Matching precision improving method and system, electronic equipment and storage medium

Abstract

The invention provides a matching precision improving method and system, electronic equipment and a storage medium. The method comprises the following steps: acquiring speckle images before and after surface deformation of an object to be measured to obtain a reference image and a deformation image; performing integer pixel matching on the reference image and the deformation image by adopting an SURF algorithm after a particle swarm optimization algorithm improves a search strategy so as to obtain deformation initial value estimation of the whole field of the image; and based on the obtained deformation initial value estimation of the whole image field, performing sub-pixel search matching on the reference image and the deformation image by adopting an IC-GN algorithm to obtain deformation displacement field information. According to the matching precision improving method, the SURF algorithm which is subjected to search strategy optimization through PSO is used in the whole pixel matching process of the laser speckle correlation method, the matching precision of the algorithm can be effectively improved on the premise that the calculation efficiency is guaranteed, and the robustness and adaptability of the laser speckle correlation method can be remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital image processing, and in particular relates to a matching accuracy improvement method, system, electronic equipment and storage medium. Background Art

[0002] At present, the laser speckle correlation method is an optical measurement method for measuring micro-strain of structures. Its basic principle is to collect speckle images before and after the deformation of the surface of the object to be measured, perform sub-pixel search and matching on the image, and return the displacement difference between each matching point to obtain the displacement field and strain distribution of the deformation. As an optical method with the advantages of non-contact, full-field measurement, simple experimental device and strong robustness, it has been increasingly used to measure the deformation of mechanical structures in recent years.

[0003] However, in actual measurement, the deformation process of the object to be measured is not continuous, that is, the process of taking images before and after deformation is separate, and the camera will be offset relative to the original position, that is, positioning error; in addition, although the use of laser irradiation to generate speckle patterns has the advantage of non-contact, the quality of laser speckle patterns is very dependent on the surface of the object to be measured. The surface of the object to be measured that is too smooth or has its own texture may have insufficient speckle grayscale information, low quality, or be covered by its own texture. These situations will cause the matching accuracy of the laser speckle correlation method to be greatly reduced when matching whole pixels (where whole pixel matching accuracy is the prerequisite for accurate sub-pixel matching). Traditional methods usually use camera-level image correction to reduce positioning errors, and use image enhancement methods to deal with insufficient speckle information. However, image correction is very susceptible to calibration parameter errors, and interpolation in correction will also cause errors in shape function parameters (where shape function parameters are important parameters for image sub-pixel self-matching), and image enhancement methods are less effective when speckle information is covered by deep textures.

[0004] Some researchers have tried to use some feature matching algorithms to optimize the whole pixel matching. Due to its scale-invariant characteristics, the feature matching algorithm can effectively avoid positioning errors. At the same time, the feature matching algorithm is based on the feature descriptor of the speckle pattern for matching. Even if the speckle grayscale quality is not very good, it can still be well matched if it has feature information. The more optimized feature matching algorithm is the SURF method (Speed ​​Up Robust Feature). Compared with the SIFT algorithm (Scale Invariant Feature Transform), its feature descriptor is reduced from 128 dimensions to 64 dimensions, and the KNN algorithm (K-Nearest Neighbors) is used to search and match the feature sub-dimensions. However, when the amount of data is large, the efficiency of searching using the KNN algorithm is still slow, and there are many matching error points, and the computational cost is high.

[0005] In summary, the current feature matching algorithm cannot meet the needs of actual engineering in essence, and cannot truly improve the matching accuracy of the digital speckle correlation method and apply it to actual engineering.

[0006] Therefore, how to provide a matching accuracy improvement method, system, electronic device and storage medium has become a technical problem that urgently needs to be solved in this field. Summary of the invention

[0007] The purpose of the present invention is to provide a matching accuracy improvement method, system, electronic device and storage medium.

[0008] According to a first aspect of the present invention, a matching accuracy improvement method is provided, the method comprising:

[0009] Step S1: acquiring speckle images of the surface of the object to be measured before and after deformation to obtain a reference image and a deformation image;

[0010] Step S3: using the SURF algorithm with improved search strategy by particle swarm optimization algorithm to perform integer pixel matching on the reference image and the deformation map to obtain an initial deformation estimate of the entire image field;

[0011] Step S4: Based on the initial deformation estimation of the entire image field, the IC-GN algorithm is used to perform sub-pixel search and matching on the reference image and the deformation map to obtain the displacement field information of the deformation.

