CN104732476A - Microstructure low-overlapping-degree three-dimensional splicing method based on optical nondestructive testing - Google Patents

Abstract

A low-overlap 3D mosaic method for microstructures based on optical non-destructive testing. It includes limiting the structural feature extraction area to the overlapping area in the measurement process based on the motion parameters of the test mechanism displacement platform in the experiment; within the area, the feature extraction is performed through the SURF algorithm; in the feature point matching stage, according to the displacement platform of the measurement system Uncertainty further proposes a method of narrowing the search range of matching point pairs to improve the reliability of feature point matching and obtains feature matching points according to the nearest neighbor method of Euclidean distance; based on the local continuity of the overlapping area, the correction matrix is calculated by the STLS algorithm And get the final stitching. The present invention is not only applicable to structures with rich features, but also to array structures with high similarity and indistinct features, which can effectively eliminate mismatching and improve splicing accuracy. Considerable additional test time comes, and large-scale measurements are successfully achieved.

Description

A kind of microstructure based on optical non-destructive detection low degree of overlapping three-dimensional splicing method

Technical field

The invention belongs to Microstructure Optics technical field of nondestructive testing, particularly relate to a kind of microstructure based on optical non-destructive detection joining method on a large scale.

Background technology

Optical non-destructive detection technology is the important content of ultraprecise detection technique, is used widely in leading-edge fields such as Aero-Space, biotechnology, communication, microelectronics.In recent years, although a large amount of new technologies and new product are constantly being pushed to application, along with the development of precision processing technology, the resolving power of optical non-destructive detection technology and measurement range, the contradiction measured between efficiency become more outstanding.Therefore, how to realize just becoming with high resolution on a large scale simultaneously microstructure measure in a major issue.

Because of the high measurement efficiency of optical non-destructive detection and parameter be easy to set, so under normal conditions, horizontal large-range measuring can realize by joining method.The overlapping region of main flow business equipment General Requirements adjacent structure now remains on 10%-25%, and lower degree of overlapping can not ensure good splicing effect, but the increase of additional measurement time that high degree of overlapping is brought is considerable when actual measurement.Also have scientific research personnel to propose new scan mode for this problem in the world, but these methods all need to transform the optical system of scanning mechanism and scanning moving mechanism, versatility is poor, and cannot avoid the Bonding Problem in two-dimentional large-range measuring.Therefore, the joining method of high robust, low degree of overlapping still has higher researching value to microstructure level three-dimensional precise measurement on a large scale.

Summary of the invention

In order to solve the problem, the object of the invention is to propose a kind of microstructure based on optical non-destructive detection joining method on a large scale, successfully to realize large-range measuring, and reduce the increase of the additional measurement time brought because of overlap in measuring process by reducing degree of overlapping.

In order to achieve the above object, the microstructure based on optical non-destructive detection provided by the invention on a large scale joining method comprise the following step carried out in order:

Step 1: the kinematic parameter of test macro displacement platform experimentally, setting characteristic matching region is the lap of two adjacent structures to be spliced;

Step 2: extract architectural feature in the characteristic matching region adopting SURF algorithm to set in described step 1;

Step 3: according to test macro displacement platform uncertainty, setting characteristic matching point search scope, and obtain characteristic matching point according to Euclidean distance nearest-neighbor method;

Step 4: with the local continuity of overlapping region for foundation, utilize STLS algorithm to calculate correction matrix, the splicing construction changing of the relative positions brought with environmental perturbation in correction measurement process, is finally spliced result thus.

In step 1, described experimentally in the kinematic parameter of test macro displacement platform, to be the concrete grammar of the lap of two adjacent structures to be spliced be in setting characteristic matching region:

According to the level of the article carrying platform that can accurately obtain in optical precision nondestructive measurement system and the relative displacement between vertical movement position and adjacent structure, delimit characteristic matching region, mate with eliminating error and improve computing velocity; Described relative displacement relation is such as formula shown in (1):

A ₁(x,y)+L＝A ₂(x,y) (1)

Wherein, (x, y) is two adjacent Structural superposition region to be spliced point midway coordinates, A ₁, A ₂be two adjacent structures to be spliced; L is x or y direction displacement;

The relation of two adjacent neighbours Structural superposition degree to be spliced and relative displacement is such as formula shown in (2):

(P-L/p)/P＝a (2)

Wherein, P is the total number of structure x direction pixel to be spliced, and p is pixel equivalent dimension, and a is two adjacent Structural superposition degree to be spliced; Delimit structure A to be spliced ₁middle matching area is that x direction P (1-a) row arrange to P, delimits structure A to be spliced ₂middle matching area is that x direction the 1st arranges P (1-a) row, when applying above-mentioned formula acquired results not for integer, for ensureing that lap is included, gets the round values solution making setting regions larger;

The measuring process in y direction is identical with above-mentioned measuring process.

