A data processing method for measuring flow field parameters based on marker spot characteristic information

Abstract

To address the problem of limited noise resistance in existing methods for measuring flow field parameters based on marker light spots, the present invention provides a data processing method for measuring flow field parameters based on marker light spot characteristic information. The present invention analyzes characteristic parameters in the original marker light spot fluorescence image and uses a generative adversarial neural network method for denoising to obtain a marker light spot fluorescence image after image preprocessing. Based on the characteristic information of the marker light spot fluorescence image, a precise spot center extraction method for saliency region detection is proposed. After saliency region detection, Otsu threshold segmentation is performed, and the spot area is obtained based on the results of maximum connected area selection. Based on the precise location of the spot center, the position difference, area, and intensity information of the marker light spot are used to obtain the flow field velocity and temperature, thereby reducing the uncertainty of flow field parameter measurement based on marker light spot characteristic information.

Description

Technical Field

[0001] The invention relates to image processing, belongs to the technical field of complex field laser spectrum diagnosis and analysis, and particularly relates to a data processing method for measuring flow field parameters based on marker light spot characteristic information. Background Art

[0002] High-temporal and spatial-resolution parameter diagnostics are playing an increasingly important role in the study of turbulence-combustion interaction mechanisms, CFD simulation verification, and engine combustion chamber design. The extreme flow conditions within engine combustion lead to more complex interactions between lasers and combustion, placing higher demands on measurement uncertainty. Molecular tagging measurement technology, which eliminates issues such as flow field tracking and uneven seeding of tagged molecules and offers high spatiotemporal resolution, offers significant advantages in measuring boundary layer flows, hypersonic flows, turbulent combustion flows, and flows in geometrically complex channels (such as those in gas turbines, combustion chambers, and high-speed nozzles). However, complex field applications still face numerous challenges, notably the low signal-to-noise ratio of the marker spot fluorescence image, which significantly impacts measurement uncertainty. When utilizing the position, area, and intensity of a marker spot to measure flow velocity and temperature, the area and intensity of the marker spot are extracted based on its center position. Therefore, the key to improving the uncertainty of flow parameter measurements based on marker spots is to accurately extract the marker spot's center position.

[0003] The noise immunity of the commonly used Gaussian fitting and correlation methods has certain limitations. The fitting method obtains the intensity distribution of the marker line profile through fitting, and the marker line position is determined by the peak of the fitting curve. Miles RB et al. (Miles RB, Zhou D, Zhang B, et al. Fundamental turbulence maesurements by relief flow tagging [J]. AIAA Journal, 1993, 31 (3): 447-452) proposed using Gaussian fitting to extract the marker line position when using RELIEF to study turbulent motion; Lempert et al. (Lempert WR, Boehm M, Jiang N, et al. Comparison of molecular tagging velocimetry data and direct simulation Monte Carlo simulations in supersonic micro jet flows [J]. Experiments in Fluids, 2003, 34 (3): 403-411) proposed using the least squares fitting method of the Gaussian curve to determine the center position of the marker line, which depends on the accuracy of the marker line intensity distribution function; Hill et al. (Hill RB, Klewiciki JC, Data reduction methods for flow tagging velocity measurements [J], Experiments in Fluids, 1996, 20(3):142-152. Second-order least squares curve fitting is used to determine the peak position of a smoothed Gaussian function. However, due to factors such as the flow field and laser effects, the assumed prior function may not match the actual situation, which will increase the uncertainty introduced. Therefore, the fitting method still has significant uncertainty in extreme measurement environments with severe background interference and large signal distortion.

[0004] The correlation method is to determine the correlation coefficient by polynomial fitting, thereby determining the displacement of the labeling signal. Gendrich et al. (Gendrich CP, Koochesfahani MM. A Spatial Correlation Technique for Estimating Velocity Fields Using Molecular Tagging Velocimetry (MTV) [J]. Experiments in Fluids, 1996, 22 (1): 67-77) proposed a direct correlation method to extract the center position of the spot; later, a two-dimensional seventh-order polynomial fitting result was used (Gendrich CP, Koochesfahani MM, Nocera DG, et al. Molecular tagging velocimetry and other novel applications of a new phosphorescent supremolecule [J]. Experiments in Fluids, 1997).

