatic displacement measuring system based on machine vision is characterized by comprising: the device comprises a positioning point group calibration module, an image plane coordinate acquisition module and an object plane coordinate system acquisition module; wherein,

the positioning point group calibration module is used for calibrating a positioning point group of the target pattern;

the image plane coordinate acquisition module is used for identifying and extracting a positioning point group of the target pattern to obtain image plane coordinates of each positioning point in the positioning point group;

the object plane coordinate system acquisition module is used for establishing a two-dimensional or three-dimensional coordinate system of the object plane according to the image plane coordinates of the positioning points, and further converting the image plane coordinates of the image to be detected into object plane coordinates.

10. The automatic displacement measuring system based on machine vision according to claim 9, wherein the image plane coordinate acquiring module comprises: the device comprises an edge detection unit, a boundary locking unit of a center positioning circle, an image plane coordinate obtaining unit of the center positioning point, a boundary locking unit of the edge positioning circle and an image plane coordinate obtaining unit of the edge positioning point; wherein,

the positioning point group in the positioning point group calibration module comprises: the positioning device comprises a central positioning point and an edge positioning point, wherein the central positioning point is the circle center of a central positioning circle, and the edge positioning point is the circle center of an edge positioning circle;

the edge detection unit is used for carrying out edge detection on the target pattern by adopting a boundary extraction method and extracting an edge topology table with edge grade information;

the boundary locking unit of the central positioning circle is used for locking the boundary of the central positioning circle by searching the edge grade information of the edge topology table;

the image plane coordinate obtaining unit of the central positioning point is used for processing the boundary of the central positioning circle by adopting least square fitting to obtain the center coordinate of the central positioning circle, namely the image plane coordinate of the central positioning point;

the boundary locking unit of the edge positioning circle is used for carrying out ellipse detection on the target pattern by adopting a boundary extraction method and locking the boundaries of the edge positioning points;

the image plane coordinate obtaining unit of the edge positioning points is used for processing the boundaries of the edge positioning points by adopting least square fitting to obtain circle center coordinates of the edge positioning points, namely the image plane coordinates of the edge positioning points.

Images