A strain measurement system and method based on in-situ image processing

Abstract

The present invention discloses a strain measurement system and method based on in-situ image processing. The system includes: an acquisition module for acquiring a reference image and an image to be corrected; an extraction module for respectively extracting feature points of the reference image and the image to be corrected to obtain reference feature points and feature points to be corrected; a matching module for matching the reference feature points and the feature points to be corrected to obtain a plurality of matching point pairs; a processing module for calculating an image rotation angle and an image displacement according to the positions of the reference feature points and the feature points to be corrected in the plurality of matching point pairs; and a correction module for correcting the image to be corrected according to the image rotation angle and the image displacement to obtain an in-situ image. The technical solution provided by the present invention has the advantages of high automation degree, high precision and high processing efficiency, and can effectively correct the offset displacement and skew angle of a digital image, making up for the deficiencies of related research.

Description

Technical Field

[0001] The present invention relates to the technology of in-plane displacement correction of digital images, and in particular to a strain measurement system and method based on in-situ image processing. Background Art

[0002] Material failure seriously threatens the safety of aeroengines, and material deformation is a direct manifestation of its mechanical behavior. Studying the deformation behavior at the mesoscopic scale of materials can provide strong support for clarifying the material failure mechanism. DIC (Digital Image Correlation) is a strain measurement method with advantages such as full-field measurement, high accuracy, low requirements for the measurement environment, and simple measurement process. It is widely used in current mechanical measurements and is a suitable strain measurement method for mesoscopic and microscopic scales. However, due to the systematic error of the in-situ loading system, a large offset will occur in the observation area, affecting the DIC displacement and strain measurement accuracy. In addition, in fatigue tests that require SEM (Scanning Electron Microscope) observation, there is a situation of repeatedly disassembling and assembling specimens and observing fatigue cracks under the SEM. It is very difficult to ensure that the position of the observation area remains unchanged in the SEM images collected before and after the test.

[0003] Therefore, it is necessary to develop an in-situ image processing method to eliminate the digital image offset displacement and skew angle caused by the systematic error of the in-situ loading system and the repeated disassembly and assembly of specimens. Summary of the Invention

[0004] The purpose of the present invention is to provide a strain measurement system and method based on in-situ image processing to solve the problem of accurately correcting the offset displacement and skew angle of images.

[0005] To achieve the above purpose, the present invention provides a strain measurement system based on in-situ image processing. The system includes: an acquisition module for acquiring a reference image and an image to be corrected; an extraction module for respectively extracting feature points of the reference image and the image to be corrected to obtain reference feature points and feature points to be corrected; a matching module for matching the reference feature points and the feature points to be corrected to obtain a plurality of matching point pairs; a processing module for calculating the image rotation angle and image displacement according to the positions of the reference feature points and the feature points to be corrected in the plurality of matching point pairs; and a correction module for correcting the image to be corrected according to the image rotation angle and the image displacement to obtain an in-situ image.

[0006] Preferably, the correction module includes: a rotation correction unit for correcting the image to be corrected according to the image rotation angle to obtain an intermediate corrected image; a displacement correction unit for correcting the intermediate corrected image according to the image displacement to obtain the in-situ image; the extraction module includes: a rotation extraction unit for respectively extracting the feature points of the reference image and the image to be corrected to obtain a reference feature point for rotation and a to-be-corrected feature point for rotation; a displacement extraction unit for respectively extracting the feature points of the reference image and the intermediate corrected image to obtain a reference feature point for displacement and a to-be-corrected feature point for displacement; the matching module includes: a rotation matching unit for matching the reference feature point for rotation and the to-be-corrected feature point for rotation to obtain a plurality of rotation matching point pairs; a displacement matching unit for matching the reference feature point for displacement and the to-be-corrected feature point for displacement to obtain a plurality of displacement matching point pairs; wherein, the processing module is further configured to calculate the image rotation angle according to the positions of the reference feature point for rotation and the to-be-corrected feature point for rotation in the plurality of rotation matching point pairs, and calculate the image displacement according to the positions of the reference feature point for displacement and the to-be-corrected feature point for displacement in the plurality of displacement matching point pairs.

[0007] Preferably, the processing module further includes: a sorting unit for: sorting the reference feature point for rotation and the to-be-corrected feature point for rotation in the plurality of rotation matching point pairs, and the sorting signs of the reference feature point for rotation and the to-be-corrected feature point for rotation in the same rotation matching point pair are the same. Wherein, two reference feature points for rotation with adjacent sorting signs form a reference line segment, and two to-be-corrected feature points for rotation with adjacent sorting signs form a to-be-corrected line segment, respectively obtaining a plurality of reference line segments and a plurality of to-be-corrected line segments, and the plurality of reference line segments and the plurality of to-be-corrected line segments correspond one by one; sorting the reference feature point for displacement and the to-be-corrected feature point for displacement in the plurality of displacement matching point pairs, and the sorting signs of the reference feature point for displacement and the to-be-corrected feature point for displacement in the same displacement matching point pair are the same; a rotation angle calculation unit for respectively calculating the rotation angle between the corresponding reference line segment and the to-be-corrected line segment to obtain a plurality of line segment rotation angles, and averaging the plurality of line segment rotation angles to obtain the image rotation angle; and a displacement calculation unit for calculating the offset displacement of the to-be-corrected feature point for displacement relative to the reference feature point for displacement in the same displacement matching point pair to obtain a plurality of feature point displacements, and averaging the plurality of feature point displacements to obtain the image displacement.

[0008] Preferably, the rotation matching unit is further configured to: calculate the Euclidean distances between each rotation-correction target feature point and all rotation-reference feature points respectively, obtain a plurality of rotation Euclidean distances corresponding to each rotation-correction target feature point, and select the rotation-reference feature point and the rotation-correction target feature point corresponding to the minimum value among the plurality of rotation Euclidean distances for each rotation-correction target feature point as the rotation matching point pair; or calculate the Euclidean distances between each rotation-correction target feature point and all rotation-reference feature points respectively, obtain a plurality of rotation Euclidean distances corresponding to each rotation-correction target feature point, calculate the rotation ratio of the minimum value and the second minimum value among the corresponding plurality of rotation Euclidean distances for each rotation-correction target feature point respectively, and use the rotation-reference feature point and the rotation-correction target feature point corresponding to the rotation ratio less than the predetermined rotation threshold as the rotation matching point pair.