[0012] Optionally, before step S3, the method further includes:

[0013] Step S2: using the CLAHE method to perform image enhancement processing on the reference image and the deformed image.

[0014] Optionally, the step S3 specifically includes:

[0015] Step S31: using a box filter to perform convolution processing on the reference image and the deformed image to construct a scale space of the image;

[0016] Step S32: Processing the scale space of the image to generate a SURF algorithm feature descriptor of the image;

[0017] Step S33: using the particle swarm optimization algorithm to perform matching processing on the SURF algorithm feature descriptor of the image to obtain an initial value estimation of the deformation of the entire image field.

[0018] Optionally, the step S32 specifically includes:

[0019] Step S321: Determine the feature key points from the scale space of the image using the Hessian matrix;

[0020] Step S322: using Haar wavelet response to assign a direction to each of the feature key points;

[0021] Step S323: Counting the response value and absolute value of the Haar wavelet in the pixel sub-region centered on each of the feature key points, and the obtained four-dimensional vector is used as the SURF algorithm feature descriptor of the image.

[0022] Optionally, the step S33 specifically includes:

[0023] Step S331: using the particle swarm optimization algorithm to search and match the SURF algorithm feature descriptor of the image to obtain the optimal particle parameters;

[0024] Step S332: determining the best matching point of the feature point of the reference image in the deformed image based on the optimal particle parameters to complete integer pixel matching.

[0025] Optionally, the step S4 specifically includes:

[0026] Step S41: Based on the obtained initial deformation estimation of the entire image field, the IC-GN algorithm is used to perform a sub-pixel search on the reference image and the deformation map to find the sub-pixel optimal matching point of each feature point of the reference image;

[0027] Step S42: Based on the optimal sub-pixel matching point of each feature point, the displacement difference between the matching point pairs is obtained, and the displacement difference between the matching point pairs is returned to the displacement field distribution of the entire deformation surface to obtain the deformation displacement field information.

[0028] Optionally, the step S41 further includes:

[0029] The deformable image and the reference image are divided into a plurality of deformable sub-regions and a plurality of reference sub-regions respectively, and a second-order shape function is selected to characterize the sub-region offset and deformation of the deformable image relative to the reference image.

[0030] According to a second aspect of the present invention, a matching accuracy improvement system is provided, the system comprising:

[0031] A first processing module is configured to obtain speckle images of the surface of the object to be measured before and after deformation to obtain a reference image and a deformation image;

[0032] The third processing module is configured to use a SURF algorithm with a search strategy improved by a particle swarm optimization algorithm to perform integer pixel matching on the reference image and the deformation map to obtain an initial value estimation of the deformation of the entire image field;

[0033] The fourth processing module is configured to, based on the obtained initial deformation estimation of the entire image field, use the IC-GN algorithm to perform sub-pixel search and matching on the reference image and the deformation map to obtain the displacement field information of the deformation.

[0034] According to a third aspect of the present invention, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps in the matching accuracy improvement method as described in any one of the first aspect of the present invention are implemented.

[0035] According to a third aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the matching accuracy improvement method as described in any one of the first aspect of the present invention are implemented.

[0036] The beneficial effects brought by the present invention are as follows:

[0037] It can be seen from the above scheme that the embodiments of the present invention provide a matching accuracy improvement method, system, electronic device and storage medium, which have the following beneficial effects:

[0038] The matching accuracy improvement method of the present invention uses the SURF algorithm after the search strategy optimization by PSO in the whole pixel matching process of the laser speckle correlation method, which can not only effectively improve the matching accuracy of the algorithm under the premise of ensuring the calculation efficiency, but also significantly increase the robustness and adaptability of the laser speckle correlation method, so that it can be more flexibly applied to the engineering practice of special deformation measurement such as various complex surfaces and large-scale deformations. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flow chart of a matching accuracy improvement method provided according to an embodiment Figure 1 ;

[0040] Figure 2 A flow chart of a matching accuracy improvement method provided according to an embodiment Figure 2 ;

[0041] Figure 3 A schematic diagram of the structure of a matching accuracy improvement system provided according to an embodiment;

[0042] Figure 4 A schematic diagram of an electronic device provided according to an embodiment. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Glossary:

[0045] SURF algorithm: The full name is Speeded Up Robust Features. It is an improvement on the SIFT algorithm. Its main features are fast speed and strong robustness.

