CN111256607A

CN111256607A - Deformation measurement method based on three-channel mark points

Info

Publication number: CN111256607A
Application number: CN202010103370.3A
Authority: CN
Inventors: 赵健; 马雪怡; 马超臣; 任晴
Original assignee: Beijing Forestry University
Current assignee: Beijing Forestry University
Priority date: 2020-02-19
Filing date: 2020-02-19
Publication date: 2020-06-09
Anticipated expiration: 2040-02-19
Also published as: CN111256607B

Abstract

The invention discloses a deformation measuring method based on three-channel mark points, which comprises the following steps: step 1, setting three-channel mark points on the surface of an object to be detected, and collecting images before and after the surface of the object to be detected is deformed; step 2, carrying out position detection and center positioning on the three-channel mark points in the image; and 3, matching the same mark point in the image before and after deformation to obtain the full-field displacement of the object. The deformation measuring method based on the three-channel mark points, provided by the invention, has the advantages that the number of the needed mark points is small, irreversible fouling on an object to be measured is avoided, and the method is suitable for measuring transparent materials, collected cultural relics and the like; the unique surrounding information and color information of the surrounding mark points of each mark point are utilized to form a feature vector, so that the matching accuracy is high, the measurement accuracy is high, and the calculation speed is high.

Description

Deformation measurement method based on three-channel mark points

Technical Field

The invention belongs to the technical field of object deformation measurement, and particularly relates to a deformation measurement method based on three-channel mark points.

Background

The object deformation is a physical phenomenon commonly existing in nature, engineering technology, cultural relic protection and daily life, and in some technical fields, such as engineering technology, cultural relic protection and the like, the object deformation needs to be accurately measured. At present, digital image correlation methods are mostly adopted for measurement, and digital image correlation technology (DIC) is originally independently proposed by Peters at the university of south Carolina of the United states and Yamaguchi in Japan in the early 80 s of the last century respectively and is an image-based non-contact optical method for measuring full-field morphology, displacement and deformation.

The DIC technology first uses digital imaging equipment (optical imaging, electronic imaging, scanning probe imaging equipment, etc.) to acquire digital images of the measured object in different states, and then uses correlation-based matching and numerical differentiation methods to perform image analysis to quantitatively calculate the full-field displacement and the full-field strain of the measured object. Specifically, a large number of speckles need to be sprayed on the surface of the object to be measured to form a speckle pattern, then the speckles are irradiated by a light source to form the speckle pattern, the speckle images before and after deformation are collected by an image capture device, and correlation calculation processing is performed to obtain an accurate deformation field of the object. The method shows great superiority in a plurality of application fields such as material mechanics, fracture mechanics, biomechanics, field real-time measurement, micro-scale deformation field measurement, electronic packaging, dynamic displacement test and the like.

However, the adoption of the digital correlation technique to measure the deformation of the object requires that a large amount of speckles are sprayed on the surface of the object, so that the object made of special materials (such as glass, transparent rubber and the like) is easy to be stained, and irreparable damage is caused to cultural relics and the like in a collection; the DIC technology for calculating deformation relates to complex algorithms such as shape functions, iterative functions, subarea selection and the like, and is low in deformation speed and efficiency, so that the digital image correlation method is not applicable any more; in addition, the matching accuracy of the same mark point in the image before and after deformation is low due to the common black and white mark points, and errors are easy to generate.

Therefore, there is a need for a three-channel-based deformation measurement method, which does not require speckle spraying on the surface of the object to be measured, and has high measurement speed and high accuracy.

Disclosure of Invention

In order to overcome the problems, the inventor of the present invention has made intensive studies to design a deformation measurement method based on three-channel marker points, which is characterized in that each marker point is accurately positioned at a sub-pixel level, a feature vector is formed by using surrounding information and color information of color marker points around one marker point, and the feature vector is used as a characteristic that the marker point is different from other marker points, so as to realize matching of the same marker point in an image before deformation and an image after deformation, thereby estimating the displacement of the whole field, wherein the required marker points are few, the matching accuracy is high, and the calculation time is short, thereby completing the present invention.

Specifically, the invention aims to provide a deformation measuring method based on three-channel mark points, wherein the method comprises the following steps:

step 1, setting three-channel mark points on the surface of an object to be detected, and collecting images before and after the surface of the object to be detected is deformed;

step 2, carrying out position detection and center positioning on the three-channel mark points in the image;

and 3, matching the same mark point in the image before and after deformation to obtain the full-field displacement of the object.