[0009] Preferably, the displacement matching unit is further configured to: calculate the Euclidean distances between each displacement-correction target feature point and all displacement-reference feature points respectively, obtain a plurality of displacement Euclidean distances corresponding to each displacement-correction target feature point, and select the displacement-reference feature point and the displacement-correction target feature point corresponding to the minimum value among the plurality of displacement Euclidean distances for each displacement-correction target feature point as the displacement matching point pair; or calculate the Euclidean distances between each displacement-correction target feature point and all displacement-reference feature points respectively, obtain a plurality of displacement Euclidean distances corresponding to each displacement-correction target feature point, calculate the displacement ratio of the minimum value and the second minimum value among the corresponding plurality of displacement Euclidean distances for each displacement-correction target feature point respectively, and use the displacement-reference feature point and the displacement-correction target feature point corresponding to the displacement ratio less than the predetermined displacement threshold as the displacement matching point pair.

[0010] Correspondingly, the present invention further provides a strain measurement method based on image in-situ processing, the method comprising: acquiring a reference image and a correction target image; respectively extracting feature points of the reference image and the correction target image to obtain reference feature points and correction target feature points; matching the reference feature points and the correction target feature points to obtain a plurality of matching point pairs; calculating an image rotation angle and an image displacement according to the positions of the reference feature points and the correction target feature points in the plurality of matching point pairs; and correcting the correction target image according to the image rotation angle and the image displacement to obtain an in-situ image.

[0011] Preferably, the strain measurement method based on in-situ image processing provided by the present invention further includes: respectively extracting the feature points of the reference image and the image to be corrected to obtain the reference feature points for rotation and the to-be-corrected feature points for rotation; matching the reference feature points for rotation and the to-be-corrected feature points for rotation to obtain a plurality of matching point pairs for rotation; calculating the image rotation angle according to the positions of the reference feature points for rotation and the to-be-corrected feature points for rotation in the plurality of matching point pairs for rotation; correcting the image to be corrected according to the image rotation angle to obtain an intermediate corrected image; respectively extracting the feature points of the reference image and the intermediate corrected image to obtain the reference feature points for displacement and the to-be-corrected feature points for displacement; matching the reference feature points for displacement and the to-be-corrected feature points for displacement to obtain a plurality of matching point pairs for displacement; calculating the image displacement according to the positions of the reference feature points for displacement and the to-be-corrected feature points for displacement in the plurality of matching point pairs for displacement; and correcting the intermediate corrected image according to the image displacement to obtain the in-situ image.

[0012] Preferably, calculating the image rotation angle and the image displacement according to the positions of the reference feature points and the to-be-corrected feature points in the plurality of matching point pairs includes: sorting the reference feature points for rotation and the to-be-corrected feature points for rotation in the plurality of matching point pairs for rotation, where the sorting signs of the reference feature points for rotation and the to-be-corrected feature points for rotation in the same matching point pair for rotation are the same. Among them, two adjacent reference feature points for rotation with the same sorting sign form a reference line segment, and two adjacent to-be-corrected feature points for rotation with the same sorting sign form a to-be-corrected line segment, respectively obtaining a plurality of reference line segments and a plurality of to-be-corrected line segments, and the plurality of reference line segments and the plurality of to-be-corrected line segments correspond one by one; respectively calculating the rotation angles between the corresponding reference line segments and to-be-corrected line segments to obtain a plurality of line segment rotation angles, and averaging the plurality of line segment rotation angles to obtain the image rotation angle; sorting the reference feature points for displacement and the to-be-corrected feature points for displacement in the plurality of matching point pairs for displacement, where the sorting signs of the reference feature points for displacement and the to-be-corrected feature points for displacement in the same matching point pair for displacement are the same; and calculating the offset displacement of the to-be-corrected feature points for displacement relative to the reference feature points for displacement in the same matching point pair for displacement to obtain a plurality of feature point displacements, and averaging the plurality of feature point displacements to obtain the image displacement.

[0013] Preferably, matching the reference feature points for rotation and the feature points to be corrected for rotation to obtain a plurality of matching point pairs for rotation, including: calculating the Euclidean distance between each feature point to be corrected for rotation and all reference feature points for rotation respectively to obtain a plurality of Euclidean distances for rotation corresponding to each feature point to be corrected for rotation, and selecting the reference feature point for rotation and the feature point to be corrected for rotation corresponding to the minimum value among the plurality of Euclidean distances for rotation for each feature point to be corrected for rotation as the matching point pair for rotation; or calculating the Euclidean distance between each feature point to be corrected for rotation and all reference feature points for rotation respectively to obtain a plurality of Euclidean distances for rotation corresponding to each feature point to be corrected for rotation, calculating the rotation ratio of the minimum value and the second minimum value among the corresponding plurality of Euclidean distances for rotation for each feature point to be corrected for rotation respectively, and taking the reference feature point for rotation and the feature point to be corrected for rotation corresponding to the rotation ratio less than the predetermined rotation threshold as the matching point pair for rotation.

[0014] Preferably, matching the reference feature points for displacement and the feature points to be corrected for displacement to obtain a plurality of matching point pairs for displacement, including: calculating the Euclidean distance between each feature point to be corrected for displacement and all reference feature points for displacement respectively to obtain a plurality of Euclidean distances for displacement corresponding to each feature point to be corrected for displacement, and selecting the reference feature point for displacement and the feature point to be corrected for displacement corresponding to the minimum value among the plurality of Euclidean distances for displacement for each feature point to be corrected for displacement as the matching point pair for displacement; or calculating the Euclidean distance between each feature point to be corrected for displacement and all reference feature points for displacement respectively to obtain a plurality of Euclidean distances for displacement corresponding to each feature point to be corrected for displacement, calculating the displacement ratio of the minimum value and the second minimum value among the corresponding plurality of Euclidean distances for displacement for each feature point to be corrected for displacement respectively, and taking the reference feature point for displacement and the feature point to be corrected for displacement corresponding to the displacement ratio less than the predetermined displacement threshold as the matching point pair for displacement.

[0015] The present invention matches the feature points extracted from the reference image and the image to be corrected, calculates the image rotation angle and image displacement according to the positions of the reference feature points and the feature points to be corrected in the matching point pairs, and corrects the image to be corrected accordingly to obtain the in-situ image. The technical solution provided by the present invention has the advantages of high automation degree, high precision and high processing efficiency, can effectively correct the offset displacement and skew angle of digital images, and makes up for the deficiencies of related research. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings are used to provide a further understanding of the embodiments of the present invention, and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation to the embodiments of the present invention. In the drawings:

[0017] Figure 1 is a block diagram of a strain measurement system based on in-situ image processing provided by the present invention;

[0018] Figure 2 is a block diagram of another strain measurement system based on in-situ image processing provided by the present invention;

[0019] Figure 3 is an illustration of the in-situ process of the image to be corrected; and

[0020] Figure 4 is a flowchart of a strain measurement method based on in-situ image processing provided by the present invention. Detailed Embodiments

[0021] The following describes in detail the specific embodiments of the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining and illustrating the present invention, and are not used to limit the scope of the present invention.

