TWI261662B - Subpixel precision estimating system of light point distance with multiple measurement purposes - Google Patents

Abstract

The present invention relates to a subpixel precision estimating system of light point distance with multiple measurement purposes, comprising: first, transforming the data captured from two light points to a first strength function and a second strength function; and estimating a precise light point distance by the use of a relevant approximate function and the multinomial interpolation method. Thus, the method of the present invention does not have to build a sample pattern of a light point model before calculating the light point distance, providing a simple and quick method for the measurement with wider application scopes.

Description

1261662 IX. Description of the invention: - Technical field of the invention The present invention relates to a sub-pixel precision estimation system for multi-measurement use spot spacing, in particular to a multi-measurement use spot using a correlation approximation function The sub-pixel precision estimation system of the pitch has the advantages and functions of not requiring the sample pattern or the spot model to be established in advance, the calculation is simple, fast and accurate, and the application range is wide. [Prior Art] • In the digital image processing analysis, the sub-pixel accuracy estimation method for single spot position has Centroid Calculation Method (CCM) and Fourier Phase Shift Method (FPS). And related approximation (Correlation Like, Method, referred to as CLM). In terms of the center of gravity estimation method, due to the reason of digital sampling, the algorithm itself generates an error term and the estimation accuracy is susceptible to noise. I In the Fourier phase shift method, although the estimation accuracy is less susceptible to noise and can overcome the DC off set, the phase analysis is more complicated. In the aspect of correlation approximation, it is necessary to establish a sample pattern or a spot model in advance, and use the interpolation method to achieve sub-pixel precision. However, it is very cumbersome to establish a sample pattern or a spot model in advance, and it can only be applied to this type, and the estimation is narrow. Therefore, it is necessary to develop a sub-pixel accuracy estimation technique that is simple, fast, and accurate. 5 1261662 SUMMARY OF THE INVENTION The main object of the present invention is to provide a multi-measurement measuring system that does not have to establish a sample pattern or a point type in advance, and can directly perform measurement. The purpose of X is to provide a human pixel accuracy estimation system for multi-measurement use, which is simple and fast to calculate. When the right is repeated multiple times, the accuracy is higher.

A further object of the present invention is to provide a sub-pixel accuracy estimation of a plurality of rangings, and a W-score estimation of the second-spot = gap of the optical sheet can also be modified into an optical strain gauge to estimate the gap between the optical strain gauges and The physical meaning of the representative; or, the wavelength of the beam can be reversed, so the range of applications is extremely wide. The present invention provides a sub-pixel precision estimation system for measuring the distance between the light and the stomach, which is used for measuring the micro-gap of the two-photo-light sheet, and includes:

A laser measuring module is used for emitting a laser beam, wherein the laser beam is deflected up and down and left and right; the second light-transmissive film is to be separated by a fine gap; the upper-image capturing device is used for After the laser beam obliquely penetrates the two light-transmissive sheets to be tested, the reflected plurality of beams become a light intensity distribution data; a metering device is used to extract the light intensity distribution data The two pre-light spot regions are respectively converted into a -th _ intensity function u(x) and a second intensity function f(x), a rib-intensity function u(x) and a second intensity function f(x) 6 1261662 And the following steps: -, taking the light spot step: reading the light spot and the second light spot intensity distribution data by the image capturing device; -, converting the first light spot step: converting the first light material Become a first strong production letter 1 ^ 'the knife is located at u (0); (1) 'and the peak of the first intensity is about two, the conversion of the second spot step: the first material is converted into a second The intensity function f (strong:: the peak of the shell is located at f(1); the brother-intensity function f〇o u(1)':= method of processing by law: using the first - After the intensity function is sampled, it becomes 8(4) "Function> 〇〇, via the digit::; get - the maximum value of r(1), r(I - 丨) is recorded. 5. Polynomial interpolation processing steps: Lixian ",...) The interpolation method of parabola,...1•(1) xc = — 2[Φ·-l) + r〇+ l)-2r(/)] : Ten different: the best center of gravity Xc, and then the distance between the spots is estimated to be - Calculate the fine gap of the transparent sheet to be tested by the ratio. The above objects and advantages are not to be understood in view of the following selected embodiments, the drawings, and the drawings. The following is the actual _ and with the H detailed Langben invention in the following: 1261662 [Embodiment] First, the relevant 仞, 本. 对 对 el el el el el el el , , , , , , , , , , , , , , , , , , , , , , , , , , , , , The function, called and ;, It is the light double (5) of the two light points and the correlation function is % (seven (10) ί+) / (Bu (2-1) = match reduction is even function, ie, (she ) can be further derived as follows: (4) - correlation function of the first intensity pass number r-> 〇〇 j