Wherein, step 2 comprises the following substeps:

step 2-1, carrying out preliminary position detection on the mark points;

step 2-2, performing integral pixel level center positioning on the mark points;

and 2-3, performing center positioning at a sub-pixel level on the mark points in the step 2-2.

Wherein, step 3 comprises the following substeps:

step 3-1, obtaining the characteristic information of each mark point in the image before the deformation of the object to be detected;

step 3-2, obtaining the characteristic information of each mark point in the deformed image of the object to be detected;

and 3-3, comparing the characteristic information of each mark point in the image before and after the deformation of the object to be detected to obtain the same matched mark point.

In step 3-1, the feature information is a feature vector, and includes position and color information.

Wherein, the step 3-1 comprises the following steps:

step 3-1-1, selecting a mark point in the image before deformation, and then selecting n mark points with the closest distance around the mark point as surrounding mark points;

3-1-2, dividing the surrounding area of the selected mark point, and numbering the divided areas in sequence;

3-1-3, calculating the information of surrounding mark points in each area according to the area number sequence to obtain the characteristic information of the selected mark points;

and 3-1-4, repeating the steps 3-1-1 to 3-1-3 to obtain the characteristic information of all the mark points in the deformed image.

In step 3-1-3, the information of the surrounding mark points in each region includes the number of the surrounding mark points in each region, the distance from each surrounding mark point to the mark point at the center of the circle, and the color of the surrounding mark points.

In step 3-3, the characteristic information of each mark point in the image of the object to be detected before deformation and after deformation is compared, and a pair of mark points with the maximum similarity is selected as the same mark point before deformation and after deformation.

Wherein the similarity comprises a Hamming distance, an Euclidean distance and a cosine similarity.

Wherein, after the step 3-3, the following steps are also included:

3-4, checking the matching condition of the mark points in the image before and after deformation to eliminate abnormal values;

and 3-5, obtaining the full-field displacement of the deformed object according to the correctly matched mark points.

The invention has the advantages that:

(1) the deformation measuring method based on the three-channel mark points is simple to operate, needs a small number of mark points, does not cause irreversible fouling on an object to be measured, and is suitable for measuring transparent materials, collected cultural relics and the like;

(2) according to the deformation measuring method based on the three-channel mark points, the surrounding information and the color information of the unique surrounding mark points of each mark point are utilized to form the characteristic vector, the matching accuracy is high, and compared with the gray mark points, the full-field displacement measuring precision is high;

(3) the deformation measurement method based on the three-channel mark points does not relate to complex algorithms such as shape functions, iteration functions, subarea selection and the like, and the calculation efficiency is improved by 75% compared with that of a digital image correlation method;

(4) the deformation measuring method based on the three-channel mark points combines the feature vector based on topological arrangement and the second harmonic spline interpolation method, makes up the defect of the displacement measuring method with few mark points, improves the measuring precision and has wide application range.

Drawings

FIG. 1 is a schematic diagram of image acquisition of an object to be measured according to a preferred embodiment of the present invention;

FIG. 2 illustrates a landmark region division diagram in accordance with a preferred embodiment of the present invention;

fig. 3 shows an overall image of an object to be measured before deformation according to embodiment 1 of the present invention;

FIG. 4 is a diagram showing a distribution of landmark points in a part of an image before deformation of an object to be measured;

FIG. 5 is a diagram showing a distribution of landmark points in a part of an image after deformation of an object to be measured;

FIG. 6 shows a vector field diagram of deformation of an object under test according to embodiment 1 of the present invention;

FIG. 7 is a diagram showing u-field deformation in example 1 of the present invention;

FIG. 8 is a view showing a v-field deformation in embodiment 1 of the present invention;

FIG. 9 is a graph showing a comparison of the accuracy of deformation detection by different methods in Experimental example 1 of the present invention;

FIG. 10 is a graph showing a comparison of the calculated time in Experimental example 2 of the present invention.

Detailed Description

The present invention will be described in further detail below with reference to preferred embodiments and examples. The features and advantages of the present invention will become more apparent from the description.