[0022] Figure 1 is a block diagram of a strain measurement system based on in-situ image processing provided by the present invention, as Figure 1 shown, the system includes an acquisition module 10, an extraction module 20, a matching module 30, a processing module 40, and a correction module 50.

[0023] The acquisition module 10 is used to acquire a reference image and an image to be corrected. The reference image is an image acquired before loading an image into the in-situ loading system, and the image to be corrected is an image acquired after loading an image into the in-situ loading system.

[0024] The extraction module 20 is used to extract feature points from the reference image and the image to be corrected respectively, to obtain reference feature points and feature points to be corrected. The SURF (Speeded Up Robust Features) feature point extraction algorithm can be used to extract the feature points of the reference image and the image to be corrected. Here, the number of the obtained reference feature points and feature points to be corrected can be the same or different. For example, 502 feature points can be obtained by extracting the feature points in the reference image, and 492 feature points can be obtained by extracting the feature points in the image to be corrected.

[0025] The matching module 30 is used to match the reference feature points and the feature points to be corrected to obtain a plurality of matching point pairs. The brute force (BF) matching can be used to match the feature points of the reference image and the image to be corrected, and the number of the obtained matching point pairs should be less than or equal to the smaller value of the number of reference feature points and the number of feature points to be corrected.

[0026] The processing module 40 is used to calculate the image rotation angle and the image displacement according to the positions of the reference feature points and the feature points to be corrected in multiple matching point pairs. The positions of the feature points can be, for example, the feature point coordinates. It should be understood that the reference feature points and the feature points to be corrected should be measured with the same coordinates.

[0027] The correction module 50 is used to correct the image to be corrected according to the image rotation angle and the image displacement to obtain an in-situ image. Specifically, the image to be corrected is rotationally corrected according to the image rotation angle, and the image to be corrected can be translationally corrected according to the image displacement, so as to obtain an in-situ image.

[0028] Figure 2 is a block diagram of another strain measurement system based on image in-situ processing provided by the present invention. As Figure 2 shown, the extraction module 20 includes an extraction unit 21 for rotation and an extraction unit 22 for displacement. The matching module 30 includes a matching unit 31 for rotation and a matching unit 32 for displacement. The correction module 50 includes a rotation correction unit 51 and a displacement correction unit 52. The specific implementation manners will be described below.

[0029] The correction module 50 includes: a rotation correction unit 51, which is used to correct the image to be corrected according to the image rotation angle to obtain an intermediate corrected image; a displacement correction unit 52, which is used to correct the intermediate corrected image according to the image displacement to obtain an in-situ image. The extraction module 20 includes: an extraction unit 21 for rotation, which is used to extract the feature points of the reference image and the image to be corrected respectively to obtain the reference feature points for rotation and the feature points to be corrected for rotation; an extraction unit 22 for displacement, which is used to extract the feature points of the reference image and the intermediate corrected image respectively to obtain the reference feature points for displacement and the feature points to be corrected for displacement. The matching module 30 includes: a matching unit 31 for rotation, which is used to match the reference feature points for rotation and the feature points to be corrected for rotation to obtain multiple matching point pairs for rotation; a matching unit 32 for displacement, which is used to match the reference feature points for displacement and the feature points to be corrected for displacement to obtain multiple matching point pairs for displacement; wherein, the processing module 40 is further used to calculate the image rotation angle according to the positions of the reference feature points for rotation and the feature points to be corrected for rotation in multiple matching point pairs for rotation, and calculate the image displacement according to the positions of the reference feature points for displacement and the feature points to be corrected for displacement in multiple matching point pairs for displacement.

[0030] In the technical solution provided by the present invention, the reference image is corrected twice to obtain an in-situ image. The first correction is to correct the image to be corrected according to the image rotation angle to obtain an intermediate corrected image, and the second correction is to correct the intermediate corrected image obtained by the first correction according to the image displacement to obtain an in-situ image. The first correction is performed by the rotation correction unit 51, and the second correction is performed by the displacement correction unit 52.

[0031] The process of the first correction includes: the rotation extraction unit 21 extracts the feature points of the reference image and the image to be corrected, obtaining the rotation reference feature points and the rotation to-be-corrected feature points; the rotation matching unit 31 matches the rotation reference feature points and the rotation to-be-corrected feature points, obtaining the rotation matching point pairs; the processing module 40 calculates the image rotation angle according to the positions of the rotation reference feature points and the rotation to-be-corrected feature points in multiple rotation matching point pairs; and the rotation correction unit 51 corrects the image to be corrected according to the image rotation angle, obtaining the intermediate corrected image.

[0032] The process of the second correction includes: the displacement extraction unit 22 extracts the feature points of the reference image and the intermediate corrected image, obtaining the displacement reference feature points and the displacement to-be-corrected feature points; the displacement matching unit 32 matches the displacement reference feature points and the displacement to-be-corrected feature points, obtaining the displacement matching point pairs; the processing module 40 calculates the image displacement according to the positions of the displacement reference feature points and the displacement to-be-corrected feature points in multiple displacement matching point pairs; and the displacement correction unit 52 corrects the intermediate corrected image according to the image displacement, obtaining the in-situ image.

[0033] Both the first correction and the second correction need to extract the feature points of the reference image. The first correction extracts the rotation reference feature points from the feature points of the reference image through the rotation extraction unit 21, and the second correction extracts the displacement reference feature points from the feature points of the reference image through the displacement extraction unit 22. It should be noted that in the process of the second correction, it is also possible not to extract the feature points of the reference image, and directly use the rotation reference feature points obtained in the process of the first correction. That is to say, the extraction of the rotation reference feature points from the feature points of the reference image by the rotation extraction unit 21 and the extraction of the displacement reference feature points from the feature points of the reference image by the displacement extraction unit 22 can be one operation (that is, only extract the feature points of the reference image once, but use them in both the first correction and the second correction) or two separate operations (that is, extract the feature points of the reference image twice).

[0034] As Figure 2 shown, the processing module 40 further includes a sorting unit 41, a rotation angle calculation unit 42, and a displacement calculation unit 43.