In step 3, described according to test macro displacement platform uncertainty, setting characteristic matching point search scope, and according to the concrete grammar that Euclidean distance nearest-neighbor method obtains characteristic matching point be;

Described matching double points range set is such as formula shown in (3) and formula (4):

x ₁+L _x-r _x≤x ₂≤x ₁+L _x+r _x(3)

y ₁+L _y-r _y≤y ₂≤y ₁+L _y+r _y(4)

Wherein, (x ₁, y ₁) and (x ₂, y ₂) be one group and belong to two adjacent structure A to be spliced respectively ₁, A ₂in matching double points, (L _x, L _y) be respectively adjacent structure A to be spliced ₁, A ₂the relative displacement of x, y direction, (r _x, r _y) respectively defining the hunting zone of x, y direction character match point, its value is determined by measuring equipment uncertainty.

In step 4, described with the local continuity of overlapping region for foundation, utilize STLS algorithm to calculate correction matrix, the splicing construction changing of the relative positions brought with environmental perturbation in correction measurement process, the concrete grammar finally being spliced result is thus:

In this step, correct-by-construction does not relate to dimensional variation and malformation, provides the matrix restraint condition in described STLS algorithm, and the correction matrix parameter described in setting is such as formula shown in (5):

$T=\left[\begin{array}{cccc}& & & p_4\\ & B& & p_8\\ & & & p_{12}\\ 0& 0& 0& 1\end{array}\right]---\left(5\right)$

Wherein, the expression formula of described B is such as formula shown in (6):

$B=\left[\begin{array}{ccc}{p}_1& p_2& p_3\\ p_5& p_6& p_7\\ p_9& p_{10}& p_{11}\end{array}\right]---\left(6\right)$

Described B is three-dimensional orthogonal matrix and mould is 1, p _i(i=4,8,12) distinguish the translational movement of denotation coordination initial point along x, y, z axle, p _i(i ∈ [1,12] ∩ i ∈ Z) is optimum solution p _optin element, wherein p _opt∈ R ^{12 × 1}.

Microstructure based on optical non-destructive detection provided by the invention is the technical characterstic of joining method and effect on a large scale:

The present invention, by setting characteristic matching region and characteristic matching point search scope, avoids the feature similarity of measurement structure or not obvious the brought matching double points location of mistakes and improves counting yield.By calculating the splicing regions defect after the structure preliminary registration that causes because of neighbourhood noise in correction matrix recoverable measuring process.Low degree of overlapping proposed by the invention greatly reduces because repeating the overlapping considerable additional measurement time brought in original measuring method, therefore, it is possible to microstructure optical non-destructive detection on a large scale under effectively realizing low degree of overlapping.

Accompanying drawing explanation

Fig. 1 gives two adjacent structures to form schematic diagram.

Fig. 2 gives multiple adjacent structure splicing schematic diagram.

Fig. 3 gives the microstructure that the present invention is based on optical non-destructive detection joining method process flow diagram on a large scale.

Fig. 4 gives measuring process schematic diagram.

Fig. 5 gives the characteristic matching region using the inventive method to obtain.

Fig. 6 gives match point region of search schematic diagram.

Fig. 7 gives characteristic matching result when not setting described Feature Points Matching region and match point region of search.

Fig. 8 gives the Feature Points Matching region described in setting and the characteristic matching result after match point region of search.

Fig. 9 (a) and Fig. 9 (b) sets forth preliminary registration structure without matrix correction and outline line thereof.

Figure 10 (a) and Figure 10 (b) sets forth the final splicing construction after matrix correction and outline line thereof.

Embodiment

Below in conjunction with drawings and embodiments, the present invention is described in further detail.

As shown in Figure 1, there are two adjacent structure A1 to be spliced and A2 of certain overlapping region, by the microstructure based on optical non-destructive detection provided by the invention on a large scale joining method obtain splice result A3, to realize large-range measuring.As shown in Figure 2, on the basis of described splicing result A3, adopt the microstructure based on optical non-destructive detection provided by the invention joining method on a large scale, multiple splicing A5-AN with the adjacent structure of overlapping region can be realized.