The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

The inventor researches and discovers that in the deformation measurement of an object, if the displacement is continuous, when one mark point moves between frames of two images, the surrounding mark points surround the mark point although different displacement changes exist, one mark point is surrounded by a plurality of surrounding mark points, and the distribution condition of the surrounding mark points is the characteristic that the mark point is different from other mark points. Based on the above, the feature vectors based on random neighborhood topological arrangement are used for matching the same mark point in the reference image and the deformed image, so that the full-field displacement is obtained.

The invention provides a deformation measuring method based on three-channel mark points, which comprises the following steps:

step 1, three-channel mark points are arranged on the surface of an object to be detected, and images before and after the surface of the object to be detected deforms are collected.

In the invention, in order to avoid the surface of the object to be detected from being stained, a sticking mode is preferably adopted to replace a spraying mode to set the mark points on the surface of the object to be detected. The three channels mean that the set mark points comprise three colors of red, blue and green, so that the specific code of each mark point comprises position information and color information, and the matching accuracy of the same mark point in the image before and after deformation is improved.

Wherein, each color mark point is not required to have mark points with the same color as much as possible, and the mark points are distributed in a pseudo-random mode on the basis of the pseudo-random distribution, namely, pseudo-random number distribution, and the pseudo-random number is a random number sequence which is calculated by a deterministic algorithm and is uniformly distributed from [0,1], is not truly random, but has the similar statistical characteristics of the random numbers, such as uniformity, independence and the like.

In a preferred embodiment, the marker points are approximately rigid circular marker points, i.e. circular after deformation, so as to avoid affecting the displacement calculation before and after deformation.

The number of the three-channel mark points arranged in the invention does not need to be too many, an image acquisition device is generally adopted to photograph the surface of an object to be detected, and the obtained image is subjected to pixel division, preferably, the set density of the mark points is 1/6561 square pixels to 1/4096 square pixels, namely, one mark point is arranged in each (64 pixels multiplied by 64 pixels) to 81 pixels multiplied by 81 pixels).

The inventor finds that by adopting the method, high estimation effect can be achieved by sacrificing local precision with a small number of mark points.

In a preferred embodiment, the image of the surface of the object to be measured is acquired by one or two synchronous image acquisition devices, such as a single digital camera or two synchronous digital cameras.

When the camera is placed, the optical axis of the camera is perpendicular to the surface of the object to be measured, as shown in fig. 1.

And 2, carrying out position detection and center positioning on the three-channel mark points in the image.

Wherein the step 2 comprises the following substeps:

and 2-1, carrying out primary position detection on the mark points.

In the invention, bright (dark) pixels are divided under dark (bright) background pixels in the image before and after the deformation of the object to be detected, namely the positions of the mark points, and the connected regions of the bright pixels corresponding to the mark points are calculated to preliminarily determine the positions of the set mark points.

Preferably, the preliminary location of the landmark points is obtained by means commonly used in the art, such as matlab software.

And 2-2, carrying out integral pixel level center positioning on the mark points.

Specifically, the centroid calculation is performed on each mark point to obtain an estimated coordinate value of the centroid of the circular mark point at the integer pixel level, and the coordinate of the estimated coordinate value is an integer.

The pixel is a basic unit of an imaging plane and is also a minimum unit, and is generally referred to as a physical resolution of an image, and the integer pixel precision refers to an original pixel precision of an original image.

In the present invention, the calculation of the centroid coordinates can be performed by software commonly used in the art, such as matlab software.

In the present invention, in order to improve the displacement calculation accuracy, it is preferable to convert the centroid coordinates of the entire pixel level of the marker point to the subpixel level.

The sub-pixel precision refers to the condition of subdivision between two adjacent pixels, namely, a sub-pixel point between two whole pixel points can be called a half sub-pixel point, and a quarter sub-pixel point can be called between two 1/2 pixel points or between one whole pixel point and one 1/2 sub-pixel point.

The invention preferably adopts a radial symmetry method to convert coordinates at the integer pixel level into coordinates at the sub-pixel level.

The principle of the radial symmetry method is as follows: neglecting asymmetric optical aberration, the intensity of the imaging single mark light spot acquired by the noise optical equipment is radially and symmetrically distributed relative to the light spot center, and for an image with ideal color intensity in radial and symmetrical distribution, any gradient orientation line of the image color intensity will intersect with a point, which is the center point.

In a preferred embodiment, a sub-region where each marker point is located is selected by a frame with the whole pixel coordinate of the center of the marker point as the center, and then the marker point in each sub-region is centered at a sub-pixel level by using the characteristic that the color intensity of the circular marker point presents radial symmetry.

