CN107462173B - Microscopic vision-based displacement measurement method and system for micro-movement platform - Google Patents

Abstract

The invention discloses a microscopic vision-based displacement measurement method and system for a micro-moving platform. The steps are as follows: image sequence acquisition: acquiring a set of image sequences through a micro-vision system; coarse matching position acquisition: using an improved particle swarm optimization algorithm to quickly perform a quick search in the entire search domain Search to obtain the rough matching position of the image block; obtain the best matching position: take the rough matching position as the center, use the full search block matching algorithm to search in a small neighborhood to obtain the best matching position; The imaging model of the vision system establishes the image Jacobian matrix, and converts the displacement corresponding to the best matching position in the image space into the actual displacement of the micro-movement platform. Compared with the existing micro-displacement measurement technology, the method has the advantages of low cost of measurement equipment, high precision and high availability. A micro-motion system with two degrees of freedom in the measurement plane.

Description

Microscopic vision-based displacement measurement method and system for micro-movement platform

technical field

The invention relates to a microscopic vision-based displacement measurement method and system for a micro-movement platform.

Background technique

Modern science and technology are rapidly developing into the micro and ultra-precision fields. The rise of micro and nanotechnology has triggered major changes in the fields of manufacturing, information, materials, biology, medical care and defense. At the same time, the development of micro-nano technology also puts forward higher requirements for ultra-precision measurement technology. The non-contact optical measurement method has been widely used in the micro-nano field due to its advantages of high measurement accuracy, fast response speed, and many degrees of measurement freedom. At present, researches at home and abroad mainly focus on computer microvision measurement, stroboscopic microscopic measurement, laser Dolphin measurement, microscopic interferometry and so on. Among them, the computer micro-vision measurement uses the micro-vision system to judge the displacement of the moving part in the micro-vision system by analyzing the motion vector in the micro-vision image. The system has low cost and can measure more degrees of freedom. However, in the image processing of microscopic vision, the common image block matching method or image feature matching is often used to estimate the displacement of the micro-movement platform, so the efficiency is low, and it is difficult to meet the fast response requirements of the measurement system.

SUMMARY OF THE INVENTION

In view of the above-mentioned prior art and the existing problems, the purpose of the present invention is to provide a microscopic vision-based micro-movement platform displacement measurement method and system, which can be directly applied to the micro-movement platform measurement and solve the technical problems in the current micro-nano control platform measurement. .

The present invention is achieved through the following technical solutions:

The microscopic vision-based displacement measurement method of the micro-movement platform, the steps are as follows:

Step (1): image sequence acquisition: collect a group of image sequences through the micro-vision system;

Step (2): coarse matching position acquisition: use the improved particle swarm optimization algorithm to quickly search in the entire search domain to obtain the coarse matching position of the image block;

Step (3): obtaining the best matching position: taking the rough matching position as the center, using the full-area search matching algorithm to search in a small neighborhood to obtain the best matching position;

Step (4): Calculation of displacement of the micro-movement platform: establish an image Jacobian matrix according to the imaging model of the micro-vision system, and convert the displacement corresponding to the best matching position in the image space into the actual displacement of the micro-movement platform.

In the step (1), the micro-vision system includes: a microscope, a CCD camera is installed on the top of the microscope, the CCD camera is connected to a computer terminal, and the XY two-degree-of-freedom micro-movement platform is fixed on the microscope stage, The XY two-degree-of-freedom micro-movement platform is driven by PZT control, and the upper surface of the XY two-degree-of-freedom micro-movement platform is affixed with a marker, and the surface of the marker is smooth; after the coaxial light is incident on the marker, it is reflected through the optical path of the microscope and is detected in the CCD camera. In the target plane imaging, the optical axis of the microscope is perpendicular to the upper surface of the marker, and at the same time, the optical axis of the microscope is also perpendicular to the target plane of the CCD camera, and the CCD target plane is parallel to the upper surface of the marker.

In the described step (2), obtaining the rough matching position includes:

Step (2-1): Determine the ROI area of the image, use the ROI area as the solution space for image block matching, and determine the ROI area according to the bit travel x _R of the XY two-degree-of-freedom micro-movement platform, microscope magnification k, camera pixel size p And the size of the image block [X, Y] is determined, and the minimum size of the selected ROI area is:

Step (2-2): Select the cumulative absolute difference (SAD) as the matching criterion. The cumulative absolute error is also the objective function of the particle swarm optimization algorithm. For each pixel in each image [x+u,y+ v] ^T , the corresponding fitness value fit is:

Among them, x is the abscissa of the image pixel, y is the ordinate of the image pixel, u is the movement amount of the image block in the horizontal axis direction, and v is the movement amount in the vertical axis direction.