[0005] Fitting and correlation methods are the most widely used. Although these image processing methods offer low measurement uncertainty when applied in specialized environments, their advantages are limited by the limitations of specific experimental data and environments. Furthermore, factors such as changes in laser and marker signals caused by heat-induced motion in complex thermal flows can also introduce measurement uncertainty. The uncertainty in velocity measurement in high-temperature, high-speed combustion flow fields is limited by strong background interference and the low signal-to-noise ratio of the image. Due to the complexity of the physical processes involved in the generation and display of photodissociated hydroxyl radicals, the intensity of the fluorescence signal from the photodissociated hydroxyl radical markers is affected by numerous factors. For example, the complex composition of the flow field can reduce the efficiency of photodissociated hydroxyl radical fluorescence generation. The optical window glass absorbs some laser energy, and the window material itself limits the incident laser energy threshold. Strong absorption of the marker laser by macromolecular fuels can also reduce dissociation efficiency. Furthermore, the diffusion and chemical reactions of photodissociated hydroxyl radicals in the flow field can weaken the signal intensity and reduce the signal-to-noise ratio, making accurate extraction of photodissociated hydroxyl radical marker information challenging. Summary of the Invention

[0006] The purpose of the present invention is to solve the problem that the existing method for measuring the uncertainty of flow field parameters based on marker light spots has great limitations in noise resistance, and to provide a data processing method for measuring flow field parameters based on marker light spot characteristic information.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A data processing method for measuring flow field parameters based on characteristic information of a marking light spot comprises the following steps:

[0009] Step 1: Use a camera to capture the original fluorescent image of the marker light spot in the flow field, analyze the characteristic parameters of the captured original fluorescent image I1, and use a generative adversarial neural network method to remove noise, thereby obtaining a preprocessed fluorescent image I2;

[0010] Step 2: Perform saliency detection and extraction on the pre-processed marked light spot fluorescence image I2 using a histogram global contrast method to obtain a light spot image I3 after saliency detection;

[0011] Step 3: Perform Otsu threshold segmentation on the obtained saliency detected image I3 to obtain connected regions and mark their area sizes. The connected region with the largest area is selected as the spot mask image I4.

[0012] Step 4: perform dot multiplication on the spot mask image I4 and the pre-processed marked spot fluorescence image I2 to obtain the spot area image I5;

[0013] Step 5. Fit and calculate the centroid position of the spot area image I5. Based on the histogram comparison method, arrange the grayscale values of the spot area image I5 in descending order, set the energy threshold Y, mark the range around the centroid where the proportion of the total energy exceeds Y, and regard the energy of the first N pixels and the pixels exceeding the total energy Y as the final spot center position d;

[0014] Step 6: Repeat steps 1 to 5 at least once. At this time, at least two or more light spot center positions d are obtained. Two of the light spot center positions d are selected and defined as d1 and d2 respectively.

[0015] Step 7: Calculate the flow field velocity v at the location of the fluorescent image of the marking spot using the spot center position difference:

[0016] v=(d1-d2) / Δt

[0017] Wherein, Δt is the time interval between the two fluorescence images of the marker spot;

[0018] Step 8: Calculate the flow field temperature using the change in the area of the marking light spot, or calculate the flow field temperature using the intensity of the marking light spot.

[0019] Among them, step seven and step eight can be calculated simultaneously or calculated separately as needed.

[0020] Furthermore, in step eight, the flow field temperature is calculated by using the change in the area of the marking light spot, specifically:

[0021] 8.1. Take the obtained spot center positions d1 and d2 as the points, and draw circles with radius a to cover the entire fluorescent image of the marking spot;

[0022] 8.2. Take the region of interest where the grayscale threshold is b in the circle as image I 61 and image I 62 , respectively calculate the image I 61 and image I 62 The pixels occupied are recorded as spot areas S1 and S2;

[0023] 8.3. Calculate the flow field temperature T using the power function relationship between the spot area change and the flow field temperature T:

[0024] T n =m(S1-S2) / Δt

[0025] Among them, m and n are power function coefficients, which are related to the labeled molecules and flow field types.

[0026] Furthermore, in step eight, the flow field temperature is calculated using the marking spot intensity, specifically:

[0027] 8.1. Take the obtained spot center positions d1 and d2 as the points, and draw circles with radius a to cover the entire fluorescent image of the marking spot;

[0028] 8.2. Take the region of interest where the grayscale threshold is b in the circle as image I 61 and image I 62 , calculate image I 61 and image I 62 The sum of the light intensities of the occupied pixels is recorded as the spot intensity F;

[0029] 8.3. Calculate the flow field temperature T using the relationship between the spot intensity F and the flow field temperature T:

[0030] F=CN H2O exp(-E / kT) / ∑exp(-E / kT)

[0031] Among them, N H2O is the water concentration, exp(-E / kT) / ∑exp(-E / kT) is the partition function of the number distribution of particles at different energy levels; the corresponding relationship between F and T is calibrated by linear transformation to obtain the coefficient C.

[0032] Furthermore, in step 7.1, a=30 pixels; and in step 7.2, b=60 pixels.

[0033] Furthermore, step 2 is specifically as follows:

[0034] 2.1、Calculate any two pixels r in the fluorescent image I2 of the marker spot i and r j The color distance D between r (r j ,r i ):

[0035] D r (r j ,r i )=f(c j )f(c i )*|c j -c i |

[0036] Among them, f(c i ) and f(c j ) represent pixel r i and r j Single channel color value of |c j -c i | represents pixel r j and r i Mahalanobis distance between coordinates;

[0037] 2.2. Calculate pixel r j The significance value X(r j ):

[0038]

[0039] Among them, w(r i ) represents pixel r i The weight of

[0040] 2.3、Get the saliency value X(r j ), and obtain the image I3 after saliency detection.

[0041] Furthermore, in step 2.2, w(r i )=5%;

[0042] In step 5, Y=70%.

[0043] Compared with the prior art, the present invention has the following beneficial technical effects:

[0044] 1. The data processing method provided by the present invention for measuring flow field parameters based on the characteristic information of the marker light spot proposes a method for accurately extracting the light spot center by detecting significant areas based on the characteristic information of the fluorescent image of the marker light spot. After the significant area is detected, Otsu threshold segmentation is performed, and the light spot area is obtained by combining the result of the maximum connected area selection. On the basis of accurately locating the light spot center, the position difference, spot area and spot intensity information of the marker light spot are used to obtain the flow field velocity and temperature, thereby reducing the uncertainty of measuring the flow field parameters based on the characteristic information of the marker light spot.

[0045] 2. The data processing method for measuring flow field parameters based on the characteristic information of the marked light spot provided by the present invention adopts the method of significant area detection, which effectively avoids the "hole" phenomenon caused by directly performing threshold segmentation on the input image based on the grayscale value when there are several pixels with brightness values lower than the surrounding pixels in the light spot area.

[0046] 3. The data processing method for measuring flow field parameters based on the characteristic information of the marking light spot provided by the present invention adopts global contrast which is superior to local contrast and can eliminate the influence of the surrounding environment on the target; the method based on histogram contrast determines the significance value of each pixel by the difference in grayscale value between it and all other pixels in the image, and obtains a significance image based on the global contrast of the histogram, which can ensure that the significance of the central (significant) area is significantly higher than that of other areas and has better anti-noise performance.