As shown in Figure 3, the microstructure based on optical non-destructive detection provided by the invention on a large scale joining method mainly comprise the following step carried out in order:

Step 1: the kinematic parameter of test macro displacement platform experimentally, setting characteristic matching region is the lap of two adjacent structures to be spliced;

Characteristic matching region described in setting is that two adjacent Structural superposition parts to be spliced can effectively reduce in microscopic field of view due to the matching double points location of mistakes that structure high similarity causes.Before feature extraction, according to the level of the article carrying platform that can accurately obtain in optical precision nondestructive measurement system and the relative displacement between vertical movement position and adjacent structure, delimit characteristic matching region, mate with eliminating error and improve computing velocity.Give measuring process schematic diagram in Fig. 4, measure test sample A, each x direction translational displacement is L.Described relative displacement relation is such as formula shown in (1):

A ₁(x,y)+L＝A ₂(x,y) (1)

Wherein, (x, y) is two adjacent Structural superposition region to be spliced point midway coordinates, A ₁, A ₂be two adjacent structures to be spliced.

The relation of two adjacent neighbours Structural superposition degree to be spliced and relative displacement is such as formula shown in (2):

(P-L/p)/P＝a (2)

Wherein, P is the total number of structure x direction pixel to be spliced, and p is pixel equivalent dimension, and a is two adjacent Structural superposition degree to be spliced.Delimit structure A to be spliced ₁middle matching area is that x direction P (1-a) row arrange to P, delimits structure A to be spliced ₂middle matching area is that x direction the 1st arranges P (1-a) row, what deserves to be explained is, when applying above-mentioned formula acquired results not for integer, for ensureing that lap is included, the present invention gets the round values solution making setting regions larger.According to described characteristic matching area setting method, setting characteristic matching region as shown in Figure 5.In Fig. 5, label 7 and 8 is the two-dimentional vertical view of two adjacent structure A1, A2 to be spliced respectively, and the characteristic matching region of setting is the overlapping region 9 and 10 of two adjacent structures.

Measuring process for y direction is similar to described above-mentioned measuring process.

Step 2: extract architectural feature in the characteristic matching region adopting SURF (Speed Up Robust Features) algorithm to set in described step 1;

Step 3: according to test macro displacement platform uncertainty, setting characteristic matching point search scope, improve coupling accuracy and precision, avoid that structural similarity is too high or feature is abundant and cause the location of mistakes of characteristic matching point, and obtaining characteristic matching point according to Euclidean distance nearest-neighbor method;

Described matching double points range set is such as formula shown in (3) and formula (4):

x ₁+L _x-r _x≤x ₂≤x ₁+L _x+r _x(3)

y ₁+L _y-r _y≤y ₂≤y ₁+L _y+r _y(4)

Wherein, (x ₁, y ₁) and (x ₂, y ₂) be one group and belong to two adjacent structure A to be spliced respectively ₁, A ₂in matching double points, (L _x, L _y) be respectively adjacent structure A to be spliced ₁, A ₂the relative displacement of x, y direction, (r _x, r _y) respectively defining the hunting zone of x, y direction character match point, its value is determined by measuring equipment uncertainty.Fig. 6 gives characteristic matching point search range set schematic diagram, and unique point 12 finds the unique point that matches in hunting zone 11, and obtains characteristic matching point 13 according to Euclidean distance nearest-neighbor method.As shown in Figure 7, clear for making the error characteristic matching double points obtained when not setting described characteristic matching region and characteristic matching point search scope show, omit proper characteristics match point in this figure.As shown in Figure 8, the correct matching double points for being obtained with characteristic matching point search scope by the characteristic matching region 9,10 described in setting.

Step 4: with the local continuity of overlapping region for foundation, utilize STLS algorithm to calculate correction matrix, the splicing construction changing of the relative positions brought with environmental perturbation in correction measurement process, is finally spliced result thus.

The high precision of measuring mechanism makes it very sensitive to the environmental perturbation in test process, and therefore splicing regions can inevitably existing defects after preliminary registration for structure.Calculate correction matrix according to the local continuity of overlapping region by STLS (Scaled Total LeastSquares) algorithm and eliminate this defect, the final precision that realizes is spliced.In this step of the present invention, correct-by-construction does not relate to dimensional variation and malformation, provides the matrix restraint condition in described STLS algorithm, and the correction matrix parameter described in setting is such as formula shown in (5):

$T=\left[\begin{array}{cccc}& & & p_4\\ & B& & p_8\\ & & & p_{12}\\ 0& 0& 0& 1\end{array}\right]---\left(5\right)$

Wherein, the expression formula of described B is such as formula shown in (6):

$B=\left[\begin{array}{ccc}{p}_1& p_2& p_3\\ p_5& p_6& p_7\\ p_9& p_{10}& p_{11}\end{array}\right]---\left(6\right)$

Described B is three-dimensional orthogonal matrix and mould is 1, p _i(i=4,8,12) distinguish the translational movement of denotation coordination initial point along x, y, z axle, p _i(i ∈ [1,12] ∩ i ∈ Z) is optimum solution p _optin element, wherein p _opt∈ R ^{12 × 1}.