[0035] The sorting unit 41 is configured to: sort the rotation reference feature points and the rotation to-be-corrected feature points in multiple rotation matching point pairs, where the sorting symbols of the rotation reference feature point and the rotation to-be-corrected feature point in the same rotation matching point pair are the same. Among them, two adjacent rotation reference feature points with sorting symbols form a reference line segment, and two adjacent rotation to-be-corrected feature points with sorting symbols form a to-be-corrected line segment, respectively obtaining multiple reference line segments and multiple to-be-corrected line segments, and the multiple reference line segments and the multiple to-be-corrected line segments correspond one by one.

[0036] Specifically, a rotation matching point pair includes a rotation reference feature point and a rotation to-be-corrected feature point. Suppose there are N (N is a positive integer) rotation matching point pairs, including the first rotation matching point pair, the second rotation matching point pair, the third rotation matching point pair,..., the Nth rotation matching point pair. Then, the rotation reference feature point in the first rotation matching point pair is the first rotation reference feature point, and the rotation to-be-corrected feature point in the first rotation matching point pair is the first rotation to-be-corrected feature point. The rotation reference feature point in the second rotation matching point pair is the second rotation reference feature point, and the rotation to-be-corrected feature point in the second rotation matching point pair is the second rotation to-be-corrected feature point, and so on. More specifically, for example, the rotation reference feature point is represented by A. Then, the first rotation reference feature point in the first rotation matching point pair is represented by A1, the second rotation reference feature point in the second rotation matching point pair is represented by A2,..., and the Nth rotation reference feature point in the Nth rotation matching point pair is represented by A N N. Correspondingly, the rotation to-be-corrected feature point is represented by B. Then, the first rotation to-be-corrected feature point in the first rotation matching point pair is represented by B1, the second rotation to-be-corrected feature point in the second rotation matching point pair is represented by B2,..., and the Nth rotation to-be-corrected feature point in the Nth rotation matching point pair is represented by B N N. Here, the subscripts 1, 2,..., N can be understood as sorting symbols. Then, A1 and A2 can form the first reference line segment l A1 =A1A2, A2 and A3 can form the second reference line segment l A2 =A2A3,..., A N-1 N-1 and A N N can form the (N-1)th reference line segment l AN-1 =A N-1 N-1 N AN, B1 and B2 can form the first to-be-corrected line segment l B1 =B1B2, B2 and B3 can form the second to-be-corrected line segment l B2 =B2B3,..., B N-1 N-1 and B N N can form the (N-1)th to-be-corrected line segment l BN-1= B N-1 B N , where the first reference line segment l A1 = A1A2 corresponds to the first line segment to be corrected l B1 = B1B2, the second reference line segment l A2 = A2A3 corresponds to the second line segment to be corrected l B2 = B2B3, ……, the (N - 1)-th reference line segment l AN-1 = A N-1 A N corresponds to the (N - 1)-th line segment to be corrected l BN-1 = B N-1 B N .

[0037] The sorting unit 41 is further configured to: sort the reference feature points for displacement and the to-be-corrected feature points for displacement in multiple displacement matching point pairs, and the sorting symbols of the reference feature points for displacement and the to-be-corrected feature points for displacement in the same displacement matching point pair are consistent.

[0038] Specifically, a displacement matching point pair includes a reference feature point for displacement and a to-be-corrected feature point for displacement. Suppose there are P (P is a positive integer) displacement matching point pairs, including the first displacement matching point pair, the second displacement matching point pair, the third displacement matching point pair, ……, the P-th displacement matching point pair. Then, the reference feature point for displacement in the first displacement matching point pair is the first reference feature point for displacement, the to-be-corrected feature point for displacement in the first displacement matching point pair is the first to-be-corrected feature point for displacement, the reference feature point for displacement in the second displacement matching point pair is the second reference feature point for displacement, and the to-be-corrected feature point for displacement in the second displacement matching point pair is the second to-be-corrected feature point for displacement, and so on. More specifically, for example, the reference feature point for displacement is represented by R. Then, the first reference feature point for displacement in the first displacement matching point pair is represented by R1, the second reference feature point for displacement in the second displacement matching point pair is represented by R2, ……, the P-th reference feature point for displacement in the P-th displacement matching point pair is represented by R P is represented, and correspondingly, the to-be-corrected feature point for displacement is represented by S. Then, the first to-be-corrected feature point for displacement in the first displacement matching point pair is represented by S1, the second to-be-corrected feature point for displacement in the second displacement matching point pair is represented by S2, ……, the P-th to-be-corrected feature point for displacement in the P-th displacement matching point pair is represented by S P is represented, and so on. Here, the subscripts 1, 2, ……, P can be understood as sorting symbols.

[0039] The rotation angle calculation unit 42 is configured to calculate the rotation angles between the corresponding reference line segments and the to-be-corrected line segments respectively, obtain a plurality of line segment rotation angles, and average the plurality of line segment rotation angles to obtain the image rotation angle.

[0040] Specifically, calculate the rotation angle between each pair of corresponding reference line segments and line segments to be corrected. In the case of having N - 1 pairs of corresponding reference line segments and line segments to be corrected, N - 1 rotation angles will be calculated. Then, the average value of these N - 1 rotation angles is the image rotation angle. For example, the feature point coordinates of the reference feature point A1 for the first rotation are (x A1 , y A1 ), the feature point coordinates of the reference feature point A2 for the second rotation are (x A2 , y A2 ), ……, the feature point coordinates of the reference feature point A N for the Nth rotation are (x AN , y AN ), the feature point coordinates of the feature point B1 to be corrected for the first rotation are (x B1 , y B1 ), the feature point coordinates of the feature point B2 to be corrected for the second rotation are (x B2 , y B2 ), ……, the feature point coordinates of the feature point B N to be corrected for the Nth rotation are (x BN , y BN ). The rotation angle between the first reference line segment l A1 = A1A2 and the first line segment to be corrected l B1 = B1B2 is represented by Δθ1, the rotation angle between the second reference line segment l A2 = A2A3 and the second line segment to be corrected l B2 = B2B3 is represented by Δθ2, ……, the rotation angle between the jth (j = 1, 2, ……, N) reference line segment l Aj = A j A j+1 and the jth line segment to be corrected l Bj = B j B j+1 is represented by Δθ j , ……, the rotation angle between the (N - 1)th reference line segment l AN-1 = A N-1 A N and the (N - 1)th line segment to be corrected l BN-1 = B N-1 B N is represented by Δθ N-1 . It should be noted that the rotation angle in this embodiment is the rotation angle of the line segment to be corrected relative to the reference line segment (for example, Δθ1 represents the rotation angle of the first line segment to be corrected l B1 = B1B2 relative to the first reference line segment l A1 = A1A2). Then, the image rotation angle can be calculated by formula (1):

[0041]

[0042] In formula (1), represents the image rotation angle (i.e., the average value of the rotation angles of all line segments to be corrected relative to the reference line segment), and Δθ j represents the rotation angle of the j-th line segment to be corrected relative to the j-th reference line segment.