% fw(r)/(r-x)/r -3⁄4 !^ι〒(4)八卜如 lThu,x,®f(x) ;% lim — fT,T 4

TAH- ~TA f (x ) exp, — i2nsx)dx ^xp(i27. TA :\P〇i -T/ tZJ (2-2) From equation (2-2), the maximum value of r(x) is located. J. Eight or more is the derivation process of the correlation function. Please refer to the second and third figures. The present invention is a sub-pixel precision estimation system for multi-measurement light 1261662 point spacing, which includes the following steps. Step 11: Set the image of the 24th stomach and the light spot and the second spot intensity distribution with the image 撷 (4). As shown in the fourth figure, the ugly has four spots 25 (including 25A, 25B, 25C and 25D), however, the gap G of the two light-transmissive sheets 22, 23 to be tested in the third figure can be calculated as: 25B and 25C.

2. Converting the first light spot Step 12: Converting the first light spot intensity distribution data into a first intensity function u(1), as shown in the fifth figure, and the peak of the first intensity is located at u(0). 3. Converting the second spot. Step 13: Converting the first spot intensity distribution data into a second intensity function f(x), as shown in the sixth figure, and the peak of the intensity function f(x) is about f. (i). — IV. Correlation Approximation Process Step 14: Using the correlation approximation function r(x) of the first intensity function U00 and the second intensity function f(x), after serial sampling, a series of r(i) can be obtained. And find a maximum r(i) value and record the adjacent r(i + l) and r(i-1). V. Polynomial interpolation processing step 15 ·· Using the interpolation method of parabola containing at least r(i), r(i + l), r(i-1), xc=_out two D-W+I) C 2k0*~l) + r(/ + l)-2r(z)] Calculate the best center of gravity Xc, and then estimate the spot spacing L as i+Xc. Since the parabolic interpolation method can also be expanded to five or seven items, it is also equivalent to the 1261662 effect replacement: However, it is best to use the following steps: 6. Repeat the estimation N times Step 16: Repeat IV best The center of gravity Xc(N) is obtained in five steps - the estimated standard deviation as. , ,, leaf method calculation can be obtained - single seven, recursive operation step 17 · use

X The following estimation formula OPW ^^〇pt{N) +ΚΝ^(χ where σ; κ can be recursively calculated - the best estimate W. β is an example of an extremely simplified simplification, assuming the above The first step = the light-taking point step u, which has been extracted - the intensity knife cloth (X = 1 to x = 16, a total of 16 pixels) containing the image of the two light spots is: ^ One, using the second step Converting the first spot step 12, and learning that the intensity function u(x) is: and its original peak is u(5)=7, and the left shift is adjusted to u(0)=5 and the 'second intensity function f(x) is: 10 1261662 and its original peak value is f(13)=8, and f(8)=8 after shifting the same distance to the left. Using step 4 of the correlation approximation of the fourth step, the correlation function can be calculated. The next three places before η are i 2, 3x0+6x2+7x5+5x8+2x4+0x2=95 i=8 , 3x2+6x5+7x8+5x4+2x2=116 i=9 5 0x2+3x5+ After 6x8+7x44-5x2+2x0 = 101, the polynomial interpolation processing step 15 of the fifth step is performed.