The sub-pixel level coordinate of the positioning is located in a sub-area coordinate system, and the sub-area coordinate system is a coordinate system which is created by taking the center of the whole pixel obtained by centroid calculation as an origin.

In the invention, the center coordinates of each sub-area are accurate to the sub-pixel level by adopting the method, so that the accuracy can be improved and the subsequent displacement calculation is facilitated.

Through the above steps, the center position coordinates of each selected marker point can be obtained.

If the full-field displacement of the object to be measured is to be obtained, the positions of the three-channel mark points in the images before and after deformation need to be calculated, and then the coordinate difference of the positions is calculated, namely the displacement difference of the same mark point in the images before and after deformation is obtained, so that the deformation condition of the object to be measured can be obtained naturally.

Specifically, step 3 comprises the following substeps:

and 3-1, obtaining the characteristic information of each mark point in the image before the deformation of the object to be detected.

According to a preferred embodiment of the invention, the feature information is a feature vector comprising position and color information.

The feature vector is defined according to the topological arrangement of the random neighborhood, and represents the situation that the periphery of a certain mark point is surrounded by other mark points, including the position information and the color information of the mark point.

Specifically, step 3-1 includes the following substeps:

and 3-1-1, selecting a mark point in the image before deformation, and then selecting n mark points with the shortest distance around the mark point as surrounding mark points.

According to a preferred embodiment of the present invention, the number n of surrounding landmark points ranges from: 2/3 Total number of marker points < n < 3/4 Total number of marker points.

Wherein the larger n is, the higher the accuracy of the subsequent calculation of the deformation of the target object is, and the higher the complexity of the deformation calculation is, the inventor finds that when the total number of 2/3 marked points is less than n < 3/4, the higher accuracy of the deformation calculation and the higher measurement efficiency can be maintained.

In a further preferred embodiment, the number n of surrounding marker points may be 8 to 12.

Preferably, the surrounding mark points each have respective color information, such as red, green, and blue.

The red, green and blue three-channel mark points are arranged in the invention, so that the surrounding condition of the surrounding mark points to the selected mark points and the respective different colors form a feature vector.

And 3-1-2, dividing the surrounding area of the selected mark point, and numbering the divided areas in sequence.

According to a preferred embodiment of the invention, a circle is drawn with the selected index point as the center of the circle, the distance of the point of the n surrounding index points which is farthest from the center of the circle as the radius,

a concentric inner circle is made in the circle, and the radius of the inner circle is one half of that of the outer circle.

Wherein the radius of the outer circle is set to a distance between the farthest ones of the n surrounding mark points from the center of the circle so as to include the n surrounding mark points in the divided region.

In the present invention, through step 2, the center position coordinates, i.e., centroid coordinates, of each marker point in the deformed image have been obtained; and then, by region division, the coordinate range of each region can be obtained, and further, the specific position information of the mark point in each region can be obtained.

And 3-1-3, calculating the information of the surrounding mark points in each area according to the area number sequence, and obtaining the characteristic information of the selected mark points.

In the present invention, as shown in fig. 2, each concentric circle is divided into four quadrants using basis vectors, whereby the area around the selected index point can be divided into eight areas, i.e., eight quadrants, numbered as I, II, III, IV, V, VI, VII, and VIII in this order.

According to a preferred embodiment of the present invention, the calculated information of the surrounding mark points in each region includes the number of the surrounding mark points in each region, the distance of each surrounding mark point from the mark point at the center of the circle, and the color of the surrounding mark points.

And each surrounding mark point is sorted from near to far according to the distance from the mark point at the circle center, the farthest distance is used as the radius of the excircle of the concentric circle when the region is divided, and the half of the farthest distance is used as the radius of the inner circle.

The inventor researches and discovers that the relative positions of n surrounding mark points can be converted into eight different regions by calculating the number, distance and color information of the surrounding mark points in each region to enable three kinds of information to be three kinds of feature vectors and encoding the three kinds of feature vectors.

According to the invention, by adopting the characteristic vector as the characteristic information of the selected mark point, not only new change caused by object deformation is considered, but also the position and color information which is beneficial to matching before and after deformation can be extracted, and the matching accuracy and the measurement precision are improved.

And sequentially taking the rest three-channel mark points of the image before the deformation of the object to be detected as the circle center, and obtaining the characteristic information of all the mark points of the image before the deformation according to the steps.