[0047] 4. The data processing method provided by the present invention for measuring flow field parameters based on the characteristic information of the marked light spot avoids high-order polynomial fitting in the area of interest near the light spot for images with complex backgrounds that emphasize the sensitivity of single pixel intensity, compared with the fitting method and the cross-correlation method, thereby improving the computational efficiency and noise resistance. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of a data processing method for measuring flow field parameters based on the characteristic information of a marking light spot according to the present invention;

[0049] Figure 2 Schematic diagram of the principle of the imaging analysis method for extracting the center of the light spot in an embodiment of the present invention;

[0050] Figure 3 Schematic diagram of the extracted spot area in an embodiment of the present invention; (a) is a schematic diagram of the extracted central spot; (b) is a schematic diagram of the central spot of the statistical spot range; (c) is a schematic diagram of the central spot after the area and intensity of the photodissociated hydroxyl labeled spot (OHp-TFS) are extracted. DETAILED DESCRIPTION

[0051] To make the objects, advantages and features of the present invention more clear, a data processing method for measuring flow field parameters based on marking spot characteristic information proposed by the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] like Figure 1 and Figure 2 As shown, the data processing method of the present invention for measuring flow field parameters based on the characteristic information of the marking spot has the following specific operating steps:

[0053] Step 1: Image preprocessing

[0054] The characteristic parameters of the noise intensity mean, noise intensity standard deviation, noise smoothness, noise relative intensity range, noise grayscale distribution skewness and kurtosis in the original fluorescent image I1 of the marker light spot were analyzed. A noise model with a mean square error of less than 10% was proposed. Based on the high-confidence noise model, a generative adversarial neural network method was used to denoise the image, and the fluorescent image I2 of the marker light spot after image preprocessing was obtained.

[0055] Step 2: Detection of salient regions

[0056] The histogram global contrast method is used to extract the saliency of the pre-processed fluorescent image I2, specifically:

[0057] 2.1、Calculate any two pixels r in the preprocessed marker spot fluorescence image I2 i and r j The color distance D between r (r j ,r i ):

[0058] D r (r j ,r i )=f(c j )f(c i )*|c j -c i | (1)

[0059] Among them, f(c i ) and f(c j ) represent pixel r i and r j Single channel color value of |c j -c i | represents pixel r j and r i Mahalanobis distance between coordinates;

[0060] 2.2. Calculate pixel r j The significance value X(r j ):

[0061]

[0062] Among them, w(r i ) represents pixel r i The weight, w(r i )=5%.

[0063] 2.3、Get the saliency value X(r j ), and obtain the image I3 after saliency detection.

[0064] After saliency extraction, the difference between the center of the marked spot and the background is enhanced, and the background interference is suppressed to a certain extent.

[0065] Step 3: Perform Otsu thresholding and select the largest connected area

[0066] The obtained saliency detected image I3 is segmented by Otsu threshold to obtain connected regions. Since the connected regions are not unique, their area sizes are marked and the connected region with the largest area is selected as the spot mask image I4.

[0067] Step 4: Get the spot area image

[0068] Perform dot multiplication on the mask image I4 and the pre-processed marked spot fluorescence image I2. The mask is a binary image. The position with a value of 1 remains the value of I2 after dot multiplication, and the position with a value of 0 is set to 0 after dot multiplication to obtain the spot area image I5.

[0069] Step 5: Fit and calculate the centroid position and extract the spot center (based on histogram comparison method)

[0070] The centroid position is calculated by fitting the spot area image I5 obtained according to the above steps. At the same time, the grayscale values of the spot area are arranged in descending order, and the energy threshold Y is set. The range around the centroid where the proportion of the total energy exceeds Y is found, and the pixels whose energy at the first N positions exceeds the total energy Y are regarded as the final spot center position d (set Y = 70%).

[0071] Step 6. Repeat steps 1 to 5 at least once to obtain the center position d of one or more light spots accordingly; among the center positions d of the light spots obtained using steps 1 to 5 and the center positions d of one or more light spots obtained by repeating steps 1 to 5, select the TFS center positions d of two of the light spots and define them as d1 and d2, respectively.

[0072] Step 7: Calculate the flow field velocity v at the location of the fluorescent image of the marker light spot using the center position difference of the marker light spot:

[0073] v=(d1-d2) / Δt (3)

[0074] Wherein, Δt is the time interval between the two fluorescent images of the marker spot;

[0075] Step 8. Calculate the flow field temperature using the change in the marking spot area or spot intensity

[0076] The flow field temperature is calculated using the change in the marking spot area, specifically:

[0077] 1. Take the extracted spot center positions d1 and d2 as the circle points and draw a circle with a radius a (covering the entire fluorescent image of the marking spot);

[0078] 2. Take the region of interest where the grayscale threshold is b in the circle as image I 61 and image I 62 , respectively calculate the image I 61 and image I 62 The pixels occupied are recorded as spot areas S1 and S2;

[0079] 3. Calculate the flow field temperature T using the power function relationship between the spot area change and the flow field temperature T: (set a = 30 pixels, b = 60 pixels)

[0080] T n =m(S1-S2) / Δt (4)

[0081] Among them, m and n are power function coefficients.

[0082] Once the marker molecule and flow field are determined, n can be determined. When measuring the temperature of a He flow field containing H2O using OH (hydroxyl group) as the marker molecule, the coefficient n = 1.76. Taking the diffusion of OH in a He flow field at 707K as an example, by measuring the area of the marker spot at delay times of 150μs and 100μs, the coefficient m in the power function curve between area change and temperature is calibrated. This provides a one-to-one correspondence between area change and flow field temperature, enabling flow field temperature measurement.

[0083] The flow field temperature is calculated using the intensity of the marker light spot, specifically:

[0084] 1. Take the extracted spot center positions d1 and d2 as the circle points and draw circles with radius a to cover the entire fluorescent image of the marking spot;

[0085] 2. Take the region of interest where the grayscale threshold is b in the circle as image I 61 and image I 62 , calculate image I 61 and image I 62 The sum of the light intensities of the occupied pixels is recorded as the spot intensity F;

[0086] 3. Using the relationship between the spot intensity F and the flow field temperature T, calculate the flow field temperature T: (set a = 30 pixels, b = 60 pixels)

[0087] F=CN H2O exp(-E / kT) / ∑exp(-E / kT) (5)

[0088] In the formula, the correlation coefficients of laser, detector, etc. are combined into coefficient C (C is independent of temperature T and can be calibrated by experiment) water concentration N H2O , the partition function of the number distribution of particles at different energy levels

[0089] exp(-E / kT) / ∑exp(-E / kT) is related to the temperature T. It can be seen that there is a corresponding relationship between the fluorescence intensity F and the temperature T. The corresponding relationship between F and T is calibrated by linear transformation to obtain the coefficient C. The absolute temperature T of the flow field can be calculated using the light spot intensity F in the flow field according to the formula.

[0090] Among them, step seven and step eight can be calculated simultaneously or calculated separately as needed.

[0091] Figure 3 Schematic diagram of spot area extraction in this embodiment, where (a) is a schematic diagram of the extracted central spot; (b) is a schematic diagram of the central spot of the statistical spot range; (c) is a schematic diagram of the central spot after the area and intensity of the photodissociated hydroxyl labeled spot (OHp-TFS) are extracted. It can be seen that after significance extraction, the significance of the central (significant) area can be ensured to be significantly higher than that of other areas, and it has better noise resistance performance.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present invention.

Claims

1. A data processing method for measuring flow field parameters based on marker spot characteristic information, characterized in that: The following steps are involved: Step 1: Use a camera to capture the original fluorescent image of the marker light spot in the flow field, analyze the characteristic parameters of the captured original fluorescent image I1, and use a generative adversarial neural network method to remove noise, thereby obtaining a preprocessed fluorescent image I2; Step 2: Perform saliency detection and extraction on the pre-processed marked light spot fluorescence image I2 using a histogram global contrast method to obtain a light spot image I3 after saliency detection; Step 3: Perform Otsu threshold segmentation on the obtained saliency detected image I3 to obtain connected regions and mark their area sizes. The connected region with the largest area is selected as the spot mask image I4. Step 4: perform dot multiplication on the spot mask image I4 and the pre-processed marked spot fluorescence image I2 to obtain the spot area image I5; Step 5. Fit and calculate the centroid position of the spot area image I5. Based on the histogram comparison method, arrange the grayscale values of the spot area image I5 in descending order, set the energy threshold Y, mark the range around the centroid where the proportion of the total energy exceeds Y, and regard the energy of the first N pixels and the pixels exceeding the total energy Y as the final spot center position d; Step 6: Repeat steps 1 to 5 at least once. At this time, at least two or more light spot center positions d are obtained. Two of the light spot center positions d are selected and defined as d1 and d2 respectively. Step 7: Calculate the flow field velocity v at the location of the fluorescent image of the marking spot using the spot center position difference: v=(d1-d2) / Δt Wherein, Δt is the time interval between the two fluorescence images of the marker spot; Step 8: Calculate the flow field temperature using the power function relationship between the change in the marking spot area and the flow field temperature, or calculate the flow field temperature using the marking spot intensity.

2. The data processing method for measuring flow field parameters based on marker light spot characteristic information according to claim 1, characterized in that: In step eight, the flow field temperature is calculated using the change in the area of the marking light spot, specifically: 8.

1. Take the obtained spot center positions d1 and d2 as the points, and draw circles with radius a to cover the entire fluorescent image of the marking spot; 8.

2. Take the region of interest where the grayscale threshold is b in the circle as image I 61 and image I 62 , respectively calculate the image I 61 and image I 62 The pixels occupied are recorded as spot areas S1 and S2; 8.

3. Calculate the flow field temperature T using the power function relationship between the spot area change and the flow field temperature T: T n =m(S1-S2) / Δt Among them, m and n are power function coefficients, which are related to the labeled molecules and flow field types.

3. The data processing method for measuring flow field parameters based on marker spot characteristic information according to claim 1, characterized in that: In step eight, the flow field temperature is calculated using the marking spot intensity, specifically: 8.

1. Take the obtained spot center positions d1 and d2 as the points, and draw circles with radius a to cover the entire fluorescent image of the marking spot; 8.

2. Take the region of interest where the grayscale threshold is b in the circle as image I 61 and image I 62 , calculate image I 61 and image I 62 The sum of the light intensities of the occupied pixels is recorded as the spot intensity F; 8.

3. Calculate the flow field temperature T using the relationship between the spot intensity F and the flow field temperature T: F=CN H2O exp(-E / kT) / ∑exp(-E / kT) Among them, N H2O is the water concentration, exp(-E / kT) / ∑exp(-E / kT) is the partition function of the number distribution of particles at different energy levels; the corresponding relationship between F and T is calibrated by linear transformation to obtain the coefficient C.

4. The data processing method for measuring flow field parameters based on marker spot characteristic information according to claim 2 or 3, characterized in that: In step 7.1, a=30 pixels; in step 7.2, b=60 pixels.

5. The data processing method for measuring flow field parameters based on the characteristic information of the marking light spot according to claim 4 is characterized in that: Step 2 is as follows: 2.1、Calculate any two pixels r in the fluorescent image I2 of the marker spot i and r j The color distance D between r (r j ,r i ): D r (r j ,r i )=f(c j )f(c i )*|c j -c i | Among them, f(c i ) and f(c j ) represent pixel r i and r j Single channel color value of |c j -c i | represents pixel r j and r i Mahalanobis distance between coordinates; 2.

2. Calculate pixel r j The significance value X(r j ): Among them, w(r i ) represents pixel r i The weight of 2.3、Get the saliency value X(r j ), and obtain the image I3 after saliency detection.

6. The data processing method for measuring flow field parameters based on marker light spot characteristic information according to claim 5, characterized in that: In step 2.2, w(r i )=5%; In step 5, Y=70%.

Images