r(/-l)-r(/ + l) — 95-101 2[r(i -1) + r(i + 1)- 2r(i)] ~ 2[95 + 101-2x116] So, most The center of gravity xc will fall at i=8+0. 083=8. 083, and then the estimated spot spacing L is i+xc. In other words, the spot distance L of the two spots is 8.083 pixels. In addition, if the recalculation step 17 of step 16 and the seventh step is further evaluated in conjunction with the sixth step, that is, the Karman filter is used to improve the estimated standard deviation, the accuracy of the present invention can be made. More improved. Please refer to Figure A of Annex I. This is a comparison of the actual curve and the theoretical curve using the Karman filter to improve the estimated standard deviation. It can be seen that the more times the repetition is repeated, the more the standard deviation is estimated. The lower the value, the more accurate the estimate is. The practical application of the present invention is numerous. The first application is to measure the gap G of two parallel glass. Referring to the third embodiment, a first embodiment of the sub-pixel precision estimation system for measuring the spot distance of the measurement uses a laser measurement module 21 for emitting a laser beam 211; 1261662 The light-transmissive sheet 22, 23 to be tested, the interval_gap G; the image capturing device 24, for capturing the teeth obliquely tilted by the laser beam 211, and reflecting the light to be measured 0 22, 23 The plurality of light beams 25B 25C, 25D become a light intensity distribution data; - the computing device 26 is configured to convert the predetermined light intensity distribution data 241 into a predetermined light spot area into a "first intensity function u (X) and - The degree function f(x)' uses the correlation function r(x) of the first intensity function u(x) and the second intensity function (X), and is digitally sampled to become ΊΑ

4, we can get a maximum value r (i), and the adjacent r (I + l) and r (IU = recorded; using at least r (1), Xian 1), parabolic interpolation, xc + Άτ {ί -1) + r(i + 1)- 2r(/)] Finally, the best center of gravity Xc is calculated, and then the spot spacing L· is i+xc Ο ^ The invention can also be applied to optical strain gauges. This is the second embodiment of the present invention (e.g., L). That is, a sub-pixel precision estimation system for multi-measurement use spot spacing, comprising: a laser summer measurement module 21 for emitting a laser beam 211; a learning strain gauge 27 having a gap G; The shirt image capturing device 24 is configured to capture a plurality of diffracted light spots 28 that are diffracted by the laser beam 211 penetrating the gap G of the optical strain gauge 27 > ^ η (refer to the eighth figure) , "Yanyankou" becomes a light intensity distribution data; 12 1261662 The foregoing computing device-calculation device 26, the details of this portion are the same as 26, and will not be repeated here. Age 匕' can accurately estimate 15l between the light spots, and then can reverse the degree of strain of =G and the gap G_ table. Therefore, the present invention can be applied to an optical strain gauge. Similarly, if the gap G is known, and the wavelength of the laser beam 211 is unknown, the wavelength of the field beam can be reversed by the distance between the spot points of the specific two diffracted light spots. It becomes a detecting instrument for the laser beam 211. In more detail, the laser measuring module 21 can be provided with a laser light source insertion hole (not shown, for example), for inserting a laser, emitting a laser beam 211, and the laser beam The measuring module 21 is provided with an adjustment button 212 for deflecting the laser beam 211 up and down and left and right; in terms of analyzing the intensity of the reflected light of the laser beam 211, the intensity distribution of the reflected light is confirmed in advance (as shown in the attached figure) ), in this study, we use the wave theory of light to estimate the reflected light intensity of R2, R3, R4, and R5. From the wave theory of light, the amplitude reflectance-s component: (2-1) (2-2) wide = -% cos 0 cos q COS0 + n2 cos02 amplitude reflectivity - p component: «2 COS0 -/2丨COS02 «2 COS^i + Nai COS02 Amplitude Transmittance - S Component: 13

1261662 ηλ cos^ +η2 cos 6^2 amplitude transmittance -ρ component: t -2% cos6^ , π---;~~~^- n2 cos0} cos 6^2 (2-3) (2-4 Since the laser light used is a non-polarized light source, the energy of both the p component and the s component is equal to half of the incident energy, so the energy reflectivity of the incident light is 1 = - r nx cos^j -n2 cos Y « 2 cos^j -nx cos02 ^ 2" 2 cos^j +n2 cos02J Kn2 cos^j +ηλ cos^ (2 - 5) If the absorption rate of the medium is neglected, then ρ + τ-\ (2-6) where r Is the transmittance of incident light. Assume: incident angle = 30.

«1 = 1 The dispersion constant of the tipping tip is calculated according to equations (2-1) to (2-6), and can be obtained as the seventh TT ^ 〇τ4ΐ 丄L 4 丄田一r*, /. ,., a ^ 5 mouth J figure does not result. From this result, D1 . n j can be further obtained.