Step (2-3): Particle Initialization: I initial particles are evenly distributed in the solution space, and each particle is calculated by formula (2) corresponding fitness value;

Step (2-4): put the i-th particle In the process of n iterations, the minimum fitness value in the fitness value [fit _i,1 ,fit _i,2 ,...fit _i,n ] is obtained as the local optimal solution pbest[i] of each particle:

pbest[i]=min{fit _i,1 ,fit _i,2 ,...fit _i,n };

Select the particle with the smallest fitness value of all particles in n iterations as the global optimal solution gbest[n], namely:

gbest[n]=min{pbest ₁ , pbest ₂ ,...pbest _I };

Each particle updates its position and velocity according to equation (3):

In the formula, n represents the nth iteration, C ₁ , C ₂ are learning factors, set to 2, R ₁ , R ₂ are random numbers, and R ₁ , R ₂ ∈ [0,1].

represents the velocity of the i-th particle during the n+1-th iteration,

represents the position of the i-th particle during the n+1-th iteration,

represents the velocity of the ith particle during the nth iteration,

Indicates the position of the i-th particle during the n-th iteration;

Step (2-5): Calculate the distance r _gbest,i between particle i and the particle of the current global optimal solution after the nth iteration:

Calculate the closest distance r _nearest,i between the particle i and the position passed by the entire particle swarm:

Among them, p∈[1,2,...I],q∈[1,2,...n-1];

After each particle updates its own position, the particle updates its own fitness value fit _{i,n according to the set rule, and the particle updates its own fitness value fit i,n according} to the set rule as follows:

a. If r _gbest,i < r ₀ , where r ₀ is the set threshold, the new particle i updates its fitness value according to formula (2);

b. If r _gbest,i >r ₀ and r _nearest,i >r ₀ , the new particle i updates its fitness value according to formula (2);

c. If r _gbest,i >r ₀ and r _nearest,i <r ₀ , replace the fitness value of the new particle with the fitness value of the new particle i from the closest point to the entire particle swarm, fit _i,n =SAD _nearest ;

Step (2-6): Repeat steps (2-4)-(2-5) until the maximum number of iterations is satisfied, and the coordinates corresponding to the calculated global optimal solution are used as the rough matching position

In the described step (3), for the rough matching position As the center, select a small search area, the size of the small search area is set to 7 × 7 pixels, search with the full search block matching algorithm in the small search area, and calculate the fitness value of 49 pixels in the small search area , select the point with the smallest fitness value as the final matching result

In the step (4), according to the step (1), the CCD target plane is parallel to the upper surface of the marker, and the imaging model is simplified to a pinhole model; the Jacobian matrix of the image space and the micro-movement platform is as follows:

where x _img , y _img are the image space coordinates, x ₀ , y ₀ are the position coordinates of the micro-movement platform, is a Jacobian matrix with constant parameters.

According to the final matching result obtained in step (3) Calculate the position corresponding to the micro-movement platform:

A microscopic vision-based micro-movement platform displacement measurement system includes: a memory, a processor, and computer instructions stored on the memory and executed on the processor, and the computer instructions complete the following steps when executed on the processor:

Step (1): image sequence acquisition: collect a group of image sequences through the micro-vision system;

Step (2): coarse matching position acquisition: use the improved particle swarm optimization algorithm to quickly search in the entire search domain to obtain the coarse matching position of the image block;

Step (3): obtaining the best matching position: taking the rough matching position as the center, using the full-area search matching algorithm to search in a small neighborhood to obtain the best matching position;

Step (4): Calculation of displacement of the micro-movement platform: establish an image Jacobian matrix according to the imaging model of the micro-vision system, and convert the displacement corresponding to the best matching position in the image space into the actual displacement of the micro-movement platform.

A computer-readable storage medium having computer instructions stored thereon, the computer instructions completing the following steps when executed on a processor:

Step (1): image sequence acquisition: collect a group of image sequences through the micro-vision system;

Step (2): coarse matching position acquisition: use the improved particle swarm optimization algorithm to quickly search in the entire search domain to obtain the coarse matching position of the image block;

Step (3): obtaining the best matching position: taking the rough matching position as the center, using the full-area search matching algorithm to search in a small neighborhood to obtain the best matching position;

Step (4): Calculation of displacement of the micro-movement platform: establish an image Jacobian matrix according to the imaging model of the micro-vision system, and convert the displacement corresponding to the best matching position in the image space into the actual displacement of the micro-movement platform.