Fig. 9 (a) and Fig. 9 (b) are the preliminary align structures without matrix correction and the outline line of black line mark part respectively, and Figure 10 (a) and Figure 10 (b) is respectively the outline line of the final splicing construction after matrix correction and black line mark part.Hardware device used in the present invention is all based on existing business equipment.

Claims (4)

1. A large-scale mosaic method of microstructure based on optical non-destructive testing, characterized in that: it comprises the following steps carried out in order: Step 1: According to the motion parameters of the test system translation stage in the experiment, set the feature matching area as the overlapping part of two adjacent structures to be spliced; Step 2: using the SURF algorithm to extract structural features in the feature matching area set in step 1; Step 3: According to the uncertainty of the displacement platform of the test system, set the search range of feature matching points, and obtain the feature matching points according to the nearest neighbor method of Euclidean distance; Step 4: Based on the local continuity of the overlapping area, the STLS algorithm is used to calculate the correction matrix to correct the dislocation of the stitching structure caused by the environmental disturbance during the measurement process, thereby obtaining the final stitching result. 2. the microstructure large-scale splicing method based on optical non-destructive testing according to claim 1, is characterized in that: in step 1, described according to the motion parameter of test system translation stage in the experiment, setting feature matching area is The specific method for the overlap of two adjacent structures to be spliced is: According to the horizontal and vertical movement positions of the object-carrying platform and the relative displacement between adjacent structures that can be accurately obtained in the optical precision non-destructive measurement system, the feature matching area is delimited to eliminate wrong matching and improve the calculation speed; the relative displacement relationship As shown in formula (1): A ₁ (x,y)+L=A ₂ (x,y) (1) Among them, (x, y) are the coordinates of the midpoint of the overlapping area of two adjacent structures to be spliced, A ₁ and A ₂ are the two adjacent structures to be spliced; L is the displacement in the x or y direction; The relationship between the overlapping degree of two adjacent structures to be spliced and the relative displacement is shown in formula (2): (P-L/p)/P=a (2) Among them, P is the total number of pixels in the x direction of the structure to be spliced, p is the pixel equivalent size, and a is the overlapping degree of two adjacent structures to be spliced; the matching area in the structure A ₁ to be spliced is defined as the P(1- a) column to column P, define the matching area in the structure to be spliced A ₂ as column 1 to column P(1-a) in the x direction, when the result obtained by applying the above formula is not an integer, in order to ensure that the overlapping part is included in In , take the integer value solution that makes the setting area larger; The measurement process in the y direction is the same as the above measurement process. 3. The microstructure large-scale mosaic method based on optical nondestructive testing according to claim 1, characterized in that: in step 3, according to the uncertainty of the displacement platform of the test system, the search range of feature matching points is set, And the specific method of obtaining the feature matching point according to the Euclidean distance nearest neighbor method is; The range setting of the matching point is shown in formula (3) and formula (4): x ₁ +L _x -r _x ≤ x ₂ ≤ x ₁ +L _x +r _x (3) y ₁ +L _y -r _y ≤y ₂ ≤y ₁ +L _y +r _y (4) Among them, (x ₁ , y ₁ ) and (x ₂ , y ₂ ) are a pair of matching points belonging to two adjacent structures A ₁ and A ₂ to be spliced respectively, and (L _x , L _y ) are adjacent The relative displacements of the structures A ₁ and A ₂ to be spliced in the x and y directions, (r _x , _ry ) respectively define the search range of feature matching points in the x and y directions, and their values are determined by the uncertainty of the measurement equipment. 4. The microstructure large-scale mosaic method based on optical non-destructive testing according to claim 1, characterized in that: in step 4, based on the local continuity of the overlapping regions, the STLS algorithm is used to calculate the correction matrix, To correct the dislocation of the splicing structure caused by the environmental disturbance during the measurement process, the specific method to obtain the final splicing result is: In this step, structural correction does not involve scale change and structural deformation. Given the matrix constraints in the STLS algorithm, set the correction matrix parameters as shown in formula (5): $\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}& & & {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 4</span>}}\\ & \mathrm{<span\; class="notranslate">\; B</span>}& & {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 8</span>}}\\ & & & {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 12</span>}}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$ Wherein, the expression of described B is as shown in formula (6): $\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{ccc}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}& {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}& {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}\\ {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 5</span>}}& {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 6</span>}}& {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 7</span>}}\\ {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 9</span>}}& {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 10</span>}}& {\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 11</span>}}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$ The B is a three-dimensional orthogonal matrix with a modulus of 1, p _i (i=4,8,12) respectively represent the translation of the coordinate origin along the x, y, z axes, p _i (i∈[1,12] ∩i∈Z) is the element in the optimal solution p _opt , where p _opt ∈R ^12×1 .