And 3-2, obtaining the characteristic information of each mark point in the deformed image of the object to be detected.

Specifically, the steps 3-1-1 to 3-1-4 are repeated on the deformed image of the object to be detected, and the characteristic information of all the mark points is obtained.

In general, points around a certain marking point tend not to change much after deformation.

According to the method, the positions of the marker points in the two images are associated with the feature vectors containing the color information according to the information of the marker points surrounding a certain marker point in the images before and after deformation, so that the marker points in the images before and after deformation are matched one by one.

According to a preferred embodiment of the present invention, the feature information of each marker point in the image of the object to be measured before and after deformation is compared, and a pair of marker points with the largest similarity is selected as the same marker point before and after deformation.

In a further preferred embodiment, the similarity includes a hamming distance, a euclidean distance, and a cosine similarity, preferably one or more of the hamming distance, the euclidean distance, and the cosine similarity, and more preferably the hamming distance.

The hamming distance represents the number of corresponding bits of two (same length) words different from each other, and d (x, y) represents the hamming distance between the two words x, y. And carrying out exclusive OR operation on the two character strings, and counting the number of 1, wherein the number is the Hamming distance. The greater the Hamming distance is, the higher the similarity of the two compared marker points is proved to be, and the two marker points can be considered as before and after the deformation of the same marker point.

Compared with the digital image correlation method in the prior art, the method of the invention has no problem of calculation failure due to low similarity.

And 3-4, checking the matching condition of the mark points in the image before and after deformation to eliminate abnormal values.

The inventor finds that in the research process, in the process of comparing the similarity of the mark points before and after deformation, the condition that the similarity of the feature information of a certain mark point before deformation is the same as the similarity of the feature information of a plurality of mark points in the image after deformation sometimes occurs, namely the problem of 'one-to-many' exists, the matching condition needs to be further detected, abnormal values are searched and eliminated, and the accuracy of subsequent detection is ensured.

According to a preferred embodiment of the present invention, the method for detecting median is used to search for abnormal values, and the specific steps are as follows:

step a, selecting a mark point to be detected and a plurality of adjacent mark points around the mark point, preferably selecting 3-5 mark points around the mark point, such as 4 mark points.

For example, the marker point to be detected is selected as a1, and 4 adjacent marker points are selected around the marker point, which are named as a2, A3, a4 and a5, respectively.

And b, calculating the displacement of the mark point to be detected and the selected adjacent mark points around the mark point to obtain the displacement median of the mark points.

Wherein, the size of the deformation displacement refers to the displacement of the mass center of the mark point before and after deformation.

And c, dividing the displacement of each mark point in the step b by a median, comparing the result with a threshold value, wherein the mark points which are larger than the threshold value are abnormal values and need to be removed.

According to a preferred embodiment of the present invention, the threshold value ranges from 19 to 21.

Specifically, the marking point a1 to be detected and 4 adjacent marking points a2, A3, a4 and a5 around the marking point are taken as examples, and the median of deformation displacement of the five points is B.

And | A1-Ai | is the distance from A1 to Ai, and the value of i is 2,3, 4 and 5. When the absolute value A1-Ai/B is larger than 19-21, the mark point matched with the value A1 in the deformed image is an error mark point and is removed as an abnormal value; and after the mark points are removed, selecting other mark points with the same similarity to continue to carry out median detection until the mark points which are correctly matched are obtained.

And 2, calculating the horizontal displacement and the vertical displacement of the mark point according to the coordinates of the same mark point in the image before and after deformation, wherein the coordinates of the center position of the mark point are accurate to the sub-pixel level in the step 2, so that the calculated horizontal displacement and the calculated vertical displacement of the mark point are both in the sub-pixel level.

In the present invention, in order to obtain the full-field displacement of the object, in addition to the mark points set in the region, the displacements of other non-mark point portions in the region need to be calculated, and therefore, it is preferable to obtain the full-field displacement, such as an interpolation function, by using a method (estimating the full-field displacement by the displacement of several points) commonly used in the prior art.

In a preferred embodiment of the present invention, a dual harmonic spline interpolation method is used to obtain the full-field displacement of the object to be measured.