[0046] IC-GN algorithm: Inverse synthetic Gauss-Newton method is another commonly used nonlinear optimization method for DIC technology.

[0047] CLAHE: Contrast Limited Adaptive Histogram Equalization, is an image enhancement technology that is mainly used to enhance the contrast of images while suppressing noise. It is an improvement on the histogram equalization (HE) algorithm and is particularly suitable for the processing of medical images and infrared images.

[0048] Hessian Matrix: Hessian Matrix, also translated as Hessian Matrix, Hessian Matrix, Hessian Matrix, etc., is a square matrix composed of the second-order partial derivatives of a multivariate function, which describes the local curvature of the function.

[0049] Embodiment 1: According to the first aspect of the present invention, a matching accuracy improvement method is provided. Figure 1 As shown, the method includes:

[0050] Step S1: acquiring speckle images of the surface of the object to be measured before and after deformation to obtain a reference image and a deformation image;

[0051] Step S3: Using the SURF algorithm with improved search strategy by particle swarm optimization algorithm to perform integer pixel matching on the reference image and the deformation map to obtain an initial deformation estimate of the entire image field;

[0052] Step S4: Based on the initial deformation estimation of the entire image field, the IC-GN algorithm is used to perform sub-pixel search and matching on the reference image and the deformation map to obtain the displacement field information of the deformation.

[0053] Optionally, the matching accuracy improvement method of the embodiment of the present invention further includes, before step S3: step S2: performing image enhancement processing on the reference image and the deformed image using the CLAHE method.

[0054] Specifically, the CLAHE method (Contrast Limited Adaptive Histogram Equalization) is an image enhancement algorithm, which is mainly used to enhance the contrast of an image while suppressing noise. The CLAHE method performs local enhancement processing on the speckle pattern based on the principle of block division. The block division processing is to divide the input image into multiple non-overlapping sub-blocks of equal size, each sub-block contains a certain number of pixels, and the common block size is a block of 8x8 pixels. Each sub-block is independently processed by histogram equalization to improve the local contrast. In the matching accuracy improvement method of this embodiment, by enhancing the speckle image, it can not only improve the problem of uneven image illumination, but also improve the average grayscale gradient of the speckle pattern, thereby improving the quality of the speckle.

[0055] Optionally, step S3 in the matching accuracy improvement method of the embodiment of the present invention specifically includes:

[0056] Step S31: using a box filter to perform convolution processing on the reference image and the deformed image to construct a scale space of the image;

[0057] Step S32: using the SURF algorithm to process the scale space of the image to generate a SURF algorithm feature descriptor of the image;

[0058] Step S33: using a particle swarm optimization algorithm to match the SURF algorithm feature descriptor of the image to obtain an initial value estimate of the deformation of the entire image.

[0059] Optionally, step S32 in the matching accuracy improvement method of the embodiment of the present invention specifically includes:

[0060] Step S321: Determine the feature key points from the scale space of the image using the Hessian matrix;

[0061] Step S322: using Haar wavelet response to assign a direction to each feature key point;

[0062] Step S323: Count the response value and absolute value of the Haar wavelet in the pixel sub-region centered on each feature key point, and the obtained four-dimensional vector is used as the SURF algorithm feature descriptor of the image.

[0063] It should be noted that the SURF algorithm is an improvement on the SIFT algorithm, and its main features are fast speed and strong robustness. The SURF algorithm mainly includes: Scale space extreme value detection: SURF uses the Hessian matrix to detect extreme points. The Hessian matrix is ​​a second-order derivative matrix, and its determinant can reflect the curvature of the image at that point. If the value of the determinant is greater than a certain threshold, the point is considered as a candidate feature point. Haar wavelet response (haar wavelet): Assign directions to candidate feature points. SURF uses the Haar wavelet operator to calculate the gradient of pixels around the candidate feature point, and the direction with the largest Haar wavelet response is considered as the direction of the feature point. Feature point positioning: Use interpolation and fitting to accurately locate the position and scale of feature points. The SURF algorithm uses quadratic interpolation to estimate the position and scale of feature points in different scale spaces, and then uses quadratic fitting to estimate the precise positioning of feature points at that position and scale.