[0043] After obtaining the image rotation angle in this embodiment, the image to be corrected can be rotated (Regarding the positive and negative values of the rotation angle, it can be determined according to the actual situation) to obtain the rotation-corrected image, that is, the intermediate-corrected image.

[0044] The displacement calculation unit 43 is used to calculate the offset displacement of the displacement to-be-corrected feature points relative to the displacement reference feature points in the same displacement matching point pairs, obtain multiple feature point displacements, and average the multiple feature point displacements to obtain the image displacement.

[0045] Specifically, calculate the offset displacement of the displacement to-be-corrected feature points relative to the displacement reference feature points in each displacement matching point pair. According to the feature point coordinates of the displacement to-be-corrected feature points and the feature point coordinates of the displacement reference feature points in each displacement matching point pair, the offset displacement of the displacement to-be-corrected feature points relative to the displacement reference feature points can be calculated. More specifically, the offset displacement of the first displacement to-be-corrected feature point S1 relative to the first displacement reference feature point R1 is represented by Δd1(x S1 -x R1 ,y S1 -y R1 ), the offset displacement of the second displacement to-be-corrected feature point S2 relative to the second displacement reference feature point S2 is represented by Δd2(x S2 -x R2 ,y S2 -y R2 ), ……, the offset displacement of the g-th (g = 1, 2, ……, P) displacement to-be-corrected feature point R j relative to the g-th displacement reference feature point S g is represented by Δd g (x Sg -x Rg ,y Sg -y Rg ), and the offset displacement of the P-th displacement to-be-corrected feature point S P relative to the P-th displacement reference feature point R P is represented by Δd P (x SP -x RP ,y SP -y RP) indicates that, then, the image displacement can be calculated by formula (2):

[0046]

[0047] In formula (2), represents the image displacement (i.e., all displacements are the average of the offset displacements of the feature points to be corrected relative to the reference feature points for displacement), Δx g represents the offset displacement x of the g-th feature point to be corrected relative to the g-th reference feature point for displacement in the x-axis direction Sg -x Rg , Δy g represents the offset displacement y of the g-th feature point to be corrected relative to the g-th reference feature point for displacement in the y-axis direction Sg -y Rg .

[0048] After obtaining the image displacement , in this embodiment, the rotated-corrected image (i.e., the intermediate-corrected image) can be translated (Regarding the positive and negative values of the displacement for translation, they can be determined according to the actual situation), to obtain the image after rotation correction and translation correction, i.e., the in-situ image.

[0049] It should be noted that the value of g and the value of j may be the same or different.

[0050] The rotation matching unit 31 is further configured to: respectively calculate the Euclidean distances between each rotated feature point to be corrected and all rotated reference feature points, obtain a plurality of rotated Euclidean distances corresponding to each rotated feature point to be corrected, and select the rotated reference feature point and the rotated feature point to be corrected corresponding to the minimum value among the plurality of rotated Euclidean distances for each rotated feature point to be corrected as the rotated matching point pairs; or respectively calculate the Euclidean distances between each rotated feature point to be corrected and all rotated reference feature points, obtain a plurality of rotated Euclidean distances corresponding to each rotated feature point to be corrected, and respectively calculate the rotated ratio of the minimum value and the second minimum value among the corresponding plurality of rotated Euclidean distances for each rotated feature point to be corrected, and use the rotated reference feature point and the rotated feature point to be corrected corresponding to the rotated ratio less than the rotated predetermined threshold as the rotated matching point pairs.

[0051] Briefly speaking, first calculate the Euclidean distances between each rotated feature point to be corrected and all rotated reference feature points, then determine the rotated matching point pairs according to the magnitudes of the obtained Euclidean distances, and the number of finally determined rotated matching point pairs should be less than or equal to the smaller value between the number of rotated reference feature points and the number of rotated feature points to be corrected.

[0052] This embodiment provides two methods for the rotation matching unit 31 to calculate the rotation matching point pairs: The first method is to calculate the Euclidean distance between each rotation to-be-corrected feature point and all rotation reference feature points respectively, obtaining multiple rotation Euclidean distances corresponding to each rotation to-be-corrected feature point. For each rotation to-be-corrected feature point, select the rotation reference feature point and the rotation to-be-corrected feature point corresponding to the minimum value among the multiple rotation Euclidean distances as the rotation matching point pair; The second method is to calculate the Euclidean distance between each rotation to-be-corrected feature point and all rotation reference feature points respectively, obtaining multiple rotation Euclidean distances corresponding to each rotation to-be-corrected feature point. For each rotation to-be-corrected feature point, calculate the rotation ratio of the minimum value and the second minimum value among the corresponding multiple rotation Euclidean distances respectively. The rotation reference feature point and the rotation to-be-corrected feature point corresponding to the rotation ratio less than the rotation predetermined threshold are used as the rotation matching point pair.

[0053] Still taking this embodiment as an example, the rotation reference feature points are represented by A. There are M (M is a positive integer and M ≥ N) rotation reference feature points in the reference image, which are A1, A2,..., A m , where 1 ≤ m ≤ M, and the feature vector of A m is represented by a m (a m1 , a m2 , a m3 ,..., a mk ). There are Q (Q is a positive integer and Q ≥ N) rotation to-be-corrected feature points in the to-be-corrected image, which are B1, B2,..., B q , where 1 ≤ q ≤ Q, and the feature vector of B q is represented by b q (b q1 , b q2 , b q3 ,..., b qk ). It can be understood that k is the dimension of the feature vector of the feature points (including rotation reference feature points and rotation to-be-corrected feature points). Then, the Euclidean distance d mq is calculated respectively between the feature vectors of each rotation to-be-corrected feature point in the to-be-corrected image and all rotation feature points in the reference image. The calculation formula is as follows:

[0054]

[0055] In formula (3), k is the dimension of the feature vector.

[0056] The first method for the rotation matching unit 31 to calculate the rotation matching point pair is as follows: Through formula (3), for the rotation to-be-corrected feature point B q , calculate the Euclidean distance dmq , the obtained d mq constitutes an Euclidean distance array {d mq}, and all elements in {d mq} are sorted by size. The smallest d mq corresponding point A m is the point in the reference image that best matches the B q point, that is, the smallest d mq corresponding rotation reference feature point and rotation to-be-corrected feature point are used as rotation matching point pairs.