The bi-harmonic spline interpolation method is to adopt a two-dimensional bi-harmonic spline interpolation method to carry out minimum curvature interpolation, and obtain the smoothest displacement field through discrete point displacement data. And performing minimum curvature interpolation on the scattered points by utilizing the Green function of the double harmonic operator. The interpolated surface is a linear combination of green's functions centered on each scattering point, satisfying a bi-harmonic equation, and therefore has minimal curvature. And solving a double harmonic function passing through the scattering sampling point by using a double harmonic spline interpolation method, and realizing by using an equation set.

In the invention, a bi-harmonic spline interpolation method is selected, and the displacements of all points in a plane are fitted by using a curve which is as smooth as possible through the displacement of the mark points, so that the displacement component in the X, Y direction in the whole plane, namely the whole deformation condition of the object to be detected, can be more accurately and effectively obtained.

The deformation measuring method based on three-channel mark points is an effective method for carrying out mark point matching and displacement calculation based on the characteristic vectors, the characteristic vectors are formed by utilizing the surrounding information of the surrounding mark points and the color information of the surrounding mark points which are respectively unique to each mark point, and the matching of the mark points is carried out based on the similarity of the characteristic vectors of each mark point. The formation of the characteristic vector uniquely codes each mark point, and the unique position information and color information of each mark point are reserved, compared with the traditional digital image correlation method, the required number of the mark points is less, and the defect of a displacement measurement method with few mark points is made up by combining the characteristic vector based on topological arrangement and a second harmonic spline interpolation method; compared with the gray scale mark points, the color information of the three-channel mark points adopted in the invention can increase the probability of correct matching and improve the matching quality.

In addition, the deformation measuring method can meet the requirement of effective calculation under the condition that a large number of mark points cannot be sprayed on an object to be measured, does not relate to complex algorithms such as shape functions, iterative functions, subarea selection and the like, and has higher calculation efficiency.

Examples

The present invention is further described below by way of specific examples, which are merely exemplary and do not limit the scope of the present invention in any way.

Example 1

Carrying out deformation measurement on the three-point bending equal-strength beam according to the following steps:

(1) every 22500mm on one side surface of the test piece²(150×150mm²) Pasting 1 red/green/blue three-channel mark points, and setting the density to be 1/4096 square pixels;

an industrial camera MV-EM130M and an industrial lens (a 16mm computer lens) are adopted to shoot a deformed image of the surface of the test piece stuck with the three-channel mark point, so that the optical axis of the industrial camera is perpendicular to the surface of the test piece, and the image before deformation is shown in FIG. 3;

and then, an electronic creep testing machine is adopted to load 1500N external force on the test piece to deform the test piece, and the same industrial camera is adopted to shoot the deformed image.

(2) Three-channel mark points in the image before deformation are identified by matlab software, the coordinates of the centroid at the whole pixel level of the mark points are obtained as (32,32), and the coordinate system of the sub-area is a coordinate system established by taking the center of the whole pixel obtained by centroid calculation as the center, so that the whole pixel coordinates of the mark points in all the selected areas (sub-areas) of the image before deformation are all (32, 32);

and then converting the centroid coordinate of the integer pixel level into a coordinate of the sub-pixel level by adopting a radial symmetry method.

Taking the 9 landmark points P1-P9 in partial area in fig. 4 as an example, the sub-pixel level centroid coordinates of the 9 landmark points in the pre-deformation image are shown in table 1.

TABLE 1

Number of mark points	Centroid coordinate (x)	Centroid coordinate (y)
			P1	32.1998	31.5265
P2	31.9994	31.6859
			P3	32.0004	33.9561
P4	31.8034	32.0036
			P5	31.6521	33.2625
P6	33.4369	32.6568
			P7	32.5614	32.9658
P8	30.6695	33.0021
			P9	33.0032	30.9586

(3) Selecting 9 surrounding mark points with the closest distance for each mark point, respectively taking the central position of each mark point as a circle center, taking the distance between the point of the 9 surrounding mark points with the farthest distance from the circle center as a radius to make a circle, and making a concentric inner circle inside the circle, wherein the radius of the inner circle is one half of that of the outer circle;

each concentric circle is divided into four quadrants by using the base vector, so that the area around the selected marking point can be divided into eight areas, namely eight quadrants, which are numbered as I, II, III, IV, V, VI, VII and VIII in sequence;

and calculating the number of surrounding mark points in each region, the distance from each surrounding mark point to the center of the circle and the number of each surrounding mark point, and converting the number into three feature vectors.

