Three-dimensional microscale velocity field measurement method, device and storage medium

Abstract

The present invention discloses a three-dimensional microscale velocity field measurement method, device, and storage medium. The three-dimensional microscale velocity field measurement method includes: obtaining a particle light field image under light field microscopy imaging; calculating a reconstructed particle distribution of the obtained particle light field image within a measurement volume; obtaining a particle center distribution of the reconstructed particle distribution; calculating a low-resolution three-dimensional displacement field within the measurement volume using a cross-correlation algorithm based on the reconstructed particle distribution; using the calculated low-resolution three-dimensional displacement field as the predicted displacement, performing motion compensation on the obtained particle center distribution to obtain a shifted particle center distribution; and performing cross-validation matching on the obtained shifted particle center distribution to obtain a high-resolution three-dimensional microscale velocity field. Based on the low-resolution three-dimensional displacement field obtained by the cross-correlation algorithm, the present invention uses a cross-validation matching algorithm to match two frames of particles. This method can accurately track individual particles at high concentrations and effectively improve the resolution of the three-dimensional microscale velocity field.

Description

Technical Field

[0001] The invention relates to a high-resolution three-dimensional microscale velocity field measurement method based on light field microscopy imaging, and belongs to the technical field of microscale fluid measurement. Background Art

[0002] With the development of microfluidic chips, the microflows within them have become increasingly complex, with the emergence of complex three-dimensional flow phenomena such as vortex flows and droplet flows. Accurately acquiring high-resolution three-dimensional microscale velocity fields of vortex flows and droplets can help provide the most intuitive understanding of microflows, further enhance the operation and control of microflows, and improve the performance of microfluidic chips. Light-field microparticle imaging velocimetry (LF-μPIV) and light-field microparticle tracking velocimetry (LF-μPTV) can achieve three-dimensional microscale velocity field measurements based on a single camera. They have the advantages of simple experimental systems and high temporal resolution, and have become a hot topic in three-dimensional microscale velocity field measurements in recent years.

[0003] LF-μPIV utilizes a cross-correlation algorithm to measure three-dimensional microscale velocity fields. This algorithm divides the measurement volume into multiple discrimination windows and calculates the velocity vector for each window. To ensure the reliability of the velocity vectors, the cross-correlation algorithm requires at least four particles within the discrimination window; that is, multiple particle pairs are required to generate a single velocity vector. Consequently, the number of velocity vectors generated by LF-μPIV is far smaller than the number of particles within the measurement volume, resulting in limited velocity field resolution.

[0004] LF-μPTV can use the nearest neighbor method to match particles, thereby achieving single-particle tracking. The upper limit of the velocity field resolution of LF-μPTV depends on the number of particles in the measurement volume. Increasing the particle concentration can increase the particle number, thereby raising the upper limit of LF-μPTV's resolution. However, increasing the particle concentration reduces the interparticle spacing, thereby increasing the ratio of particle displacement to interparticle spacing, resulting in an increased probability of mismatching using the nearest neighbor method and reduced velocity field measurement accuracy. Therefore, the nearest neighbor method is only applicable to situations with extremely low particle concentrations and has difficulty achieving high-resolution velocity field measurements. Therefore, it is necessary to propose a high-resolution three-dimensional microscale velocity field measurement method based on light field microscopy to accurately achieve dual-frame particle matching at high particle concentrations and improve velocity field resolution. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a high-resolution three-dimensional microscale velocity field measurement method, device and storage medium based on light field microscopy to improve the resolution of the three-dimensional microscale velocity field.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A high-resolution three-dimensional microscale velocity field measurement method based on light field microscopy includes the following steps:

[0008] A high-resolution three-dimensional microscale velocity field measurement method based on light field microscopy is characterized by comprising:

[0009] Acquire light field images of particles under light field microscopy;

[0010] Calculate and reconstruct the particle distribution in the measurement volume from the acquired particle light field image;

[0011] Obtaining particle center distribution for reconstructing particle distribution;

[0012] Based on the reconstructed particle distribution, a cross-correlation algorithm is used to calculate the low-resolution three-dimensional displacement field in the measurement volume;

[0013] The calculated low-resolution three-dimensional displacement field is used as the predicted displacement, and the obtained particle center distribution is motion compensated to obtain the shifted particle center distribution;

[0014] The obtained displaced particle center distribution is cross-validated and matched to obtain a high-resolution three-dimensional microscale velocity field. The core of this invention is to use the low-resolution three-dimensional displacement field obtained by the cross-correlation algorithm as the predicted displacement to perform motion compensation on the particle center distribution; on this basis, a cross-validation matching algorithm is used to perform particle matching, achieve single particle tracking at high particle concentrations, and obtain a high-resolution three-dimensional velocity field.

[0015] Beneficial effects

[0016] Compared with light field micro-particle imaging velocimetry and light field micro-particle tracking velocimetry, the present invention has the following advantages:

[0017] (1) Using the calculated low-resolution three-dimensional displacement field as the predicted displacement to perform motion compensation on the particle center distribution can significantly reduce the particle displacement during matching, lower the ratio of particle displacement to particle spacing, and improve the accuracy of particle matching;

[0018] (2) The cross-validation matching algorithm adds backward matching (taking the particles in the first frame as the target and searching for matching particles in the second frame) to the forward particle matching (taking the particles in the second frame as the target and searching for matching particles in the first frame). The forward and backward matching results are cross-validated, which can reduce the probability of particle mismatching and further improve the accuracy of particle matching.

[0019] (3) Based on cross-correlation calculation, the present invention uses a cross-validation matching algorithm to match particles in two frames, which can accurately track single particles at high concentrations and effectively improve the resolution of the three-dimensional microscale velocity field. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Flowchart of the high-resolution three-dimensional microscale velocity field measurement method based on light-field microscopy;

[0021] Figure 2 Schematic diagram of the light field microscopy velocity field measurement device;

[0022] Figure 3 Particle light field images;

[0023] Figure 4 Reconstruct particle distribution results;

[0024] Figure 5 Particle center distribution results;

[0025] Figure 6 Three-dimensional microscale velocity field results; among them, (a) the three-dimensional microscale velocity field of the present invention, (b) the three-dimensional microscale velocity field of light field microparticle imaging velocimetry technology, and (c) the three-dimensional microscale velocity field of light field microparticle tracking velocimetry technology. DETAILED DESCRIPTION

[0026] The present invention will be further described below in conjunction with the accompanying drawings and specific examples. It should be understood that these embodiments are only intended to illustrate the present invention and are not intended to limit the scope of the present invention. After reading this invention, modifications to various equivalent forms of the present invention made by those skilled in the art all fall within the scope defined by the appended claims.

[0027] A high-resolution three-dimensional microscale velocity field measurement method based on light field microscopy includes the following steps:

[0028] Step 1: Use a light field microscopy velocity field measurement device to obtain a particle light field image:

[0029] Schematic diagram of light field microscopy velocity field measurement device Figure 2 As shown. The aqueous solution containing tracer particles is injected into the microchannel through a syringe pump; the laser light emitted by the laser passes through the light guide arm and the dichroic mirror in the inverted fluorescence microscope and illuminates the microchannel; the tracer particles in the microchannel produce fluorescence after being illuminated by the laser, and then pass through the dichroic mirror and microlens array in the inverted fluorescence microscope to reach the camera, forming a particle light field image, as shown. Figure 3 As shown; the synchronization controller is used to control the double-pulse time interval of the laser so that the two laser pulses fall on the double exposure time of the camera respectively.

[0030] Step 2: Use the deconvolution algorithm to calculate the reconstructed particle distribution of the particle light field image in the measurement volume:

[0031] The particle light field image obtained in step 1 is iteratively reconstructed using the deconvolution algorithm:

[0032] Fi+1 ={(G⊕H t ) / [(F i ⊕H)⊕H t ]}F i

[0033] Where, F i and F i+1 They represent the iterative results of the i-th and i+1-th reconstructed particle distributions, G is the particle light field image, H is the point spread function, and H t is the transpose of H, and the symbol “⊕” represents a mixed operation of convolution and addition. The point spread function calculated based on wave optics theory and system optical parameters is:

[0034]

[0035] Where h(x,y,z) represents the value of the point spread function H when the source coordinates are (x,y,z), u is the imaginary unit, ξ and η are the spatial frequencies of the source coordinates x and y, respectively; F[·] and F -1 {·} represents Fourier transform and inverse Fourier transform, T(x,y) is the transmittance function of the microlens array, f l and λ are the focal length of the microscope objective and the fluorescence wavelength, and U(r,s) is the wavefront intensity distribution of the point source on the microscope imaging plane calculated by wave optics theory, expressed as:

[0036]

[0037]

[0038]

[0039] Where r and s represent the lateral and axial optical coordinates of the image plane, M is the magnification of the microscope objective, sin(α) = NA, α is half the angular aperture of the objective, NA is the numerical aperture of the objective, ρ is the normalized radial coordinate of the objective aperture, P(ρ) is the normalized pupil function, J0(·) is the zero-order Bessel function, and x1 and y1 are the coordinates of the point spread function on the image plane.

[0040] The reconstructed particle distribution obtained by the deconvolution algorithm is as follows: Figure 4 shown.

[0041] Step 3: Perform Gaussian fitting on the reconstructed particle distribution to obtain the particle center distribution:

[0042] After obtaining the reconstructed particle distribution in step 2, Gaussian fitting is performed on the light intensity of each reconstructed particle. The fitting result is:

[0043]

[0044] Where I(x,y,z) is the particle light intensity at position (x,y,z), (x p ,y p ,z p ) is the peak position of the Gaussian fit, I max is the maximum value of the Gaussian function, σ x , σ y and σ z are the standard deviations of the Gaussian function along the x, y, and z directions respectively. The peak position of the Gaussian fit (x p ,y p ,z p ) is the particle center. After traversing all reconstructed particles, the particle center distribution can be obtained, such as Figure 5 shown.

[0045] Step 4: Based on the reconstructed particle distribution, the cross-correlation algorithm is used to calculate the low-resolution three-dimensional displacement field in the measurement volume:

[0046] Establish multiple discrimination windows with sizes of I, J, and K along the X, Y, and Z axes of the measurement volume, and perform cross-correlation calculation on each discrimination window:

[0047]

[0048] Where E1 and E2 represent the intensity distribution of the discriminant window in the first and second particle distribution frames, respectively; (I, J, K) represents the total number of voxels in the discriminant window in the x, y, and z directions; (i, j, k) is the control variable; and (l, m, n) represents the three-dimensional displacement in voxels. The (l, m, n) corresponding to the maximum cross-correlation coefficient is the three-dimensional displacement of the window.

[0049] Step 5: Using the low-resolution three-dimensional displacement field obtained in step 4 as the predicted displacement, motion compensation is performed on the particle center distribution in step 3 to obtain the shifted particle center distribution:

[0050] Since different discrimination windows have different cross-correlation results, the motion compensation value needs to be determined based on the original position of the particle center, and then the particles in the first frame are moved to the new position. In the particle center distribution of the first frame, the motion compensation process of a single particle can be expressed as:

[0051]

[0052] In the formula, (x ps ,y ps ,z ps ) is the center position of the particle after motion compensation, and the symbol “[·]” represents a matrix. After applying motion compensation to all particle centers, the distribution of shifted particle centers in the first frame is obtained.

[0053] Step 6: Cross-validate and match the center distribution of the displaced particles, achieve single particle tracking at high particle concentration, and obtain a high-resolution three-dimensional velocity field:

[0054] After motion compensation is performed in step five, a cross-validation matching algorithm is used to match particles between the first frame's shifted particle center distribution and the second frame's particle center distribution. First, forward matching is performed on the two frames' particle distributions: a single particle in the first frame's shifted particle center distribution is used as the target particle, and the particle closest to the target particle is found in the second frame's particle center distribution as a matching pair. Then, backward matching is performed: a single particle in the second frame's particle center distribution is used as the target particle, and the particle closest to the target particle is found in the first frame's shifted particle center distribution as a matching pair. Finally, cross-validation is performed on the two matching pairs, i.e.

[0055] M c =M f ∩M b

[0056] Where M c , M f and M b They represent cross-validation matching pairs, forward matching pairs, and backward matching pairs respectively, and “∩” represents the intersection operation.

[0057] After obtaining the particle matching pairs of the two frames, the velocity of each particle can be obtained by substituting them into the particle center distribution of the first frame and the particle center distribution of the second frame:

[0058]

[0059] Where, v x 、v y 、v z is the velocity of the particle on the X-axis, Y-axis and Z-axis, c 1,x 、c 1,y 、c 1,z and c 2,x 、c 2,y 、c 2,z They represent the center positions of the particle matching pairs in the first frame particle center distribution and the second frame particle center distribution respectively. The final three-dimensional microscale velocity field is as follows: Figure 6 As shown in (a).

[0060] Based on the same light field image as the above embodiment of the present invention, the three-dimensional microscale velocity field is obtained by using the light field microparticle imaging velocimetry (LF-μPIV) and light field microparticle tracking velocimetry (LF-μPTV) technology. The three-dimensional microscale velocity field finally obtained by the method of the present invention is as follows: Figure 6 (a) with Figure 6(b) Compared with the three-dimensional microscale velocity field of light field microparticle imaging velocimetry (LF-μPIV), the three-dimensional microscale velocity field obtained by the present invention has 596 velocity vectors, which is higher than the 245 velocity vectors in the three-dimensional microscale velocity field of LF-μPIV; the three-dimensional microscale velocity field finally obtained by the method of the present invention is as follows Figure 6 (a) with Figure 6 (c) Compared to the three-dimensional microscale velocity field obtained using light-field microparticle tracking velocimetry (LF-μPTV), the 3D microscale velocity field obtained by the present invention has a root mean square error of 0.0024 m / s, which is lower than the 0.0138 m / s root mean square error of the 3D microscale velocity field obtained using LF-μPTV. Therefore, the high-resolution 3D microscale velocity field measurement method based on light-field microscopy proposed in this invention can accurately achieve dual-frame particle matching at high particle concentrations, enabling accurate high-resolution measurement of the 3D microscale velocity field.

Claims

1. A high-resolution three-dimensional microscale velocity field measurement method based on light field microscopy, characterized in that: include: Acquire light field images of particles under light field microscopy; Calculate and reconstruct the particle distribution in the measurement volume from the acquired particle light field image; Obtaining particle center distribution for reconstructing particle distribution; Based on the reconstructed particle distribution, the low-resolution three-dimensional displacement field in the measurement body is calculated; The calculated low-resolution three-dimensional displacement field is used as the predicted displacement, and the obtained particle center distribution is motion compensated to obtain the shifted particle center distribution; The obtained distribution of displaced particle centers is cross-validated and matched to obtain a high-resolution three-dimensional microscale velocity field; In the step of calculating a low-resolution three-dimensional displacement field in the measurement volume by using a cross-correlation algorithm based on the reconstructed particle distribution, a low-resolution three-dimensional displacement field in the measurement volume is calculated by using a cross-correlation algorithm; The steps of calculating and measuring a low-resolution three-dimensional displacement field in a body using a cross-correlation algorithm include: Establish multiple discrimination windows with sizes of I, J, and K along the X, Y, and Z axes of the measurement volume, and perform cross-correlation calculation on each discrimination window: Where E1 and E2 represent the light intensity distribution of the discrimination window in the first and second frames of particle distribution, respectively; (I, J, K) represents the total number of voxels in the x, y, and z directions within the discrimination window; (i, j, k) is the control variable; and (l, m, n) represents the three-dimensional displacement in voxels. The (l, m, n) corresponding to the maximum value of the mutual correlation coefficient R(l, m, n) of each discrimination window is determined as the three-dimensional displacement of the window; Based on the three-dimensional displacement of each discrimination window, a low-resolution three-dimensional displacement field in the measurement body is obtained; The calculated low-resolution three-dimensional displacement field is used as the predicted displacement, and the obtained particle center distribution is subjected to motion compensation to obtain the shifted particle center distribution. The steps include: Perform motion compensation on each particle in the particle center distribution of the first frame. The motion compensation process of a single particle is expressed as: In the formula, (x ps ,y ps ,z ps ) is the center position of the particle after motion compensation, and the symbol "[·]" represents a matrix; The obtained distribution of displaced particle centers is cross-validated and matched to obtain a high-resolution three-dimensional microscale velocity field. The steps include: Particle matching is performed on the first frame's shifted particle center distribution and the second frame's particle center distribution using a cross-validation matching algorithm: First, forward matching is performed on the particle distributions of the two frames: a single particle in the first frame's shifted particle center distribution is used as the target particle, and the particle closest to the target particle in the second frame's particle center distribution is found as a matching pair; Then, backward matching is performed: a single particle in the particle center distribution of the second frame is used as the target particle, and the particle closest to the target particle in the shifted particle center distribution of the first frame is found as a matching pair; Finally, cross validation is performed on the two matching pairs, namely: M c =M f ∩M b Where M c represents the cross-validation matching pair, M f Represents the forward matching pair, M b Indicates backward matching pairs, "∩" is the intersection operation; After obtaining the particle matching pairs of the two frames, substitute them into the particle center distribution of the first frame and the particle center distribution of the second frame to obtain the velocity of each particle: Where, v x 、v y 、v z is the velocity of the particle on the X-axis, Y-axis and Z-axis, c 1,x 、c 1,y 、c 1,z and c 2,x 、c 2,y 、c 2,z They represent the center positions of the particle matching pairs in the first frame particle center distribution and the second frame particle center distribution, respectively, and Δt is the time interval between the two frames; All matching particle pairs are traversed to obtain a high-resolution three-dimensional microscale velocity field.

2. The high-resolution three-dimensional microscale velocity field measurement method according to claim 1, characterized in that: In the step of reconstructing the particle distribution in the measurement volume, the particle light field image obtained by calculation uses the deconvolution algorithm to calculate the reconstructed particle distribution: Where, F i and F i+1 They represent the i-th and i+1-th iteration results of the reconstructed particle distribution respectively; G is the particle light field image; H is the point spread function; H t is the transpose of H; symbol Represents a mixed operation of convolution and addition.

3. The high-resolution three-dimensional microscale velocity field measurement method according to claim 1, wherein the step of obtaining the particle center distribution for reconstructing the particle distribution comprises: Gaussian fitting is performed on each particle of the reconstructed particle distribution, and the fitting result is: Where I(x,y,z) is the particle light intensity at position (x,y,z), (x p ,y p ,z p ) is the peak position of the Gaussian fit, I max is the maximum value of the Gaussian function, σ x , σ y and σ z are the standard deviations of the Gaussian functions along the x, y, and z directions, respectively; The peak position (x p ,y p ,z p ) is the particle center, and the particle center distribution is obtained after traversing all reconstructed particles.

4. A high-resolution three-dimensional microscale velocity field measurement device based on light field microscopy, characterized in that: include: Image acquisition module, used to obtain particle light field images; The image processing module processes the particle light field image acquired by the image acquisition module according to the high-resolution three-dimensional microscale velocity field measurement method according to any one of claims 1 to 3 to obtain a high-resolution three-dimensional microscale velocity field.

5. The high-resolution three-dimensional microscale velocity field measurement device based on light field microscopy according to claim 4 is characterized in that: The image acquisition module is a light field microscopy imaging velocity field measurement device.

6. A computer-readable storage medium storing a program or instruction, wherein the program or instruction, when executed by a processor, implements the steps of the high-resolution three-dimensional microscale velocity field measurement method according to any one of claims 1 to 3.

Images