The beneficial effects of the present invention are:

The invention combines the improved particle swarm algorithm and the full-area search algorithm, reduces the consumption of computing resources, and realizes fast matching and displacement measurement; at the same time, compared with the existing micro-displacement measurement technology, the method has the advantages of low cost of measurement equipment and high precision. High, can be used to measure the in-plane two degrees of freedom (X-Y) micro-motion system and other characteristics.

Description of drawings

The accompanying drawings that form a part of the present application are used to provide further understanding of the present application, and the schematic embodiments and descriptions of the present application are used to explain the present application and do not constitute improper limitations on the present application.

Fig. 1 is the hardware connection relation diagram of the present invention;

Figure 2 is a schematic diagram of the particle update fitness value rule in the improved particle swarm algorithm;

Figure 3 is a As the center, select a small search domain, and use the full search block matching algorithm to search for the exact solution;

4 is a schematic diagram of simplifying the imaging model into a pinhole model;

FIG. 5 is a flow chart of the method of the present invention.

Detailed ways

It should be noted that the following detailed description is exemplary and intended to provide further explanation of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

It should be noted that the terminology used herein is for the purpose of describing specific embodiments only, and is not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly dictates otherwise, the singular is intended to include the plural as well, furthermore, it is to be understood that when the terms "comprising" and/or "including" are used in this specification, it indicates that There are features, steps, operations, devices, components and/or combinations thereof.

As shown in Figure 5, the microscopic vision-based displacement measurement method of the micro-movement platform, the steps are as follows:

Step (1): image sequence acquisition: acquire a set of image sequences through a micro-vision system consisting of marked feature points, a micro-movement platform, a stereo microscope, and a CCD camera;

Step (2): obtaining the rough matching position: using the improved particle swarm optimization algorithm to quickly search in the entire search domain to obtain the rough matching position of the image block; as shown in Figure 2;

Step (3): obtaining the best matching position: taking the rough matching position as the center, using the full search block matching algorithm to search in a small neighborhood to obtain the best matching position;

Step (4): Calculation of displacement of the micro-movement platform: establish an image Jacobian matrix according to the imaging model of the micro-vision system, and convert the displacement corresponding to the best matching position in the image space into the actual displacement of the micro-movement platform.

In the step (1), the principle of image sequence acquisition is as follows: the principle of image acquisition is shown in Figure 1, the PZT drives the execution part of the XY two-degree-of-freedom micro-movement platform, and the surface of the marker is attached, and the surface of the marker is smooth. After the coaxial light is reflected by the marker, it is imaged on the CCD target plane through the optical path of the microscope. The optical axis of the microscope is perpendicular to the upper surface of the marker, and the optical axis of the microscope is also perpendicular to the camera target plane.

In the described step (2), obtaining the rough matching position includes:

Step (2-1) Determine the ROI area of the image, and use the entire ROI area as the solution space for image block matching. The entire ROI area can be determined according to the position stroke x _R of the micro-movement platform, the magnification k of the microscope, the size of the camera pixel p and the image block. The size [X, Y] is determined, and the minimum selected ROI area is:

Step (2-2) Select the cumulative absolute difference (summed absolute difference, SAD) as the matching criterion, which is also the objective function of the particle swarm optimization algorithm, for each pixel in each image [x+u,y+v] ^T , and its corresponding fitness value is:

Step (2-3) particle initialization: distribute I initial particles uniformly in the entire solution space, each particle Use the formula (2) to calculate its corresponding fitness value.

Step (2-4) i-th particle particle The smallest one of the fitness values [fit _i,1 ,fit _i,2 ,...fit _i,n ] in the n iterations process is used as the local optimal solution of each particle, that is, pbest[i]= min{fit _i,1 ,fit _i,2 ,...fit _i,n }. The particle with the smallest fitness value of all particles in n iterations is selected as the global optimal solution, that is,

gbest[n]=min{pbest ₁ , pbest ₂ , . . . pbest _I }.

Each particle updates its position and velocity as follows:

where n represents the nth iteration, C ₁ , C ₂ are learning factors, set to 2, R ₁ , R ₂ are random numbers, and R ₁ , R ₂ ∈ [0,1].

Step (2-5) Calculate the distance between particle i and the particle of the current global optimal solution after the nth iteration Calculate the closest distance between particle i and the position passed by the entire particle swarm, namely where p∈[1,2,...I],q∈[1,2,...n-1]. After each particle updates its own position, the particle updates its own fitness value fit _i,n according to the following rules, as follows:

ar _gbest,i <r ₀ , where r ₀ is the threshold, and the new particle i updates its fitness value according to formula (2);

br _gbest,i >r ₀ and r _nearest,i >r ₀ , the new particle i updates its fitness value according to formula (2);

cr _gbest,i >r ₀ and r _nearest,i <r ₀ , replace the fitness value of the new particle with the fitness value of the new particle i from the closest point to the position passed by the entire particle swarm, namely fit _i,n =SAD _nearest .