Taking 9 mark points P1-P9 in fig. 4 as an example, the distances between P1-P9 and the center of the circle gradually increase from P1 to P9, 1,2 and 3 respectively represent the three colors of red, green and blue of the three-channel mark point, and 0 represents no mark point.

The feature vectors of the center points with respect to the number are [1,1,0,0,2,2,1,2], and the feature vectors with respect to the colors are shown in table 2.

TABLE 2

Region(s)	Feature vector (image before deformation)
		I	[1,0,0]
II	[3,0,0]
		III	[0,0,0]
IV	[0,0,0]
		V	[2,3,0]
VI	[1,2,0]
		VII	[1,0,0]
VIII	[2,3,0]

The determination process of the P1-P9 marker points and all the marker points set in the image with respect to the number and the feature vectors of the colors is the same as the above-described process.

(4) And (4) repeating the steps (2) and (3), and obtaining the sub-pixel level centroid coordinates and the characteristic information of all the mark points in the deformed image.

Taking the 9 landmark points M1-M9 in partial area in fig. 5 as an example, the sub-pixel level centroid coordinates of the 9 landmark points in the deformed image are shown in table 3.

TABLE 3

Taking 9 mark points M1-M9 in fig. 5 as an example, the distances between M1-M9 and the center of the circle gradually increase from M1 to M9, 1,2, and 3 represent three colors of red, green, and blue of a three-channel mark point, and 0 represents no mark point.

The feature vectors of the circle center index points with respect to the number are [1,1,0,0,2,2,1,2], and the feature vectors with respect to the colors are shown in table 4.

TABLE 4

Region(s)	Feature vector (image after deformation)
		I	[1,0,0]
II	[3,0,0]
		III	[0,0,0]
IV	[0,0,0]
		V	[2,3,0]
VI	[1,2,0]
		VII	[3,0,0]
VIII	[2,3,0]

(5) Comparing the characteristic information of each mark point in the image before and after deformation, calculating the Hamming distance, obtaining the point with the maximum similarity with the mark point in the image before deformation from the mark points in the image after deformation, and obtaining the matched paired mark points.

Taking P1-P9 in the pre-deformed image and M1-M9 in the post-deformed image as examples, the matching results are shown in table 5.

TABLE 5

Pre-deformation image	Deformed image
		P1	M1
P2	M2
		P3	M3
P4	M4
		P5	M5
P6	M6
		P7	M7
P8	M8
		P9	M9

(6) And (3) carrying out abnormal value detection on the matching result of the mark points of the image before deformation and the image after deformation:

for example, the magnitudes of displacements before and after deformation of adjacent marker points P2, P3, P4 and P5 around the marker point P1 are calculated to be 0.045354, 0.046325, 0.045891, 0.045330 and 0.047954, the median of the displacements of the five marker points is calculated to be 0.045891, the displacement values of the five marker points P1 to P5 are divided by the median to obtain ratios 0.9883, 1.0095, 1.0000, 0.9878 and 1.0450, and the detected marker point can be judged to be not an abnormal value when the ratios are smaller than the threshold value 20.

And detecting all the mark points according to the steps, and eliminating the mark points with the ratio higher than the threshold value.

(7) The displacements before and after the deformation of the same marking point were calculated, and the displacements are shown in Table 6, taking P1-P9 as an example.

TABLE 6

Number of mark points	Displacement of
		P1	0.045354
P2	0.046325
		P3	0.045891
P4	0.045330
		P5	0.047954
P6	0.044931
		P7	0.045235
P8	0.045035
		P9	0.034340

The full-field displacement is estimated by adopting a double-harmonic spline interpolation method, and the vector field result of the deformation condition of the test piece is shown in fig. 6 (wherein the arrow direction is the displacement direction, the arrow length represents the relative magnitude of the displacement, the abscissa represents the position in the horizontal direction, and the ordinate represents the position in the vertical direction).

The u-field (horizontal direction) deformation is shown in fig. 7, and the v-field (vertical direction) deformation is shown in fig. 8, where the abscissa represents the length of the calculation region in pixels, and different colors represent different degrees of displacement. The full-field displacement change of the test piece can be visually seen from fig. 7 and 8.

Example 2

The method used in this example is similar to that of example 1, except that the test piece was loaded with an external force of 2000N.

Example 3

The method used in this example is similar to that of example 1, except that the external force applied to the test piece was 2500N.