[0057] The second way for the rotation matching unit 31 to calculate the rotation matching point pairs is as follows: Through formula (3), for the rotation to-be-corrected feature point B q , calculate the Euclidean distance d mq , the obtained d mq constitutes an Euclidean distance array {d mq}, and all elements in {d mq} are sorted by size. Select the minimum value (d mq ) mq and the second smallest value (d min ) mq in {d 2ndmin}, and calculate the distance ratio of the minimum value (d mq ) min to the second smallest value (d mq ) 2ndmin , that is, (d mq ) min / (d mq ) 2ndmin . And set a predetermined threshold, which can be set to 0.2. When (d mq ) min / (d mq ) 2ndmin is less than 0.2, the rotation reference feature point and rotation to-be-corrected feature point corresponding to d mq are used as rotation matching point pairs.

[0058] The displacement matching unit 32 is further configured to: calculate the Euclidean distance between each displacement to-be-corrected feature point and all displacement reference feature points respectively, obtain a plurality of displacement Euclidean distances corresponding to each displacement to-be-corrected feature point, and select the displacement reference feature point and the displacement to-be-corrected feature point corresponding to the minimum value among the plurality of displacement Euclidean distances for each displacement to-be-corrected feature point as a displacement matching point pair; or calculate the Euclidean distance between each displacement to-be-corrected feature point and all displacement reference feature points respectively, obtain a plurality of displacement Euclidean distances corresponding to each displacement to-be-corrected feature point, calculate the displacement ratio of the minimum value and the second minimum value among the corresponding plurality of displacement Euclidean distances for each displacement to-be-corrected feature point respectively, and use the displacement reference feature point and the displacement to-be-corrected feature point corresponding to the displacement ratio less than the displacement predetermined threshold as a displacement matching point pair.

[0059] Here, two methods for the displacement matching unit 32 to calculate the displacement matching point pair are given. In this application, the two methods for the displacement matching unit 32 to calculate the displacement matching point pair are similar to the two methods for the rotation matching unit 31 to calculate the rotation matching point pair, and thus will not be elaborated here.

[0060] Figure 3 It is a diagram of the process of in-situ processing of the to-be-corrected image provided by the present invention. As Figure 3 shown, the whole process includes:

[0061] Step S301: Extract the rotation reference feature points and the rotation to-be-corrected feature points.

[0062] Step S302: Perform feature point matching on the rotation reference feature points and the rotation to-be-corrected feature points to obtain rotation matching point pairs.

[0063] Step S303: Calculate the image rotation angle, that is, calculate the image rotation angle according to the positions of the rotation reference feature points and the rotation to-be-corrected feature points in the rotation matching point pairs.

[0064] Step S304: Rotate the to-be-corrected image, that is, rotate the to-be-corrected image according to the image rotation angle to obtain an intermediate corrected image.

[0065] Step S305: Extract the displacement reference feature points and the displacement to-be-corrected feature points.

[0066] Step S306: Perform feature point matching on the displacement reference feature points and the displacement to-be-corrected feature points to obtain displacement matching point pairs.

[0067] Step S307: Calculate the image displacement, that is, calculate the image displacement according to the positions of the displacement reference feature points and the displacement to-be-corrected feature points in the displacement matching point pairs.

[0068] Step S308, translate the image to be corrected, and perform translational correction according to the intermediate corrected image obtained in Step S304 to obtain an in-situ image.

[0069] Figure 4 is a flowchart of the strain measurement method based on image in-situ processing provided by the present invention. As Figure 4 shown, the method includes:

[0070] Step S401, obtain a reference image and an image to be corrected;

[0071] Step S402, respectively extract the feature points of the reference image and the image to be corrected to obtain reference feature points and feature points to be corrected;

[0072] Step S403, match the reference feature points and the feature points to be corrected to obtain multiple pairs of matching points;

[0073] Step S404, calculate the image rotation angle and the image displacement according to the positions of the reference feature points and the feature points to be corrected in multiple pairs of matching points;

[0074] Step S405, correct the image to be corrected according to the image rotation angle and the image displacement to obtain an in-situ image.

[0075] Among them, the strain measurement method based on image in-situ processing provided by the present invention further includes: respectively extracting the feature points of the reference image and the image to be corrected to obtain reference feature points for rotation and feature points to be corrected for rotation; matching the reference feature points for rotation and the feature points to be corrected for rotation to obtain multiple pairs of matching points for rotation; calculating the image rotation angle according to the positions of the reference feature points for rotation and the feature points to be corrected for rotation in multiple pairs of matching points for rotation; correcting the image to be corrected according to the image rotation angle to obtain an intermediate corrected image; respectively extracting the feature points of the reference image and the intermediate corrected image to obtain reference feature points for displacement and feature points to be corrected for displacement; matching the reference feature points for displacement and the feature points to be corrected for displacement to obtain multiple pairs of matching points for displacement; calculating the image displacement according to the positions of the reference feature points for displacement and the feature points to be corrected for displacement in multiple pairs of matching points for displacement; correcting the intermediate corrected image according to the image displacement to obtain an in-situ image.

[0076] Among them, calculating the image rotation angle and the image displacement according to the positions of the reference feature points and the feature points to be corrected in the multiple matching point pairs includes: sorting the reference feature points for rotation and the feature points to be corrected for rotation in the multiple matching point pairs for rotation, where the sorting signs of the reference feature points for rotation and the feature points to be corrected for rotation in the same matching point pair for rotation are the same. Among them, two adjacent reference feature points for rotation with different sorting signs form a reference line segment, and two adjacent feature points to be corrected for rotation with different sorting signs form a line segment to be corrected, respectively obtaining a plurality of reference line segments and a plurality of line segments to be corrected, and the plurality of reference line segments correspond to the plurality of line segments to be corrected one by one; respectively calculating the rotation angles between the corresponding reference line segments and the line segments to be corrected, obtaining a plurality of line segment rotation angles, and averaging the plurality of line segment rotation angles to obtain the image rotation angle; sorting the reference feature points for displacement and the feature points to be corrected for displacement in the multiple matching point pairs for displacement, where the sorting signs of the reference feature points for displacement and the feature points to be corrected for displacement in the same matching point pair for displacement are the same; and calculating the offset displacement of the feature points to be corrected for displacement relative to the reference feature points for displacement in the same matching point pair for displacement, obtaining a plurality of feature point displacements, and averaging the plurality of feature point displacements to obtain the image displacement.