Among them, r ₀ is a set threshold, which can be determined by an empirical formula according to the search window and the number of particles.

Step (2-6) Repeat steps (2-4), (2-5) until the maximum number of iterations is satisfied, and the coordinates corresponding to the global optimal solution calculated by the above method are used as the rough matching position

In the described step (3), for In the center, select a small search field and set its size to 7×7pixel, as shown in Figure 3. Search the entire area with the full search block matching algorithm, calculate the fitness value of 49 pixels in the area, and select the point with the smallest fitness value as the final matching result

In the step (4), according to the step (1), the CCD target plane is parallel to the upper surface of the marker, so the imaging model can be simplified to a pinhole model, as shown in FIG. 4 . According to the illustrated geometric relationship, the Jacobian matrix of the image space and the micro-movement platform can be derived as follows:

where x _img , y _img are the image space coordinates, x ₀ , y ₀ are the position coordinates of the micro-movement platform, is a Jacobian matrix with constant parameters. According to the requirements in step (3), the position corresponding to the micro-movement platform can be deduced:

The above descriptions are only preferred embodiments of the present application, and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included within the protection scope of this application.

Claims (8)

1. the micromotion platform displacement measurement method based on micro-vision, characterized in that steps are as follows:

Step (1): one group of image sequence image sequence acquisition: is acquired by micro- vision system；

Step (2): thick matching position obtains: utilizing improved particle swarm optimization algorithm fast search in entire region of search, obtains Obtain the thick matching position of image block；

Step (3): best match position obtains: again centered on thick matching position, using area coverage matching algorithm small Search in neighborhood, obtains best match position；

Step (4): micromotion platform displacement calculates: image Jacobin matrix is established according to micro- vision system imaging model, by image The corresponding displacement of best match position is converted to micromotion platform actual displacement in space；

In the step (2), thick matching position acquisition includes:

Step (2-1): determining image ROI region, and using ROI region as the matched solution space of image block, ROI region determines basis The position stroke x of XY two degrees of freedom micromotion platform_R, microscope magnification k, camera pixel dimension p and image block size size [X, Y] is determined, the ROI region size of selection is minimum are as follows:

Step (2-2): select accumulative absolute error as matching criterior, accumulative absolute error is also the mesh of particle swarm optimization algorithm Scalar functions, to pixel [x+u, y+v] each in each image^T, corresponding adaptive value fit are as follows:

Wherein, x is image slices vegetarian refreshments abscissa, and y is image slices vegetarian refreshments ordinate, and u is image block X direction amount of exercise, and v is Y direction amount of exercise；f_j(x, y) represents the position corresponding to each pixel in j moment each image；

Step (2-3): particle initialization: being evenly distributed on solution space for I primary, calculates each particle using formula (2)Corresponding adaptive value；

Step (2-4): by i-th of particleAdaptive value [fit is obtained in n times iterative process_i,1,fit_i,2, ...fit_i,n] locally optimal solution pbest [i] of the smallest adaptive value in the inside as each particle:

Pbest [i]=min { fit_i,1,fit_i,2,...fit_i,n}；

Select all particles in n times iterative process the smallest particle of adaptive value as globally optimal solution gbest [n], it may be assumed that

Gbest [n]=min { pbest₁,pbest₂,...pbest_I}；

Each particle updates the position and speed of oneself with formula (3):

In formula, n indicates nth iteration, C₁,C₂For Studying factors, it is set as 2, R₁,R₂For random number, R₁,R₂∈[0,1]；

Indicate speed of i-th of particle in (n+1)th iterative process,

Indicate position of i-th of particle in (n+1)th iterative process,

Indicate speed of i-th of particle during nth iteration,

Indicate position of i-th of particle during nth iteration；

Step (2-5): after calculating nth iteration, the distance between particle of particle i and current globally optimal solution r_gbest,i:

Calculate particle i and entire population by position minimum distance r_nearest,i:

Wherein, p ∈ [1,2 ... I], q ∈ [1,2 ... n-1]；

After each particle updates oneself position, particle is according to the adaptive value fit for setting Policy Updates oneself_i,n；

Step (2-6): step (2-4)-(2-5) is repeated, until meeting maximum number of iterations, and calculated globally optimal solution pair The coordinate answered is as thick matching position