[0064] Optionally, step S33 in the matching accuracy improvement method of the embodiment of the present invention specifically includes:

[0065] Step S331: using a particle swarm optimization algorithm to search and match the SURF algorithm feature descriptor of the image to obtain the optimal particle parameters;

[0066] Step S332: Determine the best matching point of the feature point of the reference image in the deformed image based on the optimal particle parameters to complete the whole pixel matching.

[0067] It should be noted that Particle Swarm Optimization (PSO) is an optimization algorithm based on swarm intelligence. It finds the best solution by moving a group of particles (representing potential solutions) in a multidimensional search space. Each particle has its own position and speed, and gradually approaches the best solution.

[0068] For details, see Figure 2 As shown, in this embodiment, the obtained reference image and deformed image are first processed by a box filter to construct the scale space of the image. In this embodiment, a box filter is used instead of a Gaussian filter to speed up the convolution of the image f(x,y); box filters of different sizes and different blur coefficients are used to convolve the image to generate image pyramids of different scales (composed of multiple groups, each group of multiple layers of images).

[0069] D xx(x,σ)=f(x+1,y)+f(x-1,y)-2f(x,y) (1)

[0070] The Hessian matrix can be approximately expressed as:

[0071] det(H)=D xx (x,σ)D yy (x,σ)-[0.9D xy (x,σ)] 2 (2)

[0072] Among them, D xx , D xy and D yy is the convolution of the image and the box filter at a certain point, x is the coordinate of a certain point P in the image, 0.9 is an empirical value, σ is the feature point scale, x represents the horizontal coordinate of the pixel point, and y represents the vertical coordinate of the pixel point. Image pyramids of different scales are constructed, consisting of multiple groups, each with multiple layers of images.

[0073] Secondly, generate the SURF algorithm feature descriptor of the image. Specifically including:

[0074] 1) Determine whether each pixel point processed by the Hessian matrix is ​​an extreme point, thereby obtaining a candidate key point, and then compare the 26 candidate key points in the 3×3×3 area of ​​this point. If it is still an extreme point in the area, it is determined to be a feature key point.

[0075] 2) Assign a direction to each key feature point, and describe the direction of the feature point using the Gaussian-weighted Haar wavelet response in the x and y directions in a circular area with the key feature point as the center and 6σ as the radius. Use a fan-shaped window with an opening angle of π / 3 to slide and count the Haar wavelet feature response in the fan-shaped area. After scanning a circle, take the direction of the maximum response as the main direction of the current feature point.

[0076] 3) Generate SURF algorithm feature descriptors, which are used to characterize the coordinate position and direction of the matching points. With the current feature key point as the center, the square area of ​​20σ is divided into 4×4 grid areas, and the Haar wavelet response characteristics of 5×5 pixels are counted in each area. Finally, the response value and its absolute value are counted, and a four-dimensional vector descriptor V = [∑dx ∑dy |∑dx| |∑dy|] can be obtained in each sub-area, that is, the vector of the feature descriptor is a 64-dimensional feature vector with 4×4×4 dimensions.

[0077] Finally, the PSO algorithm is used to match the image feature points. The PSO algorithm seeks the optimal solution by sharing information between individual particles in the group. The optimal solution at this time is the best matching point of the image. The specific process is:

[0078] 1) Initialize particles (here, particles are descriptors of certain feature points), set the total number of particles NP, and the properties of the i-th particle: speed v i and position x i , set the number of iterations N, and generate the first generation of particles, namely the initial particles.

[0079] 2) Calculate the fitness of the current generation of particles (the number of generations of the population in this iteration, called the current generation), and use the Euclidean distance between the particle and the reference image to be matched as the particle's fitness function. Calculate the fitness of each particle, compare it with the entire population, and record the position of the particle with the maximum fitness.

[0080] 3) Update the particle's speed and position according to the following formula:

[0081]

[0082] Among them, ω is the inertia weight, which indicates the ability of the particle to inherit velocity, c1 and c2 are learning factors, r1 and r2 are random numbers distributed between [0,1], t is the latest iteration number, and Pbest i is the optimal fitness position of the i-th particle, Gbest i is the optimal fitness position of the population in the global process.

[0083] 4) Iterate according to the set number of iterations and output the best particle position, so as to find the best matching point of the feature point in the deformed image and complete the whole pixel matching.