Comparative example

Comparative example 1

The comparative example used a method similar to that of example 1 except that the set marking points were black marking points, i.e., the characteristic information of each marking point did not include color information.

Comparative example 2

The comparative example used a method similar to that of example 2 except that the set marking points were black marking points, i.e., the characteristic information of each marking point did not include color information.

Comparative example 3

The comparative example used a method similar to that of example 3 except that the set marking points were black marking points, i.e., the characteristic information of each marking point did not include color information.

Comparative example 4

The test piece and Image acquisition apparatus used in this comparative example were the same as in example 1 except that the full field displacement of the test piece was measured by Digital Image Correlation (DIC) according to the procedures described in the documents "J.Blaber, B.Adair, A.Antoniou.Ncor: Open-Source 2D Digital Image Correlation matrix Software [ J ]. Experimental memories, 2015, Vol.55(6), pp.1105-1122".

Comparative example 5

The test piece and the image acquisition device used in the comparative example are the same as those in example 2, except that a Digital Image Correlation (DIC) method is used to measure the full-field displacement of the test piece.

Comparative example 6

The test piece and the image acquisition device used in the comparative example are the same as those in example 3, except that the Digital Image Correlation (DIC) method was used to measure the full field displacement of the test piece.

Examples of the experiments

Experimental example 1

The results of comparing the full-field displacement results estimated by the methods described in examples 1 to 3 and comparative examples 1 to 6 are shown in fig. 9.

As can be seen from fig. 9, under different loading forces, the full-field displacement measurement methods used in examples 1 to 3 and comparative examples 1 to 3 achieve a higher estimation effect with a small amount of point sacrifice for local accuracy, and compared with the DIC measurement methods described in comparative examples 4 to 6, the estimation error is less than 0.01 mm.

Compared with the black mark point estimation methods used in comparative examples 1 to 3, the three-channel color mark point estimation methods used in examples 1 to 3 of the present invention have higher measurement accuracy, the estimation errors of the methods and the DIC methods described in examples 1 to 3 are ± 0.0055mm, and the estimation errors of the methods and the DIC methods described in comparative examples 1 to 3 are ± 0.0075mm, which indicates that the deformation measurement method based on three-channel mark points described in the examples of the present invention has higher measurement accuracy.

Experimental example 2

The full field displacement calculation times for example 1 and comparative example 4 were compared and the results are shown in fig. 10.

As can be seen from fig. 10, the calculation time decreases as the calculation step increases in embodiment 1; in comparative example 4, the calculation time gradually increased as the size of the sub-region increased. Compared with comparative example 4 adopting DIC method, the calculation speed of the deformation measurement method based on three-channel marking point is increased by about 75%, and the calculation efficiency is obviously improved.

In conclusion, the deformation measuring method based on the three-channel mark points can realize the measurement of the full-field displacement quickly, efficiently and accurately under the condition that the object to be measured is not stained.

The invention has been described in detail with reference to specific embodiments and illustrative examples, but the description is not intended to be construed in a limiting sense. Those skilled in the art will appreciate that various equivalent substitutions, modifications or improvements may be made to the technical solution of the present invention and its embodiments without departing from the spirit and scope of the present invention, which fall within the scope of the present invention.

Claims

1. A deformation measurement method based on three-channel mark points is characterized by comprising the following steps:

2. The method according to claim 1, characterized in that step 2 comprises the following sub-steps:

step 2-1, carrying out preliminary position detection on the mark points;

3. The method according to claim 1, characterized in that step 3 comprises the following sub-steps:

4. The method according to claim 3, wherein in step 3-1, the feature information is a feature vector comprising position and color information.

5. The method of claim 3, wherein step 3-1 comprises the steps of:

6. The method according to claim 5, wherein in step 3-1-3, the information of the surrounding mark points in each area comprises the number of the surrounding mark points in each area, the distance of each surrounding mark point from the mark point at the center of the circle, and the color of the surrounding mark points.

7. The method according to claim 3, wherein in step 3-3, the feature information of each mark point in the image of the object to be detected before and after deformation is compared, and the pair of mark points with the largest similarity is selected as the same mark point before and after deformation.

8. The method of claim 7, wherein the similarity comprises a hamming distance, a euclidean distance, and a cosine similarity.

9. The method of claim 3, further comprising, after step 3-3, the steps of: