Defocused speckle deblurring method based on Wiener filtering

Abstract

The present application provides a defocused speckle deblurring method based on Wiener filtering, comprising: acquiring an original speckle image, including a reference image and at least one deformed image, and evaluating the speckle quality in the original speckle image using an average intensity gradient to obtain an evaluation result; deblurring the deformed image having lower speckle quality based on the evaluation result to obtain a restored image corresponding to the deformed image; dividing the reference image into at least two sub-regions using a virtual grid, and obtaining a correlation between any of the sub-regions and the deformed image to obtain a best matching position; obtaining a full-field displacement corresponding to any of the deformed images based on the best matching position, and further obtaining a full-field strain. By comparing the strain results before and after deblurring, the method can obtain more strain results in a greater depth direction, thereby achieving a range expansion of digital image correlation.

Description

Technical Field

[0001] The present application relates to the field of non-contact material mechanics testing, and in particular to a defocused speckle deblurring method based on Wiener filtering. Background Art

[0002] Digital image correlation technology is one of the non-contact optical measurement methods with application prospects. By tracking the deformation of the speckle pattern on the surface of the material, it can provide in-plane displacement analysis and strain calculation. It has the advantages of being non-destructive, high-precision, and highly efficient, and is widely used in fields such as materials science, mechanical engineering, and civil engineering. However, the use of this technology for image correlation analysis is affected by the imaging depth range. When the displacement of the material in the depth direction exceeds the range of the optical system, its speckle image will appear blurred due to defocus. This ambiguity will significantly increase the difficulty of feature matching between subsequent images, and may even cause matching failure, thereby limiting the range that can be accurately measured in the depth direction. In fields such as structural dynamic testing, material fatigue analysis, and structural health monitoring, the ability to measure depth over a wide range helps to reveal the behavior of materials and structures under extreme conditions.

[0003] To overcome this limitation and increase the measurement range, the imaging system can be improved. For example, the depth range can be extended by increasing the aperture of the lens or using multi-focus fusion technology. A large-aperture lens can increase the depth of field, thereby capturing clear images at a greater depth range, but this requires a specially designed lens and increases the system size and cost. Multi-focus fusion technology takes multiple images at different focus positions and synthesizes them into a high-depth-of-field image, but this method usually requires additional hardware support, such as a multi-camera system, which greatly increases the cost.

[0004] Therefore, a defocused speckle deblurring method that can realize speckle tracking under defocus conditions using digital image correlation technology and overcome the speckle image correlation mismatch problem caused by limited range is urgently needed to be developed. Summary of the Invention

[0005] The purpose of this application is to provide a defocused speckle deblurring method based on Wiener filtering to at least solve one technical problem in the prior art.

[0006] The technical solution of this application is:

[0007] A defocused speckle deblurring method based on Wiener filtering, comprising:

[0008] Acquiring an original speckle image, including a reference image and at least one deformed image, and evaluating the speckle quality in the original speckle image by using an average intensity gradient to obtain an evaluation result;

[0009] deblurring the deformed image with low speckle quality according to the evaluation result to obtain a restored image corresponding to the deformed image;

[0010] Dividing the reference image into a virtual grid to obtain at least two sub-regions, and obtaining a correlation between any of the sub-regions and the deformed image to obtain a best matching position;

[0011] The full-field displacement corresponding to any of the deformed images is obtained based on the optimal matching position, and the full-field strain is further obtained. By comparing the strain results before and after deblurring, the method can obtain more strain results in a larger depth direction, thereby achieving a range expansion of digital image correlation.

[0012] Deblurring the deformed image with low speckle quality according to the evaluation result to obtain a restored image corresponding to the deformed image includes:

[0013] According to the evaluation result, the blur kernel radius and signal-to-noise ratio corresponding to the speckle in the original speckle image are initialized:

[0014] Acquire an initial restored image according to the blur kernel radius and the signal-to-noise ratio;

[0015] The restoration standard of the blurred speckle image is set, and the blur kernel radius and the signal-to-noise ratio are adjusted to obtain a restored image corresponding to the deformed image.

[0016] The obtaining of the initial restored image according to the blur kernel radius and the signal-to-noise ratio comprises:

[0017] Assuming that the blur kernel radius is h, the blur model is expressed as: g = h*f + n;

[0018] Where g is the blurred image, f is the original speckle image, * represents the convolution operation, and n is the noise;

[0019] The estimated value of the original speckle image is expressed as:

[0020] Where F' is the estimated value of the original speckle image, H is the Fourier transform of the blur kernel h, |H 2 ∣ is the amplitude spectrum of the blur kernel, SNR is the signal-to-noise ratio; G represents the Fourier transform of the blurred image g.

[0021] The setting of the standard for restoring the blurred speckle image and adjusting the blur kernel radius and the signal-to-noise ratio include:

[0022] The half-maximum width of the grayscale histogram of the clear speckle image is used as a criterion for restoring the blurred speckle image, thereby adjusting the blur kernel radius and the signal-to-noise ratio. The half-maximum width FWHM of the grayscale histogram is expressed as:

[0023] FWHM=D right -D left ;

[0024] Among them, D right and D left They are the grayscale values ​​corresponding to the first one on the right and left sides of the grayscale histogram that reaches or exceeds half the height of the peak.

[0025] The adjusting of the blur kernel radius and the signal-to-noise ratio comprises:

[0026] Gradually increase the blur kernel radius R and reduce the signal-to-noise ratio SNR to obtain the half-height width of the grayscale histogram of the restored image;

[0027] Comparing the half-height width of the grayscale histogram of the restored image with the half-height width of the clear speckle image within the focus to obtain a comparison result;

[0028] According to the comparison results, the blur kernel radius R and the signal-to-noise ratio SNR are continuously adjusted until the half-maximum width of the grayscale histogram of the restored image reaches the half-maximum width of the grayscale histogram of the clear speckle image in focus. The blur kernel radius R and the signal-to-noise ratio SNR at this time are taken as the optimal values.

[0029] When adjusting the blur kernel radius and the signal-to-noise ratio, the following steps are also included:

[0030] By combining spatial proximity and grayscale similarity weights, weighted averaging is performed on any pixel in the restored image to adjust the ringing effect caused by the blur kernel radius and signal-to-noise ratio, thereby obtaining the final restored image, including:

[0031] Determining a spatial proximity weight by obtaining a spatial distance between any pixel in the restored image and pixels in its neighborhood;

[0032] Determine the grayscale similarity weight by obtaining the grayscale value similarity between any pixel in the restored image and pixels in its neighborhood;

[0033] For the pixel, the spatial proximity weight and the grayscale similarity weight are used to obtain a weighted average of all pixels in the neighborhood, which is used as the final value of the pixel to suppress the ringing effect and obtain the final restored image.

[0034] The step of dividing the reference image into a virtual grid to obtain at least two sub-regions, and obtaining a correlation between any of the sub-regions and the deformed image to obtain the best matching position includes:

[0035] Selecting an area of ​​interest in the reference image as a search window and dividing the area into virtual grids to obtain at least two sub-areas;

[0036] Treating the sub-region composed of the virtual grid as a rigid motion, and obtaining the correlation between any of the sub-regions and the deformed image;

[0037] A bilinear interpolation method is used to search the speckle image at a sub-pixel level to obtain the grayscale value corresponding to the image function of the deformed image at the coordinate (x′, y′) until the best matching position is obtained.

[0038] The use of bilinear interpolation to search the speckle image at a sub-pixel level to obtain a grayscale value corresponding to the image function of the deformed image at the coordinate (x′, y′) includes:

[0039] Assume that f(x1,y1), f(x1,y2), f(x2,y2), and f(x2,y1) represent the grayscale values ​​corresponding to the image functions of the deformed image at coordinates (x1,y1), (x1,y2), (x2,y2), and f(x2,y1), respectively;

[0040] Get the linear interpolation in the x′ direction:

[0041]

[0042]

[0043] Get the linear interpolation in the y′ direction:

[0044] Then the grayscale value g(x′,y′) is obtained.

[0045] Obtaining the full-field displacement corresponding to any one of the deformed images according to the best matching position to obtain the full-field strain includes:

[0046] Taking the sub-region center point Q(x0, y0) in the reference image as the center, search in the deformed image to obtain the best matching position Q1(x0+u, y0+v);

[0047] According to the change of the deformed image at any point P1(x*,y*) in the sub-region, the full-field displacement is obtained, including:

[0048] The change of any point P1(x*, y*) in the deformed image in the sub-region is represented by the zero-order displacement characterization function P(x, y) in the reference image sub-region: Among them, u and v are the displacements in the x and y directions respectively;

[0049] The full-field strain is obtained by estimating the displacement ladder using the central difference method:

[0050]

[0051] in, and They represent the rate of change of displacement u in the x-direction and y-direction respectively, and Represents the rate of change of v in the x and y directions respectively; u(x,y) and v(x,y) represent the displacement along the x and y directions at the point (x,y); Δx and Δy are the step sizes; ε xx and ε yy represent the normal strain in the x and y directions, ε xy represents the shear strain.

[0052] The evaluating the speckle quality in the original speckle image by using the average intensity gradient includes:

[0053] Get the average grayscale gradient δ of the speckle image:

[0054]

[0055] Where W is the pixel width of the speckle pattern, H is the pixel height of the speckle pattern, and f x (x,y) represents the first-order derivative of the pixel point (x,y) in the x direction, f y (x,y) represents the first-order derivative of the pixel point (x,y) in the y direction.

[0056] The beneficial effects of this application include at least:

[0057] The method described in this application first acquires an original speckle image, including a reference image and at least one deformed image, and evaluates the speckle quality in the original speckle image using the average intensity gradient to obtain an evaluation result. Then, based on the evaluation result, the deformed image with lower speckle quality is deblurred to obtain a corresponding restored image of the deformed image. The reference image is then divided into at least two sub-regions using a virtual grid, and the correlation between any of the sub-regions and the deformed image is obtained to obtain the best matching position. Finally, the full-field displacement corresponding to any of the deformed images is obtained based on the best matching position to obtain the full-field strain, thereby improving the digital image range of the original speckle image. The method described in this application restores the original image by adaptively adjusting the filter parameters based on the blur type and intensity of the speckle. This method can stably implement speckle tracking using digital image correlation technology under defocus conditions, overcoming the speckle image correlation mismatch problem caused by limited range. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a flow chart of the method described in this application;

[0059] Figure 2 is the grayscale histogram;

[0060] Figure 3 It is a schematic diagram of the bilinear interpolation algorithm;

[0061] Figure 4 This is the result of digital image correlation strain calculation. DETAILED DESCRIPTION

[0062] The present application will be further described below with reference to the accompanying drawings.

[0063] Specific embodiment 1:

[0064] This application provides an embodiment:

[0065] like Figure 1 A defocused speckle deblurring method based on Wiener filtering specifically includes the following steps: S1. Speckle image production:

[0066] First, the white test sample is cut into the required shape and size and its surface is polished. Then, the surface is cleaned with water and organic solvent to enhance the sample's surface adhesion to the speckle. Finally, matte black paint is sprayed in spots to form a black and white contrasting speckle pattern on the surface. S2. Reference Image Acquisition:

[0067] Before applying a specific load to the sample, the matching image acquisition software is used to capture a high-contrast and clear speckle pattern as the initial condition of the sample in its natural state, and a reference image is also collected.

[0068] S3. Apply load:

[0069] When the camera's field of view remains unchanged, the sample is controlled to move at fixed intervals through a two-dimensional linear displacement platform, so that the camera can obtain the deformed speckle image and obtain the original speckle image.

[0070] S4. Deformation image acquisition:

[0071] During the sample loading process, the camera and its supporting image acquisition software are used to capture the sample surface several times at a fixed acquisition frame rate to obtain a series of deformation images.

[0072] S5. Speckle Quality Assessment:

[0073] After collecting a series of deformed images, the quality of these images is evaluated. The average grayscale gradient δ of the speckle image is calculated as follows:

[0074] Where W is the pixel width of the speckle pattern, H is the pixel height of the speckle pattern, and f x(x,y) represents the first-order derivative of the pixel point (x,y) in the x direction, f y (x,y) represents the first-order derivative of the pixel (x,y) in the y direction. A larger average intensity gradient indicates better speckle image quality. If this result meets the target, proceed directly to step 8 for subsequent calculations. If the result does not meet the target, perform step 7 to deblur the speckle image before continuing with subsequent calculations.

[0075] S6. Speckle Deblurring:

[0076] The specific steps are as follows:

[0077] S601. Initialize the blur kernel radius R and signal-to-noise ratio SNR:

[0078] Since the original clear image corresponding to the blurred image is unknown, the blur kernel size and signal-to-noise ratio cannot be directly determined. In this case, it is necessary to set a reasonable initial value. To facilitate the subsequent optimization of R and SNR, this patent selects a smaller R and a larger SNR as the initial values.

[0079] S602. Wiener filter deblurring:

[0080] Based on the initialized blur kernel radius R and signal-to-noise ratio (SNR), a blur model is first established. According to the characteristics of the imaging system, speckle images outside the depth of field are primarily affected by out-of-focus blur. For out-of-focus blur, the blur kernel is typically represented as a Gaussian function. Assuming the blur kernel is h, the blur model can be expressed as:

[0081] g=h*f+n (2)

[0082] Where g is the blurred image, f is the original image, * represents the convolution operation, and n is the noise. By performing Fourier transform, the blurred model can be expressed in the frequency domain as follows:

[0083] G=H·F+N (3)

[0084] Where G, H, F, and N represent the Fourier transform of g, h, f, and n respectively. The purpose of deblurring is to obtain an estimate of the original image, which can be expressed in the frequency domain as:

[0085] F'=H w ·G (4)

[0086] Where F' is the estimated value of the original image, H w is the transfer function of the Wiener filter, which can be simplified as:

[0087] Where H(f) is the Fourier transform of the blur kernel h, |H(f)2 | is the amplitude spectrum of the blur kernel, and SNR is the signal-to-noise ratio. Therefore, the estimated value of the original image can be expressed as:

[0088] Substitute the blur kernel radius R and signal-to-noise ratio SNR in step 1 into formula (5) to obtain the initial restored image F1.

[0089] S603. Optimization of blur kernel radius R and signal-to-noise ratio SNR:

[0090] In order to obtain a restored image that is clearer and more accurate than the initial restored image F1, it is necessary to optimize R and SNR. During the entire defocusing process, the blurred speckle image has a smaller grayscale histogram half-width than the clear speckle image. For the blurred speckle image obtained at the defocused position, the restored grayscale histogram half-width is close to the grayscale histogram half-width of the clear speckle image obtained within the focus. Therefore, the grayscale histogram half-width of the clear speckle image obtained within the focus is used as a criterion for restoring a blurred speckle image, so as to select the appropriate R and SNR. The half-width FWHM of the grayscale histogram is as follows: Figure 2 As shown, it can be expressed as:

[0091] FWHM=D right -D left (7)

[0092] Among them, D right and D left They are the grayscale values ​​corresponding to the first one on the right and left sides of the grayscale histogram that reaches or exceeds half the height of the peak.

[0093] The specific optimization process is as follows: ① Gradually increase the blur kernel radius R and reduce the signal-to-noise ratio SNR. Repeat step (6-Ⅱ) after each adjustment and calculate the half-height width of the grayscale histogram of the restored image. ② Compare the half-height width after each adjustment with the half-height width of the clear speckle image within the focal length. Select the R and SNR that make the half-height width close to the clear image as the optimal value to obtain the restored image F n .

[0094] S604. Adaptive local smoothing strategy to suppress ringing effect:

[0095] The above steps will result in the restored image F nObvious ringing will occur, manifested as alternating bright and dark stripes or banded structures near the edges of the speckle image. This phenomenon will cause distortion of image details and affect image quality. This is because Wiener filtering involves deconvolution of the image's blur kernel. If the blur kernel contains zero or near-zero frequency components, deconvolution may lead to amplification of high-frequency components, resulting in a ringing effect. The spectral characteristics of an image determine its components at different spatial frequencies. If the original image has strong mutations at high frequencies, ringing effects may also occur during the restoration process. An adaptive local smoothing strategy is used here to suppress the ringing effect. The adaptive local smoothing strategy effectively suppresses the ringing effect by combining spatial proximity and grayscale similarity weights to perform weighted averaging on each pixel.

[0096] The specific process is as follows:

[0097] S6041. Spatial proximity weight:

[0098] The spatial proximity weight is determined by calculating the spatial distance between each pixel and its neighboring pixels. The closer the pixel is to the center pixel, the greater its weight. The spatial proximity weight w s (q,p) can be expressed as:

[0099] Among them, p represents the center pixel, q represents the pixels in the area, σ d is the standard deviation of the spatial proximity weight. S6042. Gray similarity weight:

[0100] The grayscale similarity weight is determined by calculating the grayscale similarity between each pixel and its neighboring pixels. The closer the grayscale value of the pixel is to the center pixel, the greater its weight. r (q,p) can be expressed as:

[0101] Among them, I(p) and I(q) are the grayscale values ​​of the central pixel and the pixels in the neighborhood, respectively, and σ r is the standard deviation of the grayscale similarity weights.

[0102] S6043. Weighted average:

[0103] For each pixel, the weighted average of all pixels in the neighborhood is calculated using the above spatial proximity and grayscale similarity weights, which is used as the final value of the pixel. The pixel value I'(p) after ringing suppression processing can be expressed as:

[0104] Where N(p) represents the neighborhood of pixel p. After suppressing ringing, the final restored image F is obtained.

[0105] S7. Image search positioning:

[0106] Perform digital image correlation analysis on the restored image F. First, select a region of interest in the reference image as a search window and divide it into a virtual grid. Each subregion composed of the virtual grid is treated as a rigid motion. Then, use the normalized covariance cross-correlation function to calculate the correlation between the deformed image and each reference subregion. The closer the calculated result is to 1, the more accurate the matching result. The normalized covariance cross-correlation function can be expressed as:

[0107] Among them, f(x,y) and g(x′,y′) represent the grayscale values ​​corresponding to the image functions of the reference image and the deformed image at coordinates (x,y) and (x′,y′), respectively. Then, bilinear interpolation is used to search the speckle image at the sub-pixel level to obtain g(x′,y′), as follows: Figure 3 shown.

[0108] Assuming that f(x1,y1), f(x1,y2), f(x2,y2), and f(x2,y1) represent the grayscale values ​​corresponding to the image functions of the deformed image at coordinates (x1,y1), (x1,y2), (x2,y2), and f(x2,y1), respectively, the calculation process is as follows:

[0109] S701. Calculate the linear interpolation in the x′ direction:

[0110]

[0111] S702. Calculate the linear interpolation in the y′ direction:

[0112] Substituting formula (12) and formula (13) into formula (14) can obtain the grayscale value g(x′,y′).

[0113] S8. Displacement and strain field reconstruction:

[0114] Taking the center point Q(x0,y0) of the reference image sub-region as the center, the best matching position Q1(x0+u,y0+v) is obtained through the search in the previous step, where u and v are the displacements in the x and y directions respectively. * ,y * ) can be represented by the zero-order displacement characterization function in the reference image subregion P(x,y):

[0115] After calculating each sub-region, the full-field displacement can be obtained. Then the central difference method is used to estimate the displacement gradient to obtain the full-field strain, which is calculated as follows:

[0116]

[0117] in, and They represent the rate of change of displacement u in the x-direction and y-direction respectively, and Represents the rate of change of v in the x and y directions respectively; u(x,y) and v(x,y) represent the displacement along the x and y directions at the point (x,y); Δx and Δy are the step sizes; ε xx and ε yy represent the normal strain in the x and y directions, ε xy Represents shear strain. Application examples:

[0118] A two-dimensional displacement stage was used to move the sample from 0 mm to 6.0 mm, with speckle images captured every 0.5 mm. When the displacement exceeded 4.5 mm, the speckle image began to blur, resulting in poor speckle quality. To verify the feasibility of the proposed method, speckle images were analyzed over a depth range of 4.5 mm to 6.0 mm. Figure 4 The strain calculation results in the x-direction using digital image correlation technology are shown before and after deblurring the speckle image, and the errors in the two cases are compared. Figure 4 (a) shows the strain calculation results of the digital image without deblurring. It can be seen that as the displacement in the depth direction increases, the blurriness of the acquired speckle image becomes increasingly greater. When the displacement in the depth direction reaches 6.0 mm, the speckle image matching fails due to the large blurriness of the speckle image, and the strain cannot be calculated. Figure 4 After deblurring the speckle image using the method mentioned in this paper, the quality of the speckle image is improved, and the matching can be successful at the position where the original speckle image failed to match, as shown in (a-4). Figure 4 (b-4) Next, the error calculation is performed on the strain calculation results in the two cases, as shown in Figure 4 As shown in (c), it can be seen that after deblurring using the method of this patent, the error of the strain calculation results is reduced. When the depth displacement reaches 6.0 mm, a good strain calculation result can also be obtained, thus achieving the expansion of the depth range of digital image correlation.

[0119] Specific embodiment II:

[0120] This application also provides an embodiment:

[0121] An electronic device comprises: a storage medium and a processing unit; wherein the storage medium is used to store a computer program, and the processing unit exchanges data with the storage medium, and is configured to execute the computer program via the processing unit when performing defocused speckle deblurring to perform the steps of the defocused speckle deblurring method based on Wiener filtering as described in Specific Embodiment 1.

[0122] Specific embodiment III:

[0123] A computer-readable storage medium having a computer program stored therein; when the computer program is run, the computer program executes the steps of the defocused speckle deblurring method based on Wiener filtering as described in Specific Embodiment 1.

[0124] In this application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. Furthermore, in this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination thereof.

[0125] The above disclosures are only a few specific implementation scenarios of this application, but this application is not limited to them. Any changes that can be thought of by those skilled in the art should fall within the scope of protection of this application. The above application numbers are for descriptive purposes only and do not represent the advantages or disadvantages of the implementation scenarios.

Claims

1. A defocused speckle deblurring method based on Wiener filtering, characterized in that: include: Acquiring an original speckle image, including a reference image and at least one deformed image, and evaluating the speckle quality in the original speckle image by using an average intensity gradient to obtain an evaluation result; deblurring the deformed image with low speckle quality according to the evaluation result to obtain a restored image corresponding to the deformed image; Dividing the reference image into a virtual grid to obtain at least two sub-regions, and obtaining a correlation between any of the sub-regions and the deformed image to obtain a best matching position; Obtaining a full-field displacement corresponding to any of the deformed images according to the optimal matching position, further obtaining a full-field strain, and improving a digital image-related range of the original speckle image by comparing the strain results before and after deblurring; Deblurring the deformed image with low speckle quality according to the evaluation result to obtain a restored image corresponding to the deformed image includes: According to the evaluation result, the blur kernel radius and signal-to-noise ratio corresponding to the speckle in the original speckle image are initialized: Acquire an initial restored image according to the blur kernel radius and the signal-to-noise ratio; Setting a standard for restoring the blurred speckle image, adjusting the blur kernel radius and the signal-to-noise ratio, and obtaining a restored image corresponding to the deformed image; The obtaining of the initial restored image according to the blur kernel radius and the signal-to-noise ratio comprises: A blur model is established. The speckle image outside the depth of field range has defocus blur. For defocus blur, the blur kernel is represented by a Gaussian function. Assuming that the blur kernel radius is h, the blur model is expressed as: g=h*f+n; Where g is the blurred image, f is the original speckle image, * represents the convolution operation, and n is the noise; The estimated value of the original speckle image is expressed as: Where F' is the estimated value of the original speckle image, H is the Fourier transform of the blur kernel h, |H 2 ∣ is the amplitude spectrum of the blur kernel, SNR is the signal-to-noise ratio; G represents the Fourier transform of the blurred image g; The setting of the standard for restoring the blurred speckle image and adjusting the blur kernel radius and the signal-to-noise ratio include: The half-maximum width of the grayscale histogram of the clear speckle image is used as a criterion for restoring the blurred speckle image, thereby adjusting the blur kernel radius and the signal-to-noise ratio. The half-maximum width FWHM of the grayscale histogram is expressed as: <h2 style=";text-align:left;direction:ltr">FWHM=D<h2 style=";text-align:left;direction:ltr"> right <h2 style=";text-align:left;direction:ltr"> -D<h2 style=";text-align:left;direction:ltr"> left <h2 style=";text-align:left;direction:ltr"> ; Among them, D right and D left They are the grayscale values ​​corresponding to the first one on the right and left sides of the grayscale histogram that reaches or exceeds half the height of the peak; The evaluating the speckle quality in the original speckle image by using the average intensity gradient includes: Get the average grayscale gradient δ of the speckle image: Where W is the pixel width of the speckle pattern, H is the pixel height of the speckle pattern, and f x (x,y) represents the first-order derivative of the pixel point (x,y) in the x direction, f y (x,y) represents the first-order derivative of the pixel point (x,y) in the y direction.

2. The defocused speckle deblurring method based on Wiener filtering according to claim 1, characterized in that: The adjusting of the blur kernel radius and the signal-to-noise ratio comprises: Gradually increase the blur kernel radius R and reduce the signal-to-noise ratio SNR to obtain the half-height width of the grayscale histogram of the restored image; Comparing the half-height width of the grayscale histogram of the restored image with the half-height width of the clear speckle image within the focus to obtain a comparison result; According to the comparison results, the blur kernel radius R and the signal-to-noise ratio SNR are continuously adjusted until the half-maximum width of the grayscale histogram of the restored image reaches the half-maximum width of the grayscale histogram of the clear speckle image in focus. The blur kernel radius R and the signal-to-noise ratio SNR at this time are taken as the optimal values.

3. The defocused speckle deblurring method based on Wiener filtering according to claim 1, characterized in that: When adjusting the blur kernel radius and the signal-to-noise ratio, the following steps are also included: By combining spatial proximity and grayscale similarity weights, weighted averaging is performed on any pixel in the restored image to adjust the ringing effect caused by the blur kernel radius and signal-to-noise ratio, thereby obtaining the final restored image, including: Determining a spatial proximity weight by obtaining a spatial distance between any pixel in the restored image and pixels in its neighborhood; Determine the grayscale similarity weight by obtaining the grayscale value similarity between any pixel in the restored image and pixels in its neighborhood; For the pixel, the spatial proximity weight and the grayscale similarity weight are used to obtain a weighted average of all pixels in the neighborhood, which is used as the final value of the pixel to suppress the ringing effect and obtain the final restored image.

4. The defocused speckle deblurring method based on Wiener filtering according to claim 1, characterized in that: The step of dividing the reference image into a virtual grid to obtain at least two sub-regions, and obtaining a correlation between any of the sub-regions and the deformed image to obtain the best matching position includes: Selecting an area of ​​interest in the reference image as a search window and dividing the area into virtual grids to obtain at least two sub-areas; Treating the sub-region composed of the virtual grid as a rigid motion, and obtaining the correlation between any of the sub-regions and the deformed image; A bilinear interpolation method is used to search the speckle image at a sub-pixel level to obtain the grayscale value corresponding to the image function of the deformed image at the coordinate (x′, y′) until the best matching position is obtained.

5. The defocused speckle deblurring method based on Wiener filtering according to claim 4, characterized in that: The use of bilinear interpolation to search the speckle image at a sub-pixel level to obtain a grayscale value corresponding to the image function of the deformed image at the coordinate (x′, y′) includes: Assume that f(x1,y1), f(x1,y2), f(x2,y2), and f(x2,y1) represent the grayscale values ​​corresponding to the image functions of the deformed image at coordinates (x1,y1), (x1,y2), (x2,y2), and f(x2,y1), respectively; Get the linear interpolation in the x′ direction: Get the linear interpolation in the y′ direction: Then the grayscale value g(x′,y′) is obtained.

6. The defocused speckle deblurring method based on Wiener filtering according to claim 1, characterized in that: Obtaining the full-field displacement corresponding to any of the deformed images according to the best matching position, and further obtaining the full-field strain, includes: Taking the sub-region center point Q(x0, y0) in the reference image as the center, search in the deformed image to obtain the best matching position Q1(x0+u, y0+v); According to the change of the deformed image at any point P1(x*,y*) in the sub-region, the full-field displacement is obtained, including: The change of the deformed image at any point P1(x*,y*) in the sub-region is represented by the zero-order displacement characterization function P(x,y) in the reference image sub-region: Among them, u and v are the displacements in the x and y directions respectively; The full-field strain is obtained by estimating the displacement ladder using the central difference method: in, and They represent the rate of change of displacement u in the x-direction and y-direction respectively, and Represents the rate of change of v in the x and y directions respectively; u(x,y) and v(x,y) represent the displacement along the x and y directions at the point (x,y); Δx and Δy are the step sizes; ε xx and ε yy represent the normal strain in the x and y directions, ε xy represents the shear strain.

Images