[0084] Optionally, step S4 in the matching accuracy improvement method of the embodiment of the present invention specifically includes:

[0085] Step S41: Based on the initial deformation estimation of the entire image, the IC-GN algorithm is used to perform a sub-pixel search on the reference image and the deformation map to find the sub-pixel optimal matching point of each feature point of the reference image;

[0086] Step S42: Based on the optimal sub-pixel matching point of each feature point, the displacement difference between the matching point pairs is obtained, and the displacement difference between the matching point pairs is returned to the displacement field distribution of the entire deformation surface to obtain the deformation displacement field information.

[0087] Optionally, step S41 in the matching accuracy improvement method of the embodiment of the present invention further includes:

[0088] The deformed image and the reference image are divided into a plurality of deformed sub-regions and a plurality of reference sub-regions respectively, and a second-order shape function is selected to characterize the sub-region offset and deformation of the deformed image relative to the reference image.

[0089] It should be noted that the IC-GN algorithm (Inverse Compositional Gauss-Newton method) in this embodiment is a nonlinear optimization method used for high-precision deformation measurement in digital image correlation technology.

[0090] For details, see Figure 2 As shown, the specific process of sub-pixel level matching in this embodiment is:

[0091] First, the sub-area offset mode during the speckle pattern search is selected to describe the deformation of the sub-area. The second-order shape function is selected to characterize the offset and deformation of the deformation image relative to the reference image sub-area. The second-order shape function can describe the displacement, stretching and distortion of the sub-area, which is suitable for various nonlinear deformations and large deformations, and has smaller system errors. It can be expressed as

[0092]

[0093] in, x o is the coordinate of the center point P of the reference sub-area, Represents the offset of the deformation sub-region point relative to the center point of the reference sub-region,

[0094] PSF 2 =[uu x u y u xx u xy u yy vv x v y v xx v xy v yy ] T

[0095] is the shape function parameter of the second-order shape function. The process of finding the best match is actually the process of solving the optimal shape function parameters.

[0096] To further understand the connotation of shape function parameters, a point in the known deformation sub-region The coordinates of can be written as the coordinates of the corresponding point in the reference sub-area (x, y) plus the offset in the X and Y directions. It can also be seen that the sub-area offset has a certain relationship with the coordinates of the corresponding point in the reference sub-area, so the following relationship can be obtained:

[0097]

[0098] Performing the second-order Taylor expansion on the two equations, we can obtain

[0099]

[0100] It can be seen that the shape function parameters are the coefficients of each term in the expansion.

[0101] Secondly, based on the initial deformation estimation of the entire image field obtained after integer pixel matching, the IC-GN algorithm is combined to perform sub-pixel level search and obtain the optimal matching point.

[0102] 1) Calculate the gradient of the image sub-region at pixel i and the Jacobian matrix of the 2nd-order shape function The formula is

[0103]

[0104] It is also the offset of the deformation sub-region point relative to the center point of the reference sub-region.

[0105] 2) Calculate the Hessian matrix H, and construct the shape function expression SF(W) of the deformed sub-area based on the initial deformation estimate obtained by the integer pixel matching algorithm, that is, the initial value of the shape function.

[0106]

[0107] 3) Calculate the increment ΔP of the shape function parameter, and iterate continuously based on the following formula until it is determined that ΔP is less than a certain infinitesimal threshold, at which time the best sub-pixel matching point is output.

[0108]

[0109] Among them, the Hssian matrix, Calculated, G(x o +W(Δx i ,P)) represents the deformation sub-area,

[0110] f m , g m are the average grayscale values ​​of the reference sub-area and the deformation sub-area respectively, and the image size is (2M+1)×(2M+1)

[0111]

[0112] Finally, based on the sub-pixel matching points found for each point, the displacement difference between the matching point pairs is obtained, thereby returning the displacement field distribution in the x and y directions of the entire deformation surface, and based on this, the strain field distribution is obtained.