Rl : R2 : R3 : R4 : R5 = 35.4 : 20.6 : 28 :16.2 ; 〇 After continuous image acquisition, the result is shown in Figure D of Annex I. ==Analytical verification' The result is as follows: R1: R2: R3: R4 = 28 ·· 24 : 26 : 23 ... 14 1261662 and the reflected light of the laser beam 211 In terms of geometric size analysis, in order to obtain the geometric relationship between the distance between the reflected light intensity R2 and R3 of the laser beam 211 and the gap size between the two glasses, the reflection law and the law of refraction of the geometrical optics can be obtained (2-7). ) p 2 sin6} In addition, in the analysis of the light field distribution of the laser beam 211, as shown in Figure F of Annex I, the optical system of the present study is shown. The laser beam 211 is reflected by the glass and then incident on the lens and sensed. The wafer (CCD) imaging, the light field distribution of the laser beam 211 after imaging through the optical path can be derived by the Fourier optical theory or the ABCD law of the Gaussian beam. The following study is based on the Gaussian beam. The light field distribution of the laser is assumed to be

E(r,z)=E0 and %w(z)

Kz-tan-1 Z 11 >exp jkr2 kz〇y. JJ _ 2q(z)_ (2-8) -j(~ λ q(z) R(z) 7rw(z) (2-9) where r -"x2 + y2,

2π Ύ z is the propagation direction of the Gaussian beam. a is the wavelength of light. 4 is the amplitude of the light field. % is the minimum spot size of the Nantes beam. 15 1261662 W is the spot size of the Nans beam at position Z. The radius of curvature of the wavefront of the light wave. And many measurement techniques are based on single-slit diffraction, such as optical strain gauges. In practical applications, since 1965, due to the homology of lasers, many optical strain gauges have been developed. The Tuckerman optical strain gauge is one of the simple and excellent optical strain gauges, such as Annex I. As shown in Figure G, the optical strain gauge is attached to the test piece by two blades. Usually, the blade pitch is a small distance, perhaps only a few micrometers. When the test piece is applied, the strain is caused by the strain. The blade distance increases, so when we inject the laser light into the slit between the blades, we will see that as the applied force increases, the spot is dense and dense, and the slit diffraction formula, we know the slit The larger the distance of the diffracted spot, the higher the strain value can be obtained. Interferometric optical strain gauges have the following advantages: [1] provides a method of measuring strain without the need for expensive strain gauges, and is easy to set up and the difficulty of bonding is reduced. [2] eliminates the problem of the enhancement effect of the material to be tested. [3] is a contactless measurement. [4] If the blade is made of the same material as the object to be tested, the result of automatic temperature compensation will be obtained. [5] For testing in high temperature environments. Since 1984, research on optical strain gauge experiments has gradually increased, but most of them use various fiber optic devices to perform strain measurement. It is worth noting that the multicolor optical strain gauge proposed by Brown in 1990 (POLYCHROMATIC OPTICAL STRAIN) GUAGE), and 1988 16 1261662

The grating and fiber optic (IMpRESSING GRATINGS WITHIN FIBER OPTICS) optical strain gauges proposed by Glenn use grating diffraction to perform strain measurement. The optical strain gauge experiment of this study uses single slit diffraction. Very similar. In recent years, the rapid development of laser precision measurement technology is due to the fact that the laser beam has the characteristics of twist collimation, directivity and homology.

Photon diffraction has been widely used. Huygens Curry·) — Fresnel principle can be used to explain the diffraction phenomenon. In short, the intensity of the light wave that is difficult to change in the space is from the point source on the wavefront. The sum on. Every time through the slit: Γ =: When the surface wave reaches the screen, it will interfere due to its optical path difference. Therefore, diffraction, ν π, Q 1 ～ ^ can be regarded as interference of many light rays. ~疒::』H shows a single-slit diffraction, with the PQ point at the intermediate point: each has the same optical path, which is a constructive interference and is a bright spot. The difference between rl λ/2 is 1/2; (4) ——When the light reaches the screen, the optical path can be found... The light should be destructively interfered, so Pi is a dark point. 772乂=lansin0 Constructive interference: Destructive interference is known from the above equation, the relationship sin 0 = again /a regular relationship, when subtracting from the first dark point position, its relative slit angle • the width of the diffraction fringe The gap of the slit width a and the wavelength of the J slit or the increase of the incident light wavelength can widen the 1261662 stripe. For the bright spot P1, the smaller the slit width a is, the larger the sing is. Thus, the diffraction phenomenon has a function of small signal amplification. As shown in m1, the synthesized electric field on the screen can be expressed as [2-3]: it = E\+E2+A +En ', when the destructive interference light privately is λ /2, the two adjacent lights When the phase difference of the point source is △, the vector sum is zero, that is, the dark point P1, and the light intensity is zero. As shown in the top of Annex I, the light intensity of the remaining P on the screen can be expressed by the following formula: where φ = ηΔ The light intensity of φ Ι〇 is the Ρ〇 point as shown in the Annex - Figure 为 is the single-slit diffraction pattern contrast, the ratio of the light intensity of the middle bright spot to the light intensity of other bright spots is: 〇047.