[0077] Among them, matching the reference feature points for rotation and the feature points to be corrected for rotation to obtain a plurality of matching point pairs for rotation includes: respectively calculating the Euclidean distances between each feature point to be corrected for rotation and all the reference feature points for rotation, obtaining a plurality of Euclidean distances for rotation corresponding to each feature point to be corrected for rotation, and selecting the reference feature point for rotation and the feature point to be corrected for rotation corresponding to the minimum value among the plurality of Euclidean distances for rotation for each feature point to be corrected for rotation as the matching point pair for rotation; or respectively calculating the Euclidean distances between each feature point to be corrected for rotation and all the reference feature points for rotation, obtaining a plurality of Euclidean distances for rotation corresponding to each feature point to be corrected for rotation, and respectively calculating the rotation ratios of the minimum value and the second minimum value among the corresponding plurality of Euclidean distances for rotation for each feature point to be corrected for rotation, and taking the reference feature point for rotation and the feature point to be corrected for rotation corresponding to the rotation ratio less than the predetermined rotation threshold as the matching point pair for rotation.

[0078] Among them, matching the displacement reference feature points and the displacement to-be-corrected feature points to obtain a plurality of displacement matching point pairs includes: calculating the Euclidean distance between each displacement to-be-corrected feature point and all displacement reference feature points respectively to obtain a plurality of displacement Euclidean distances corresponding to each displacement to-be-corrected feature point, and selecting the displacement reference feature point and the displacement to-be-corrected feature point corresponding to the minimum value among the plurality of displacement Euclidean distances for each displacement to-be-corrected feature point as the displacement matching point pair; or calculating the Euclidean distance between each displacement to-be-corrected feature point and all displacement reference feature points respectively to obtain a plurality of displacement Euclidean distances corresponding to each displacement to-be-corrected feature point, calculating the displacement ratio of the minimum value and the second minimum value among the corresponding plurality of displacement Euclidean distances for each displacement to-be-corrected feature point respectively, and using the displacement reference feature point and the displacement to-be-corrected feature point corresponding to the displacement ratio less than the displacement predetermined threshold as the displacement matching point pair.

[0079] It should be noted that the specific details and benefits of the strain measurement method based on image in-situ processing provided by the present invention are similar to those of the strain measurement system based on image in-situ processing provided by the present invention, and will not be elaborated here.

[0080] The optional implementation manners of the embodiments of the present invention have been described in detail above with reference to the drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation manners. Within the technical concept scope of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention.

[0081] Through the technical solution provided by the present invention, the automatic in-situ correction of digital images is well realized, which has the advantages of high automation degree, high precision and high processing efficiency, and can effectively correct the offset displacement and skew angle of digital images. After obtaining the in-situ image, the reference image and the corrected in-situ image can be imported into the Digital Image Correlation (DIC) software, so that the strain generated by the object to be measured can be calculated.

[0082] In addition, it should be noted that, in the case of no contradiction, the various specific technical features described in the above specific implementation manners can be combined in any suitable manner. To avoid unnecessary repetition, the embodiments of the present invention will not separately describe various possible combination manners.

[0083] In addition, any combination can be made among various different implementation manners of the embodiments of the present invention, as long as it does not violate the idea of the embodiments of the present invention, and it should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A strain measurement system based on in-situ image processing, characterized in that, The system includes: An acquisition module for acquiring a reference image and an image to be corrected; An extraction module for respectively extracting feature points of the reference image and the image to be corrected to obtain reference feature points and feature points to be corrected; A matching module for matching the reference feature points and the feature points to be corrected to obtain a plurality of matching point pairs; A processing module for calculating an image rotation angle and an image displacement according to the positions of the reference feature points and the feature points to be corrected in the plurality of matching point pairs; and A correction module for correcting the image to be corrected according to the image rotation angle and the image displacement to obtain an in-situ image, including: A rotation correction unit for correcting the image to be corrected according to the image rotation angle to obtain an intermediate corrected image; A displacement correction unit for correcting the intermediate corrected image according to the image displacement to obtain an in-situ image; The extraction module includes: A rotation extraction unit for respectively extracting feature points of the reference image and the image to be corrected to obtain rotation reference feature points and rotation feature points to be corrected; A displacement extraction unit for respectively extracting feature points of the reference image and the intermediate corrected image to obtain displacement reference feature points and displacement feature points to be corrected; The matching module includes: A rotation matching unit for matching the rotation reference feature points and the rotation feature points to be corrected to obtain a plurality of rotation matching point pairs; A displacement matching unit for matching the displacement reference feature points and the displacement feature points to be corrected to obtain a plurality of displacement matching point pairs; Wherein, the processing module is further configured to calculate the image rotation angle according to the positions of the rotation reference feature points and the rotation feature points to be corrected in the plurality of rotation matching point pairs, and calculate the image displacement according to the positions of the displacement reference feature points and the displacement feature points to be corrected in the plurality of displacement matching point pairs, and Wherein, the processing module further includes: A sorting unit for: Sorting the rotation reference feature points and the rotation feature points to be corrected in the plurality of rotation matching point pairs, the sorting signs of the rotation reference feature points and the rotation feature points to be corrected in the same rotation matching point pair are the same, wherein, two adjacent rotation reference feature points with sorting signs form a reference line segment, and two adjacent rotation feature points to be corrected with sorting signs form a to-be-corrected line segment, respectively obtaining a plurality of reference line segments and a plurality of to-be-corrected line segments, and the plurality of reference line segments and the plurality of to-be-corrected line segments correspond one by one; Sorting the displacement reference feature points and the displacement feature points to be corrected in the plurality of displacement matching point pairs, and the sorting signs of the displacement reference feature points and the displacement feature points to be corrected in the same displacement matching point pair are the same; A rotation angle calculation unit for respectively calculating the rotation angles between the corresponding reference line segments and the to-be-corrected line segments to obtain a plurality of line segment rotation angles, and averaging the plurality of line segment rotation angles to obtain the image rotation angle; and A displacement calculation unit for calculating the offset displacement of the displacement feature points to be corrected relative to the displacement reference feature points in the same displacement matching point pair to obtain a plurality of feature point displacements, and averaging the plurality of feature point displacements to obtain the image displacement.

2. The strain measurement system based on in-situ image processing according to claim 1, characterized in that, The rotation matching unit is further configured to: Calculate the Euclidean distances between each to-be-corrected feature point for rotation and all reference feature points for rotation respectively, obtaining multiple Euclidean distances for rotation corresponding to each to-be-corrected feature point for rotation. For each to-be-corrected feature point for rotation, select the reference feature point for rotation and the to-be-corrected feature point for rotation corresponding to the minimum value among the multiple Euclidean distances for rotation as the matching point pair for rotation; Or Calculate the Euclidean distances between each to-be-corrected feature point for rotation and all reference feature points for rotation respectively, obtaining multiple Euclidean distances for rotation corresponding to each to-be-corrected feature point for rotation. For each to-be-corrected feature point for rotation, calculate the ratio of the minimum value and the second minimum value among the corresponding multiple Euclidean distances for rotation respectively. The reference feature point for rotation and the to-be-corrected feature point for rotation corresponding to the ratio for rotation less than the predetermined threshold for rotation are used as the matching point pair for rotation.

3. The strain measurement system based on in-situ image processing according to claim 1, characterized in that, The displacement matching unit is further configured to: Calculate the Euclidean distances between each to-be-corrected feature point for displacement and all reference feature points for displacement respectively, obtaining multiple Euclidean distances for displacement corresponding to each to-be-corrected feature point for displacement. For each to-be-corrected feature point for displacement, select the reference feature point for displacement and the to-be-corrected feature point for displacement corresponding to the minimum value among the multiple Euclidean distances for displacement as the matching point pair for displacement; or Calculate the Euclidean distances between each to-be-corrected feature point for displacement and all reference feature points for displacement respectively, obtaining multiple Euclidean distances for displacement corresponding to each to-be-corrected feature point for displacement. For each to-be-corrected feature point for displacement, calculate the ratio of the minimum value and the second minimum value among the corresponding multiple Euclidean distances for displacement respectively. The reference feature point for displacement and the to-be-corrected feature point for displacement corresponding to the ratio for displacement less than the predetermined threshold for displacement are used as the matching point pair for displacement.

4. A strain measurement method based on in-situ image processing, characterized in that, The method includes: Obtain a reference image and an image to be corrected; Extract the feature points of the reference image and the image to be corrected respectively, obtaining reference feature points and to-be-corrected feature points; Match the reference feature points and the to-be-corrected feature points to obtain multiple matching point pairs; Calculate the image rotation angle and the image displacement according to the positions of the reference feature points and the to-be-corrected feature points in the multiple matching point pairs; and Correct the image to be corrected according to the image rotation angle and the image displacement to obtain an in-situ image, The method further includes: Extract the feature points of the reference image and the image to be corrected respectively, obtaining reference feature points for rotation and to-be-corrected feature points for rotation; Match the reference feature points for rotation and the to-be-corrected feature points for rotation to obtain multiple matching point pairs for rotation; Calculate the image rotation angle according to the positions of the reference feature points for rotation and the to-be-corrected feature points for rotation in the multiple matching point pairs for rotation; Correct the image to be corrected according to the image rotation angle to obtain an intermediate corrected image; Extract the feature points of the reference image and the intermediate corrected image respectively, obtaining reference feature points for displacement and to-be-corrected feature points for displacement; Match the reference feature points for displacement and the to-be-corrected feature points for displacement to obtain multiple matching point pairs for displacement; Calculate the image displacement according to the positions of the reference feature points for displacement and the to-be-corrected feature points for displacement in the multiple matching point pairs for displacement; Correct the intermediate corrected image according to the image displacement to obtain the in-situ image, wherein calculating the image rotation angle and the image displacement according to the positions of the reference feature points and the feature points to be corrected in multiple matching point pairs includes: Sort the reference feature points for rotation and the feature points to be corrected for rotation in multiple matching point pairs for rotation. The sorting signs of the reference feature points for rotation and the feature points to be corrected for rotation in the same matching point pair for rotation are the same. Among them, the reference feature points for rotation of two adjacent sorting signs form a reference line segment, and the feature points to be corrected for rotation of two adjacent sorting signs form a line segment to be corrected, respectively obtaining multiple reference line segments and multiple line segments to be corrected. The multiple reference line segments and the multiple line segments to be corrected correspond one by one; Calculate the rotation angles between the corresponding reference line segments and the line segments to be corrected respectively to obtain multiple line segment rotation angles, and average the multiple line segment rotation angles to obtain the image rotation angle; Sort the reference feature points for displacement and the feature points to be corrected for displacement in multiple matching point pairs for displacement. The sorting signs of the reference feature points for displacement and the feature points to be corrected for displacement in the same matching point pair for displacement are the same; and Calculate the offset displacement of the feature points to be corrected for displacement relative to the reference feature points for displacement in the same matching point pair for displacement to obtain multiple feature point displacements, and average the multiple feature point displacements to obtain the image displacement.

5. The strain measurement method based on in-situ image processing according to claim 4, wherein, Matching the reference feature points for rotation and the feature points to be corrected for rotation to obtain multiple matching point pairs for rotation includes: Calculate the Euclidean distances between each feature point to be corrected for rotation and all the reference feature points for rotation respectively to obtain multiple Euclidean distances for rotation corresponding to each feature point to be corrected for rotation. For each feature point to be corrected for rotation, select the reference feature point for rotation and the feature point to be corrected for rotation corresponding to the minimum value among the multiple Euclidean distances for rotation as the matching point pair for rotation; or Calculate the Euclidean distances between each feature point to be corrected for rotation and all the reference feature points for rotation respectively to obtain multiple Euclidean distances for rotation corresponding to each feature point to be corrected for rotation. For each feature point to be corrected for rotation, calculate the rotation ratio of the minimum value and the second minimum value among the corresponding multiple Euclidean distances for rotation respectively. The reference feature point for rotation and the feature point to be corrected for rotation corresponding to the rotation ratio less than the predetermined rotation threshold are used as the matching point pair for rotation.

6. The strain measurement method based on in-situ image processing according to claim 4, wherein, Matching the reference feature points for displacement and the feature points to be corrected for displacement to obtain multiple matching point pairs for displacement includes: Calculate the Euclidean distances between each feature point to be corrected for displacement and all the reference feature points for displacement respectively to obtain multiple Euclidean distances for displacement corresponding to each feature point to be corrected for displacement. For each feature point to be corrected for displacement, select the reference feature point for displacement and the feature point to be corrected for displacement corresponding to the minimum value among the multiple Euclidean distances for displacement as the matching point pair for displacement; or Calculate the Euclidean distance between each displacement's to-be-corrected feature point and all displacements' reference feature points respectively, obtaining multiple displacements' Euclidean distances corresponding to each displacement's to-be-corrected feature point. For each displacement's to-be-corrected feature point, calculate the ratio of the displacement corresponding to the minimum value and the second minimum value among the corresponding multiple displacements' Euclidean distances. The displacement's reference feature point and displacement's to-be-corrected feature point corresponding to the displacement ratio less than the displacement's predetermined threshold are used as displacement matching point pairs.

Images