[0113] Embodiment 2: According to the second aspect of the present invention, a matching accuracy improvement system 200 is provided. Figure 3 As shown, the system 200 includes:

[0114] The first processing module 201 is configured to obtain speckle images of the surface of the object to be measured before and after deformation to obtain a reference image and a deformation image;

[0115] The third processing module 203 is configured to use the SURF algorithm with improved search strategy of the particle swarm optimization algorithm to perform integer pixel matching on the reference image and the deformation map to obtain an initial value estimation of the deformation of the entire image field;

[0116] The fourth processing module 204 is configured to perform sub-pixel search and matching on the reference image and the deformation map using the IC-GN algorithm based on the obtained initial value estimation of the deformation of the entire image field, so as to obtain the displacement field information of the deformation.

[0117] Optionally, the matching accuracy improvement system of the embodiment of the present invention further includes: a second processing module 202 configured to perform image enhancement processing on the reference image and the deformed image by using a CLAHE method.

[0118] Optionally, the third processing module 203 in the matching accuracy improvement system of the embodiment of the present invention specifically includes:

[0119] A first processing unit is configured to perform convolution processing on the reference image and the deformed image using a box filter to construct a scale space of the image;

[0120] The second processing unit is configured to process the scale space of the image to generate a SURF algorithm feature descriptor of the image;

[0121] The third processing unit is configured to use a particle swarm optimization algorithm to perform matching processing on the SURF algorithm feature descriptor of the image to obtain an initial value estimation of the deformation of the entire image field.

[0122] Optionally, the second processing unit in the matching accuracy improvement system of an embodiment of the present invention is specifically configured to use the Hessian matrix to determine the feature key points from the scale space of the image; use the Haar wavelet response to assign a direction to each feature key point; and count the response value and absolute value of the Haar wavelet in the pixel sub-region centered on each feature key point, and the obtained four-dimensional vector is used as the SURF algorithm feature descriptor of the image.

[0123] Optionally, the third processing unit in the matching accuracy improvement system of an embodiment of the present invention is specifically configured to use a particle swarm optimization algorithm to search and match the SURF algorithm feature descriptor of the image to obtain the optimal particle parameters; based on the optimal particle parameters, determine the best matching point of the feature point of the reference image in the deformed image to complete integer pixel matching.

[0124] Optionally, the fourth processing module 204 in the matching accuracy improvement system of the embodiment of the present invention specifically includes:

[0125] The fourth processing unit is configured to, based on the obtained initial deformation estimation of the entire image field, use the IC-GN algorithm to perform a sub-pixel search on the reference image and the deformation map to find the sub-pixel optimal matching point of each feature point of the reference image;

[0126] The fifth processing unit is configured to obtain the displacement difference between the matching point pairs based on the sub-pixel level optimal matching point of each feature point, and return the displacement difference between the matching point pairs to the displacement field distribution of the entire deformation surface to obtain the deformation displacement field information.

[0127] Optionally, the fourth processing unit in the matching accuracy improvement system of an embodiment of the present invention is also configured to divide the deformed image and the reference image into multiple deformed sub-areas and multiple reference sub-areas, respectively, and select a second-order shape function to characterize the sub-area offset and deformation of the deformed image relative to the reference image.

[0128] The matching accuracy improvement method of the embodiment of the present invention adopts the SURF algorithm to perform whole pixel matching in the digital speckle correlation method. Due to its scale invariance and the principle based on feature matching, it can not only well avoid the matching inaccuracy caused by the positioning error of the image, but also perform matching based on feature information when the speckle grayscale quality is not good. In addition, the use of the SURF algorithm also enables the digital speckle correlation algorithm to be applied to the measurement of various large-scale deformation processes such as rotation and distortion; the PSO algorithm is used to optimize the search strategy of the SURF algorithm, which not only greatly improves the matching efficiency of the SURF algorithm, but also reduces the wrong matching points to a certain extent. In this way, the computational efficiency and matching accuracy of the optimized SURF algorithm for whole pixel matching are greatly improved compared with the past, and this improved laser speckle correlation algorithm can be applied to engineering practice.

[0129] In summary, the matching accuracy improvement method of the embodiment of the present invention uses the SURF algorithm after the search strategy optimization by PSO in the whole pixel matching process of the laser speckle correlation method, which can not only effectively improve the matching accuracy of the algorithm under the premise of ensuring the calculation efficiency, but also significantly increase the robustness and adaptability of the laser speckle correlation method, so that it can be more flexibly applied to the engineering practice of special deformation measurement such as various complex surfaces and large-scale deformations.

[0130] Embodiment 3: The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps in any one of the matching accuracy improvement methods disclosed in the first aspect of the present invention are implemented.