0.0"7;0·008;... 'The main bright band is at the center point, and the center points of the other crust strips are respectively at the phase difference of the two adjacent spot sources: + 1· 4303 7Γ ; ± 2· 4560 7Γ; soil 3 47Π7 7Γ · a π -· 707 τγ,···. As for the dark line, the phase difference between the two adjacent spot sources is: ± redundant; ± 2 dead; + as for the center bright bandwidth:

a bright band with no light, 1 degree is 2 times the other secondary bright bandwidth is decreasing in order. 18 1261662 = The constant slit b is - a small distance, only a few micrometers, when the test distance is 'extended deformation effect', the slit b between the copper sheets has ^/laser light incident on the copper sheet Between the slits b, there will be a light-emitting point on the screen. As the applied force increases, the larger the slit b(4), the more the spot will be, and the strain gauge will use this characteristic to make the test piece strain. Measure. False π and b are the slits of the two copper sheets pasted, then y where η is the number of dark points from the middle of the diffraction pattern, and the distance between the intermediate points (as shown in Figure L of Annex I). The test piece is subjected to: Day: The amount of deformation causes the gap width to change, that is, Shou, £ △ b. The distance between the dark point and the midpoint before the deformation is y. When the distance between the test/dark point and the midpoint is yi, we can Obtained: 4 o' yx b + Ab The relationship between b and y before deformation is as follows:

Under XR: The two equations are subtracted and combined to obtain the average green 曰% and Ϊ1 iyxy after the applied force. In the inspection example of the diffraction pattern of the grating, the diffraction pattern of the light 19 1261662 grid is often used. To test the spectrum and wavelength of the light wave, the early spectrometer used surface etching to form a grating to split the incident light. From the diffraction formula of light: d sin Θ = η λ Take the differential on both sides: D cos θ 5 <9 5 λ Here, the angular dispersion power of the grating is set to Όθ — δθ/ δλ

It can be seen from the above equation that the angular dispersion power is proportional to the grating pitch d, the diffraction order η, and also to the total number of slits of the diffraction grating. The angular scatter ability can only tell us how much the center of the line is separated, but it does not reflect whether the edges of these center-separated lines overlap or not. If you want to know whether the two lines overlap, you must know each The half angle of the bar line # u θ, according to the previous raster pattern, we know that this is related to the number of the light thumb, because when the incident light passes through the number of grating stripes, the diffraction spot is fine, and when When the incident light is incident only on the -first grid, we will find that the diffracted point is pulled long and the dark band is almost only a narrow point. According to Raileigh's law (Raileigh "sa Criteeri〇n), as shown in the attached figure, you can get:

Asin( 6» )-η Δ λ /d δίη(^+Δ^ )-(η+1/Ν) Λ/d 20 1261662 sin( ^ +Δ ^ )- sin( Θ )= Αθ δ sin( 0) / δ Θ =Δ Θ cos( θ ) According to the grating pattern, we know that this is related to the number of gratings, because when the incident light passes through a large number of grating stripes, the diffraction spot is fine, and when the incident light is only When we shoot a pair of gratings, we will find that the diffracted bright band is very long, and the dark band has almost only one narrow point. Δ^ = λ/(Ν d cos^ ) where N is the total number of slits of incident light passing through the diffraction grating. In summary, the advantages and effects of the present invention can be summarized as [1]. It is not necessary to establish a sample pattern or a spot model in advance. The invention can directly measure the sub-pixel precision of the two-spot spacing, especially for measuring the gap between two liquid crystal display (LCD) panels, which is of great benefit to the optoelectronic industry. ° [2] · Calculation is simple, fast and accurate. The present invention does not require a cumbersome mathematical conversion as in the conventional Fourier phase shift method, and is more accurate than the conventional center of gravity estimation method. In practice, the accuracy of the present invention can be as high as 〇· 〇 5 pixels, overall, The calculation is simple, fast and accurate. _ is to use the principle of Kalman filter to improve the accuracy of the estimation if it is repeated with N times and recursive steps. [3] · Wide range of applications. Since the system of the present invention can be used not only to estimate the spot pitch, but also to push back the gap between the two light-transmissive sheets to be tested. It can also be modified into an optical strain gauge, using the estimated light-off distance, and then deriving the gap between the optical strain gauges and the physical meaning of the representation. Of course, if the gap is set to a known item, the wavelength of the laser beam can be reversed, so the range of application 1261662 is extremely wide. The present invention has been described in detail by the preferred embodiments of the present invention. From the above detailed description, those skilled in the art can understand that the present invention can achieve the foregoing objects, and the invention has been in compliance with the provisions of the patent law. [Annex 1] Figure A is a comparison between the actual curve and the theoretical curve using the Karman filter to improve the estimated standard deviation. Figure B is a schematic diagram of the distribution of reflected light intensity. Figure C is the theoretical calculation result of the reflected light intensity distribution. FIG. D is a schematic diagram of the actual image capturing result of the reflected light. FIG. E is a schematic diagram of the intensity intensity profile measurement result of the reflected light. FIG. F is a schematic diagram of the glass gap measuring optical system. FIG. G is an optical strain gauge embodiment. Schematic diagram of the first diagram is a schematic diagram of single-slit diffraction. Figure I is a vector diagram of the synthetic electric field. Figure J is a vector diagram of the dark-spot light field. Figure 示意图 is a schematic diagram of the single-slit diffraction pattern. Schematic diagram of the geometric relationship between the diffracted spot and the slit. Figure 2 is a schematic diagram of the grating pattern. 22 1261662 [Simplified illustration of the drawing] The first drawing is a schematic diagram of the related approximation theory of the present invention. The second drawing is the flow of the method of the present invention. Figure 3 is a schematic view of a first embodiment of the present invention. Figure 4 is a schematic diagram showing the intensity distribution of a plurality of beams of the present invention. A schematic view of a sixth schematic view showing a second function of the strength of the present invention

Figure 7 is a schematic view of a reflective embodiment of the second embodiment of the present invention. 11: Picking up a light spot, Step 13 converting a second light spot, Step 15, Polynomial interpolation processing, Step 17, Recursive operation step 211, Laser beam = Eight figure, the present invention The reflective spectrometer of the third embodiment shows the symbolic description of the symbolic component] %, FIG. 12 converts the first light spot, step 14 is related to the approximation, and the processing step 16 repeats the estimation step 21, the laser measurement module 212 adjusts Buttons 22, 23 to be tested light transmissive sheet 241 light intensity distribution data 26 computing device 28 diffracts the point f (x) dienergetic intensity function X. Best center of gravity Xopt Best estimate 24 Image capture device 25 '25A, 25B, 25C, 25D Spot 27 Optical strain gauge u(x) First intensity function r(x) Correlation approximation σ s Single estimation standard Difference L spot spacing G gap R b R2, R3, /7, spacing of the spacing, R5 reflected light intensity 23 1261662 Z Gaussian beam propagation direction A Light field amplitude size ^ Incident light transmittance A light wavelength % Gauss The minimum spot size of the beam The spot size of the beam at position z is the radius of curvature of the wavefront. The point of the crust (ie, the intermediate point) Pi dark point I 〇 is the light intensity of the point Δ φ phase difference

δ slit visibility (diffraction in the early slit) b slit width (in the single slit diffraction pattern contrast) y dark point and intermediate point distance ε average strain d grating spacing η diffraction order ΝSlit total 6 θ half-width 24

Claims (1)

