Flow field measurement method and system based on multi-color laser synthesis of new spectrum and light field imaging

Abstract

The invention discloses a flow field measurement method based on multi-color laser synthesis of a novel spectrum and light field imaging, comprising the following steps: step 1, building an experimental optical path for multi-color laser synthesis, using three lasers emitting RGB three colors as excitation sources, and synthesizing a novel laser spectrum whose color changes along the depth direction; step 2, performing a calibration experiment, and extracting the axial depth point spread function of the novel laser spectrum; step 3, exciting a flow field to be measured with the novel laser spectrum, and imaging the flow field to be measured with a light field camera, and collecting a light field image; step 4, performing three-dimensional light field reconstruction on the light field image, and obtaining a three-dimensional particle field of the light field with color coding information; step 5, performing color / depth decoding on the three-dimensional particle field of the light field with color coding information, and obtaining a three-dimensional distribution image of particles at different depths; step 6, reconstructing a three-dimensional velocity field, and using a variational optical flow model based on a physical model to solve and obtain a three-dimensional velocity vector field of the flow field to be measured.

Description

Technical Field

[0001] The present invention relates to the field of optical measurement technology, and in particular to a flow field measurement method and system based on multi-color laser synthesis of a novel spectrum and light field imaging. Background Art

[0002] Fluid imaging has many important applications in science and engineering. Three-dimensional unsteady flow and turbulence are very common, so the main task of fluid imaging is to detect fluid motion on a series of length scales. The ultimate goal is to obtain 3D dense measurements of the three components of the velocity vector, namely 3D-3C. Since the 1980s, two-dimensional particle image velocimetry technology has emerged and gradually expanded its application range. This technology can realize multi-point non-contact transient measurement in a two-dimensional plane. Therefore, this technology is widely used in many fields such as fluid mechanics, combustion and biomedicine. However, many flow phenomena in nature and actual engineering problems are complex three-dimensional flows. The two-dimensional velocity field is not enough to fully reveal the mechanism of three-dimensional flow phenomena. Therefore, it is necessary to study how to perform accurate 3D-3C velocity field measurements on fluids.

[0003] Tomo-PIV is a PIV testing technology based on multi-eye vision. It has the advantages of high measurement accuracy, large measurement area, and is suitable for high particle concentrations. However, this technology has many shortcomings, such as complex calibration operations, the need for multiple cameras, and cannot be used when the optical measurement window is relatively narrow. Due to the instant volume imaging capability of light field technology, a light field PIV (LF-PIV) system has been developed in recent years. It only requires a single light field camera and does not require optical or mechanical scanning. Compared with traditional modes such as Tomo-PIV, LF-PIV greatly simplifies the imaging system. The simplicity of the system and the convenience of light field imaging give it great application freedom under conditions of limited space and limited optical channels. Although LF-PIV can realize the acquisition of three-dimensional position information, its axial resolution is relatively poor, and the reconstructed particles usually have an elongated shape with much larger axial dimensions than lateral dimensions. The low axial resolution severely limits the accuracy of LF-PIV in particle position positioning, thereby affecting the accuracy of the calculated velocity vector.

[0004] In order to improve the axial resolution of light field PIV technology, the color and depth encoding (Cade) method has been developed recently. The principle is to change the color of the light beam so that particles at different depths are irradiated with different wavelengths. When photographed with a color camera, particles at different depths appear in different colors, forming color and depth encoding in the particle image. By accurately parsing the relationship between color and depth, the axial position of the particle can be determined more accurately than monochrome PIV. The currently proposed Cade method of generating spectra has the problems of low light source power, unsuitability for variable-scale light fields, and complex customized diffraction optical elements, and its application is limited to low-speed flow imaging. Summary of the invention

[0005] Purpose of the invention: In view of the disadvantage of poor axial resolution of the prior art, the present invention provides a flow field measurement method and system based on multi-color laser synthesis of a new spectrum and light field imaging.

[0006] Technical solution: To solve the above problems, the present invention adopts a flow field measurement method based on multi-color laser synthesis of a new spectrum and light field imaging, including the following steps:

[0007] Step 1, build an experimental optical path for multi-color laser synthesis, use three lasers emitting RGB colors as excitation sources, the wavelengths of the three lasers are 405nm, 532nm, and 650nm respectively, the laser beams emitted by the three lasers are modulated by three spatial light modulators respectively, and synthesize a new laser spectrum whose color changes along the depth direction;

[0008] Step 2: Perform a calibration experiment to extract the axial depth point spread function of the new laser spectrum;

[0009] Step 3, using a new laser spectrum to excite the flow field to be measured, and using a light field camera to image the flow field to be measured, and collecting a light field image;

[0010] Step 4, reconstructing the light field image into three-dimensional light field to obtain a light field three-dimensional particle field with color coding information;

[0011] Step 5, color / depth decoding is performed on the light field three-dimensional particle field with color coding information to obtain a three-dimensional distribution image of particles at different depths;

[0012] Step 6: Reconstruct the three-dimensional velocity field, and use a variational optical flow model based on a physical model to solve and obtain the three-dimensional velocity vector field of the flow field to be measured.

[0013] The present invention also provides the measurement system, comprising:

[0014] (1) Light field cameras, including macro lenses, microlens arrays, primary lenses, and industrial cameras;

[0015] (2) An imaging system for synthesizing a new spectrum using multi-color lasers, including three-color lasers with wavelengths of 405nm, 532nm, and 650nm, three spatial light modulators (DMDs), and an optical lens combination;

[0016] (3) A data processor is used to reconstruct the three-dimensional flow field velocity vector field by obtaining the time-series particle light field images.

[0017] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0018] (1) Fewer devices are required for measurement. A single light field camera replaces multiple camera systems or special optical components, which reduces system costs and greatly simplifies system configuration and operation.

[0019] (2) High imaging spatial resolution, using color depth encoding method to improve axial spatial resolution;

[0020] (3) It can be better applied to variable-scale imaging. The three DMDs can modulate the three-color laser to generate a new laser spectrum, which increases the color change rate per unit depth and is beneficial to depth recognition. At the same time, the beam width and color change rate can be freely adjusted through different DMD modulation patterns to match the imaging depth of the experimental field, making the system easier to adapt to different experimental scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic flow chart of the flow field measurement method of the present invention;

[0022] Figure 2 It is a schematic diagram of a flow field measurement system for synthesizing a novel laser spectrum and light field imaging according to the present invention;

[0023] Figure 3 It is a new type of laser spectrum and light intensity distribution diagram synthesized by the present invention;

[0024] Figure 4 It is a device and experimental principle diagram of the calibration experiment of the present invention;

[0025] Figure 5 It is a schematic diagram of the principle of the light field refocusing algorithm of the present invention. DETAILED DESCRIPTION

[0026] like Figure 1 As shown, a flow field measurement method based on multi-color laser synthesis of a new spectrum and light field imaging in this embodiment includes the following steps:

[0027] Step 1: Build the experimental optical path for multi-color laser synthesis, using three lasers emitting RGB colors as excitation sources. The wavelengths of the three lasers are 405nm, 532nm, and 650nm respectively. Figure 2 As shown, the laser beams emitted by the three lasers are modulated by the gradient patterns pre-loaded by the three spatial light modulators (DMDs) to synthesize a new laser spectrum whose color changes along the depth direction, such as Figure 3 The beam width and color change rate of the new laser spectrum are adjusted through the modulation pattern of the spatial light modulator and the lens combination to match the imaging depth of the experimental field, making the system more adaptable to different experimental scenarios and effectively promoting high-precision measurement research on three-dimensional flow fields.

[0028] Step 2: Perform a calibration experiment to extract the axial depth point spread function of the new laser spectrum. First, place a tracer particle with a diameter of 500 μm on the top of the bracket, and fix the bracket on the electric translation stage. The electric translation stage controls the particle to translate through the excitation beam. The light field camera collects images of the particles at certain intervals to obtain light field image data at different depths. Then use the light field reconstruction algorithm to reconstruct the particle light field image obtained at each depth, and extract the PSF color information at each depth, such as Figure 4 As shown, the depth and color calibration is completed as the basis for estimating the corresponding color when the particle is at the depth. Through this calibration experiment, the intensity of the synthesized excitation beam is evenly distributed in different depth planes, and the intensity distribution of each color channel along the depth direction is close to the ideal design state to the greatest extent, so that the color in the depth direction changes regularly.

[0029] Step 3: Evenly spread the tracer particles in the flow field to be measured, use the calibrated new laser spectrum to excite the flow field to be measured, and use the light field camera to image the flow field to be measured, ensure that the depth of field of the light field camera is greater than the depth of the illumination light, and collect the time-series light field image of the tracer particles in the flow field.

[0030] Step 4: Perform three-dimensional light field reconstruction on the light field image to obtain a three-dimensional particle field with color coding information. The collected light field image is reconstructed by the light field refocusing algorithm to obtain a three-dimensional particle field with color coding information. Light field reconstruction is a process of refocusing the tracer particles at different depths in the image based on the light field principle to obtain a three-dimensional image. The geometric relationship between the lens plane, the sensor plane and the refocusing plane is shown in Figure 5 As shown, the light field refocusing algorithm equation is as follows:

[0031]

[0032] Where: The light field camera obtains a blurred image E at an image distance of l = F F (x, y), a clear image E is obtained on the image plane at image distance l′=αF αF (x, y′), α is the coefficient for adjusting the image distance l′, and the light passing through the point (x, y) on the image plane l=F is recorded as L F(u, u, x, y), when the light reaches the αtF image plane, the projection coordinate will change and be recorded as L αF .

[0033] Step 5: Color / depth decoding is performed on the light field three-dimensional particle field with color coding information to obtain a three-dimensional distribution image of particles at different depths. The identified color coding information is compared with the color / depth coding index calibrated by the system, and the depth estimation function is solved and iteratively calculated by the alternating direction multiplier method (ADMM) solver to obtain a high-precision three-dimensional particle field intensity distribution. The comparison process with the color / depth coding index calibrated by the system is as follows:

[0034] i c (x) = ∫ Λ ∫ X g C (xsx′,λ)·i r (x, λ)·P(x, λ)dx′dλ,

[0035] Where: the wavelength λ of the light beam varies with depth (z coordinate), the pixel position of the particle in the light plane can be expressed as (x, λ) = (x, y, λ), the intensity points occupied by the particle in the reconstructed three-dimensional field can be expressed by the occupancy probability rate P(x, λ), (x′, λ) is the image position where the calibration color is located, and the position function corresponding to each color channel is g c (x,λ),i C (x) is the color channel of the collected RGB image, i r (x, λ) is the corresponding spectral distribution pattern incident on the camera sensor.

[0036] After the comparison is completed, the difference between the reconstructed RGB image and the RGB image collected by the light field camera is calculated by the alternating direction multiplier method (ADMM), thereby improving the axial resolution. The depth estimation function p* is expressed as:

[0037]

[0038] Where: i is the color-coded particle image, A is the convolution operator, p is the three-dimensional particle distribution position, and α is the weight factor for adjusting the spatial particle sparsity.

[0039] For the point spread functions obtained at different depths, we first construct the convolution operator matrices A of the three RGB channels respectively, and then construct the depth estimation function for solving the position P. We also add relevant weight parameters between depth / color to compensate for the camera spectral sensitivity, and solve the depth estimation function by the alternating direction multiplier method (ADMM):

[0040]

[0041] subject to p j -y=0

[0042] Among them: j represents different color channels, p j is a relative local variable, and y is a global consistent variable; the depth estimation problem is decomposed into three sub-problems for iterative solution, and the iterative update expression is as follows:

[0043]

[0044]

[0045]

[0046] Where: k represents the number of iterations, is the transposed matrix of the A convolution operator, I is the identity matrix, q j is the Lagrange multiplier, and Represents the color channel p j and q j is the average value of , ρ is the penalty parameter, and it is a positive value.

[0047] Step 6, reconstruct the three-dimensional velocity field, and use a variational optical flow model based on a physical model to solve the three-dimensional velocity vector field of the flow field to be measured. The optical flow method based on the global solution calculates the velocity field with high axial resolution. Optical flow is essentially the brightness change of pixels projected onto the two-dimensional image plane by the motion of objects in the three-dimensional scene. Optical flow calculation is to estimate the motion of objects from image data. The objective function of the classic variational optical flow model is a minimization representation of an energy function containing a data term and a smoothing term. The data term is mainly based on the brightness conservation assumption, that is, the brightness (grayscale value) of the same pixel remains unchanged in two adjacent frames of the image, that is:

[0048] I(x+dx,y+dy,t+dt)=I(x,y,t)

[0049] Where: I(x, y, i) represents the brightness (grayscale value) of the pixel with coordinates (x, y) at time t.

[0050] The optical flow model expression used is:

[0051] E total =E date +γ1E smooth +γ2E corres +γ3E aiv

[0052] Among them: γ1, γ2, γ3 are the balance factors of each item, E dateand E smooth From the basic optical flow method, derived from the constant brightness assumption, E corres is the tracer particle counterpart, E div is a divergence-free term.

[0053] The present invention also provides a measurement system for the above flow field measurement method, comprising:

[0054] (1) Light field cameras, including macro lenses, microlens arrays, primary lenses, and industrial cameras;

[0055] (2) An imaging system for synthesizing a new spectrum using multi-color lasers, including three-color lasers with wavelengths of 405nm, 532nm, and 650nm, three spatial light modulators (DMDs), and an optical lens combination;

[0056] (3) A data processor is used to reconstruct the three-dimensional flow field velocity vector field by obtaining the time-series particle light field images.

[0057] In summary, the measurement of the present invention requires less equipment, and a single light field camera replaces multiple camera systems or special optical elements, which reduces the system cost and greatly simplifies the configuration and operation of the system; the imaging spatial resolution is high, and the axial spatial resolution is improved by using a color depth encoding method; it can be better applied to variable-scale imaging, and a new laser spectrum generated by modulating three-color lasers respectively by three DMDs increases the color change rate per unit depth, which is beneficial to depth recognition; at the same time, the beam width and color change rate can be freely adjusted to match the imaging depth of the experimental field, making the system more adaptable to different experimental scenarios.

Claims

1. A flow field measurement method based on multi-color laser synthesis of new spectrum and light field imaging, characterized in that: The following steps are involved: Step 1, build an experimental optical path for multi-color laser synthesis, use three lasers emitting RGB colors as excitation sources, the wavelengths of the three lasers are 405nm, 532nm, and 650nm respectively, the laser beams emitted by the three lasers are modulated by three spatial light modulators respectively, and synthesize a new laser spectrum whose color changes along the depth direction; Step 2: Conduct a calibration experiment to extract the axial depth point spread function of the new laser spectrum; Step 3, using a new laser spectrum to excite the flow field to be measured, and using a light field camera to image the flow field to be measured, and collecting a light field image; Step 4: Perform three-dimensional light field reconstruction on the light field image to obtain a three-dimensional particle field with color coding information; the three-dimensional light field reconstruction is specifically: reconstruct the collected light field image through a light field refocusing algorithm to obtain a three-dimensional particle field with color coding information. The light field refocusing algorithm equation is as follows: Where: The light field camera obtains a blurred image E at an image distance of l = F F (x, y), a clear image E is obtained on the image plane at image distance l′=αF αF (x, y), α is the coefficient for adjusting the image distance l', and the light passing through the point (x, y) on the image plane l = F is recorded as L F (u, v, x, y), when the light reaches the αF image plane, the projection coordinates will change and be recorded as L αF ; Step 5, color / depth decoding is performed on the light field three-dimensional particle field with color coding information to obtain a three-dimensional distribution image of particles at different depths; Step 6: Reconstruct the three-dimensional velocity field, and use a variational optical flow model based on a physical model to solve and obtain the three-dimensional velocity vector field of the flow field to be measured.

2. The flow field measurement method according to claim 1, characterized in that: In step 1, the laser beams emitted by the three lasers are modulated by the gradient patterns pre-loaded by the three spatial light modulators respectively.

3. The flow field measurement method according to claim 1, characterized in that: The beam width and color change rate of the new laser spectrum described in step 1 are adjusted by the modulation pattern of the spatial light modulator and the lens combination.

4. The flow field measurement method according to claim 1, characterized in that: The step 2 is specifically as follows: a single tracer particle is placed on the top of the bracket, and the bracket is fixed on the electric translation stage, and is moved sequentially along the depth direction to obtain light field image data at different depths, from which the PSF color information of each depth is extracted, and the depth and color calibration is completed, which is used as the basis for estimating the corresponding color when the particle is at its depth.

5. The flow field measurement method according to claim 1, characterized in that: The step 3 specifically includes: evenly spreading tracer particles in the flow field to be measured, exciting the flow field to be measured using a calibrated new laser spectrum, imaging the flow field to be measured using a light field camera, and collecting time-series light field images of the tracer particles in the flow field.

6. The flow field measurement method according to claim 1, characterized in that: The color / depth decoding described in step 5 is specifically as follows: the recognized color coding information is compared with the color / depth coding index calibrated by the system, the depth estimation function is solved by the alternating direction multiplier method solver and iterative calculation is performed to obtain a high-precision three-dimensional particle field intensity distribution; the comparison process with the color / depth coding index calibrated by the system is as follows: i c (x)=∫∫g c (x-x′,λ)·i r (x,λ)·P(x,λ)dx′dλ, Where: the wavelength λ of the light beam varies with depth, the pixel position of the particle in the light plane is represented by (x, λ) = (x, y, λ), the intensity point occupied by the particle in the reconstructed three-dimensional field is represented by the occupancy probability P(x, λ), (x′, λ) is the image position where the calibration color is located, and the position function corresponding to each color channel is g c (x, λ), i c (x) is the color channel of the collected RGB image, i r (x, λ) is the corresponding spectral distribution pattern incident on the camera sensor.

7. The flow field measurement method according to claim 6, characterized in that: The depth estimation function p* is expressed as: Where: i is the color-coded particle image, A is the convolution operator, p is the three-dimensional particle distribution position, and α is the weight factor for adjusting the spatial particle sparsity; For the point spread functions obtained at different depths, we first construct the convolution operator matrices A of the three RGB channels respectively, and then construct the depth estimation function for solving the position P. We also add relevant weight parameters between depth / color to compensate for the camera spectral sensitivity, and solve the depth estimation function by the alternating direction multiplier method: subject to p j -y=0 Among them: j represents different color channels, p j is a relative local variable, and y is a global consistent variable; the depth estimation problem is decomposed into three sub-problems for iterative solution, and the iterative update expression is as follows: Where: k represents the number of iterations, is the transposed matrix of the A convolution operator, I is the identity matrix, qj is the Lagrange multiplier, and Represents the color channel p j and q j is the average value of , ρ is the penalty parameter, and it is a positive value.

8. The flow field measurement method according to claim 1, characterized in that: The objective function of the variational optical flow model in step 6 is a minimization representation of an energy function including a data term and a smoothing term. The data term is based on the brightness conservation assumption, that is, the brightness of the same pixel remains unchanged in two adjacent frames of images, that is: I(x+dx,y+dy,t+dt)=I(x,y,t) Where: I(x, y, t) represents the brightness of the pixel with coordinates (x, y) at time t; The optical flow model expression used is: E total =E date +γ1E smoth +γ2E corres +γ3E div Among them: γ1, γ2, γ3 are the balance factors of each item, E date and E smooth From the basic optical flow method, derived from the constant brightness assumption, E corres is the tracer particle counterpart, E div is a divergence-free term.

9. A measurement system for implementing the flow field measurement method according to any one of claims 1 to 8, characterized in that: include: (1) Light field cameras, including macro lenses, microlens arrays, primary lenses, and industrial cameras; (2) An imaging system for synthesizing a new spectrum using multi-color lasers, including three-color lasers with wavelengths of 405nm, 532nm, and 650nm, three spatial light modulators, and an optical lens combination; (3) A data processor is used to reconstruct the three-dimensional flow field velocity vector field by obtaining the time-series particle light field images.

Images