[0131] Figure 4 is a structural diagram of an electronic device according to an embodiment of the present invention, such as Figure 4As shown, the electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, near field communication (NFC) or other technologies. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device can be a touch layer covered on the display screen, or a button, a trackball or a touch pad set on the housing of the electronic device, or an external keyboard, touch pad or mouse, etc.

[0132] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a structural diagram of the part related to the technical solution of the present disclosure, and does not constitute a limitation on the electronic device to which the technical solution of the present application is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0133] Embodiment 4: The fourth aspect of the present invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any one of the matching accuracy improvement methods disclosed in the first aspect of the present invention are implemented.

[0134] The above are preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A matching accuracy improvement method, characterized in that: include: The method comprises: Step S1: acquiring speckle images of the surface of the object to be measured before and after deformation to obtain a reference image and a deformation image; Step S3: using the SURF algorithm with improved search strategy using the particle swarm optimization algorithm to perform integer pixel matching on the reference image and the deformation map to obtain an initial deformation estimate of the entire image field; Step S4: Based on the initial value estimation of the deformation of the entire image field, the IC-GN algorithm is used to perform sub-pixel search and matching on the reference image and the deformation map to obtain the displacement field information of the deformation.

2. The matching accuracy improvement method according to claim 1, characterized in that: Before step S3, the method further includes: Step S2: using the CLAHE method to perform image enhancement processing on the reference image and the deformed image.

3. The matching accuracy improvement method according to claim 1, characterized in that: The step S3 specifically includes: Step S31: using a box filter to perform convolution processing on the reference image and the deformed image to construct a scale space of the image; Step S32: Processing the scale space of the image to generate a SURF algorithm feature descriptor of the image; Step S33: using the particle swarm optimization algorithm to perform matching processing on the SURF algorithm feature descriptor of the image to obtain an initial value estimation of the deformation of the entire image field.

4. The matching accuracy improvement method according to claim 3, characterized in that: The step S32 specifically includes: Step S321: Determine the feature key points from the scale space of the image using the Hessian matrix; Step S322: using Haar wavelet response to assign a direction to each of the feature key points; Step S323: Counting the response value and absolute value of the Haar wavelet in the pixel sub-region centered on each of the feature key points, and the obtained four-dimensional vector is used as the SURF algorithm feature descriptor of the image.

5. The matching accuracy improvement method according to claim 3, characterized in that: The step S33 specifically includes: Step S331: using the particle swarm optimization algorithm to search and match the SURF algorithm feature descriptor of the image to obtain the optimal particle parameters; Step S332: determining the best matching point of the feature point of the reference image in the deformed image based on the optimal particle parameters to complete integer pixel matching.

6. The matching accuracy improvement method according to claim 1, characterized in that: The step S4 specifically includes: Step S41: Based on the obtained initial deformation estimation of the entire image field, the IC-GN algorithm is used to perform a sub-pixel search on the reference image and the deformation map to find the sub-pixel optimal matching point of each feature point of the reference image; Step S42: Based on the optimal sub-pixel matching point of each feature point, the displacement difference between the matching point pairs is obtained, and the displacement difference between the matching point pairs is returned to the displacement field distribution of the entire deformation surface to obtain the deformation displacement field information.

7. The matching accuracy improvement method according to claim 6, characterized in that: The step S41 further includes: The deformable image and the reference image are divided into a plurality of deformable sub-regions and a plurality of reference sub-regions respectively, and a second-order shape function is selected to characterize the sub-region offset and deformation of the deformable image relative to the reference image.

8. A matching accuracy improvement system, characterized in that: The system comprises: A first processing module is configured to obtain speckle images of the surface of the object to be measured before and after deformation to obtain a reference image and a deformation image; The second processing module is configured to use a SURF algorithm with a search strategy improved by a particle swarm optimization algorithm to perform integer pixel matching on the reference image and the deformation map to obtain an initial value estimation of the deformation of the entire image field; The third processing module is configured to use the IC-GN algorithm to perform sub-pixel search and matching on the reference image and the deformation map based on the initial value estimation of the deformation of the entire image field, so as to obtain the displacement field information of the deformation.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps in a matching accuracy improvement method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in a matching accuracy improvement method described in any one of claims 1 to 7 are implemented.

Images