1261662 X. Patent application scope: 1 · A sub-pixel precision estimation system for multi-measurement spot spacing, which is used to measure the fine gap of the two transparent sheets to be tested, including: · a laser measurement The module is configured to emit a laser beam with a laser beam on the upper, lower, left and right side of the adjustment button; - a transparent light body to be tested, separated by a fine gap; an image capturing device for capturing the laser beam After the light beam obliquely penetrates the two light-transmissive sheets to be tested, the reflected plurality of light beams become a light intensity distribution data; a computing device is configured to respectively convert the two predetermined light spot regions of the extracted light intensity distribution data Forming a first intensity function u(x) and a first-intensity function f(X), a (four) first-intensity function u(X) and a second intensity function f(X) and the following steps: 1. Stepping up the spot : reading a first light spot and a second light spot intensity distribution data by using an image capturing device; 1. converting the first light spot step: converting the first light spot intensity distribution data into a -first intensity function u(x), and the peak of the first intensity is located at U(0); — ------ The spot intensity distribution metric is converted into a second intensity function f(1), and the peak of the second intensity function f(X) is located at f(i); 4. Correlation approximation processing steps The heartbeat uses the correlation approximation function r(x) of the first intensity function u(x) and the second intensity function f(x), and after digit sampling, it becomes a maximum value r(i) And record the adjacent r(I + l) and r(Il); • 5. Polynomial interpolation processing steps: using a parabola containing at least r(i), r(i + l), rQ-1) Interpolation method, xc = _ 2 [Φ * -1) + γ ( / + 1) - 2 γ (〇) I ° ten = out of the center of gravity Xc, and then estimated the spot spacing is i + Xc, by the ratio The micro-gap of the light-transmissive sheet to be tested is different. 2 · ^ Applying the sub-pixel precision estimation system of the multi-measurement spot spacing described in item i of the range, wherein the computing device further comprises the steps: 6. Repeat the N steps: repeat the first to fifth steps to obtain the N best center of gravity Xe(N), and obtain a single estimated standard deviation by the existing statistical method (Js; #七, Recursive operation Step: Use the following estimation formula: xmw = χ〇Ρ_+ ΚΝ_'{χΝ 一乂郇(4), where κ = ~~^2_ ~^σ〇ρί{Ν-\) Then estimate the light spot Recursively calculates a sub-pixel accuracy estimation system with a best estimated value distance of i+xc 〇 a multi-measurement spot spacing, and 26 1261662 includes: a laser measurement module for issuing a laser beam having an adjustment button for deflecting the laser beam up and down and left and right; an optical strain gauge having a gap; and an image capturing device for capturing the penetration of the laser beam After measuring the light-transmissive sheet, the reflected plurality of beams become a light intensity distribution data; a computing device for converting the two predetermined spot regions of the extracted light intensity distribution data into a first intensity function u(x) and a second intensity function f(x), respectively, using the first intensity function u(1) The correlation function r(x) of the second intensity function f(X) is matched with the following steps: 1. Step of capturing the light spot: reading a first light spot and a second light spot intensity by an image capturing device Distribution data; one, the first light spot conversion step · the first light spot intensity distribution The lean material is converted into a -first-strength function u(x), and the peak of the first intensity is located at u(0); a third, the second light spot is converted: the first light spot intensity distribution is converted a second intensity function f(x), and the peak of the second intensity function f(X) is located approximately; then the money processing step: using the first intensity function u(x) and the second strength function The correlation approximation function Γ(χ), after the digital sampling becomes double j, can obtain a maximum value r(i) and record the adjacent r(I + l) and 27 1261662 r(Il); 5. Polynomial interpolation processing Step: obtain a maximum value r(i), and record the adjacent r(i + 1Mr(H); use a parabolic interpolation containing at least r(i), r(i + l), ^b丨) Complement, C 2[r(z -1) + r(i + 1)- 2r{i)] Finally, calculate the best center of gravity Xc, and then estimate the spot spacing as 'strain from the proportional optical strain gauge value. .4 ·- Multi-quantity _ pass-by-pixel sub-pixel accuracy estimation system, the basin ^ includes: a laser measurement module, which is provided with a laser light source insertion hole for inserting a laser and emitting a laser a beam, the laser measuring module is provided with an adjusting button for deflecting the laser beam up and down and left and right; a grating; an image capturing device for capturing the tilting of the laser beam; After the light-transmissive sheet, the reflected plurality of beams become a light intensity distribution data; a computing device is configured to convert the two predetermined light spot regions of the extracted light intensity distribution data into a first intensity function u(x) And a second intensity function f(X), using the correlation function r(x) of the first intensity function u(x) and the second intensity function f(X), and obtaining a maximum after being digitally sampled into J The value r(i), and the adjacent r(i + i) and - D are recorded; using a parabola 28 1261662 containing at least 1*〇), 1*〇+ 1), and steam Line interpolation method, χ — r(zl)-r(/ + l) c ~2[r(/-1) +r(z+ 1)-2r(〇] Finally, calculate the best center of gravity X. Then estimate the spot spacing as i + Xc and calculate the strain value of the optical strain gauge from the ratio.