The particle is according to the adaptive value fit for setting Policy Updates oneself_i,nIt is as follows:

A. if r_gbest,i<r₀, r in formula₀For given threshold, then new particle i updates its adaptive value according to formula (2)；

B. if r_gbest,i>r₀And r_nearest,i>r₀, then new particle i updates its adaptive value according to formula (2)；

C. if r_gbest,i>r₀And r_nearest,i<r₀, then with new particle i apart from entire population institute by position closest approach fitting The adaptive value instead of new particle, fit should be worth_i,n=SAD_nearest。

2. the micromotion platform displacement measurement method based on micro-vision as described in claim 1, characterized in that the step (1) in, micro- vision system, comprising: CCD camera is installed at microscope, the microscope top, and the CCD camera and computer are whole End connection, XY two degrees of freedom micromotion platform are fixed on microscope carrier, and the XY two degrees of freedom micromotion platform is controlled by PZT Marker is posted in device drive control processed, the XY two degrees of freedom micromotion platform upper surface, and marker surface is smooth；Axis light is incident Be imaged to marker back reflection by microscopes optical path in CCD camera target plane, microscopes optical axis perpendicular to marker upper surface, Meanwhile microscopes optical axis is also perpendicularly to CCD camera target plane, CCD target plane is parallel to marker upper surface.

3. the micromotion platform displacement measurement method based on micro-vision as described in claim 1, characterized in that the step (3) in, with thick matching positionCentered on, a small region of search is selected, the size of the small region of search is set as 7 × 7 pixels calculate in small region of search 49 pixels with the search of area coverage matching algorithm in small region of search Adaptive value selects wherein adaptive value is the smallest to put as final matching results

4. the micromotion platform displacement measurement method based on micro-vision as claimed in claim 3, characterized in that the step (4) in, according to step (1), CCD target plane is parallel to marker upper surface, and imaging model is reduced to pin-hole model； The Jacobin matrix of image space and micromotion platform is as follows:

X in formula_img, y_imgFor image space coordinate, x₀, y₀For micromotion platform position coordinates,Be a parameter be constant Jacobin matrix.

5. the micromotion platform displacement measurement method based on micro-vision as claimed in claim 4, characterized in that

According to final matching results required in step (3)Calculate position corresponding to micromotion platform:

6. the micromotion platform displacement measurement system based on micro-vision, characterized in that include: memory, processor and be stored in The computer instruction run on memory and on a processor completes such as right when the computer instruction is run on a processor It is required that the step of micromotion platform displacement measurement method described in 1 based on micro-vision:

Step (1): one group of image sequence image sequence acquisition: is acquired by micro- vision system；

Step (2): thick matching position obtains: utilizing improved particle swarm optimization algorithm fast search in entire region of search, obtains Obtain the thick matching position of image block；

Step (3): best match position obtains: again centered on thick matching position, using area coverage matching algorithm small Search in neighborhood, obtains best match position；

Step (4): micromotion platform displacement calculates: image Jacobin matrix is established according to micro- vision system imaging model, by image The corresponding displacement of best match position is converted to micromotion platform actual displacement in space.

7. the micromotion platform displacement measurement system based on micro-vision as claimed in claim 6, characterized in that the step (1) in, micro- vision system, comprising: CCD camera is installed at microscope, the microscope top, and the CCD camera and computer are whole End connection, XY two degrees of freedom micromotion platform are fixed on microscope carrier, and the XY two degrees of freedom micromotion platform is driven by PZT Marker is posted in dynamic control, the XY two degrees of freedom micromotion platform upper surface, and marker surface is smooth；Axis light is incident to label Object back reflection by microscopes optical path CCD camera target plane be imaged, microscopes optical axis perpendicular to marker upper surface, meanwhile, Microscopes optical axis is also perpendicularly to CCD camera target plane, and CCD target plane is parallel to marker upper surface.

8. a kind of computer readable storage medium, is stored thereon with computer instruction, characterized in that the computer instruction is being located Manage the step of micromotion platform displacement measurement method based on micro-vision as described in claim 1 is completed when running on device:

Step (1): one group of image sequence image sequence acquisition: is acquired by micro- vision system；

Step (2): thick matching position obtains: utilizing improved particle swarm optimization algorithm fast search in entire region of search, obtains Obtain the thick matching position of image block；

Step (3): best match position obtains: again centered on thick matching position, using area coverage matching algorithm small Search in neighborhood, obtains best match position；

Step (4): micromotion platform displacement calculates: image Jacobin matrix is established according to micro- vision system imaging model, by image The corresponding displacement of best match position is converted to micromotion platform actual displacement in space.