CN103245309A - Method for compensating laser flatness measurement error - Google Patents

Abstract

The invention discloses a method for compensating a laser flatness measurement error. The method is characterized by consisting of sub-pixel image scattering compensation, benchmark dip angle compensation and spot abnormal change alignment compensation, so that the measurement accuracy of a laser flatness measuring system is effectively increased. In the sub-pixel image scattering compensation, the problem of scattering caused by long-distance transmission of laser beams is solved by using a pixel edge identifying method, and the vertical accuracy of planar measurement can be ensured. In the benchmark dip angle compensation, the problem of benchmark deviation of laser target placement is solved by using a dip angle height changing method, and the height benchmark of a laser target can be ensured; and in the spot abnormal change alignment compensation, the problem of abnormal change of spot size along with the change of a target distance is solved by using a reciprocating inching method, and the projection accuracy of a laser point is increased.

Description

A kind of laser evenness measurement Error Compensation method

Technical field

The present invention relates to the plane and detect and surface topography evaluation field, especially relate to a kind of laser evenness measurement Error Compensation implementation method.

Background technology

Along with the sustainable development of science and technology, the modern industry competition is growing more intense, and model change is accelerated, and accuracy requirement improves constantly.The continuous appearance of all kinds of new technologies, material, technology, design, processing, detection, the control of equipping to modern industry bring greater impact.Flatness is as one of the most basic device fabrication and detection technique index, and its technical requirement also improves constantly, especially in special equipment fields such as metallurgy, mining, electric power, space flight, aviation, boats and ships and large molds.

Flatness is called flatness again, refers to the deviation on the macroscopical concavo-convex height relative ideal plane that measured target spare has.Flatness belongs to the shape error in the morpheme error, and its tolerance range is two zones between the parallel plane.As previously mentioned, along with the development of modern industrial technology, the precision measurement demand on large scale plane becomes increasingly conspicuous.So-called large scale plane refers to that generally horizontal vertical two-dimension sizes all greater than the member plane of 1000mm, supports steel structure, blades of large-scale wind driven generator, large-sized rolling mills working face, heavy duty machine tools guide rail as large bridge.The common feature on above-mentioned large scale plane is: physical dimension is bigger, and flatness error is had relatively high expectations.For example, the flatness of the workplace of large-sized rolling mills is during greater than 100: 0.025 (100 centimetres concavo-convex amplitude＜=0.025 centimetre), and its defect rate will raise 30%.Therefore, the flatness of large scale flat work pieces detects with the evaluation of surface topography level and has very important practical significance and application prospect at the related a plurality of industrial circles of great equipment manufacturing and detection.

In recent years, both at home and abroad colleges and universities, scientific research institution are based on the difference of elevation principle between the fixing laser beam emitting device in measuring position and the mobile high precision linear array photoelectric coupled device receiving target, carry out the large scale plane and detect and the research of surface topography assessment technique, and be applied in traffic and field of civil engineering.2006, people such as Ni Fujian analyzed respectively logistic recurrence, multiple regression, these 3 kinds of modeling methods of time series, according to highway flatness measured data, had set up several time series surface evenness forecast models that the varying number lagged value is arranged.According to the comparison of measured value, find out optimum time series surface evenness forecast model [Southeast China University's journal (natural science edition), 2006].2008, people such as Ma Ronggui proposed a kind of dual sensor vertical section and have detected principle, have provided computer artificial result, and have designed the laser evenness detection system based on embedded IP-2022.This system can calculate the vertical curve on tested road surface, calculates flatness index [Wuhan University of Technology's journal (traffic science and engineering version), 2008] thus.2011, people such as P.B.Tang proposed a kind of method measuring flatness based on laser scanning.This method is on the basis of three kinds of flatness data processing algorithms of contrast, the measuring system embodiment is carried out preferably, and then make measuring accuracy obtain to guarantee [Journal of Computing in Civil Engineering, 2011].

Present high-precision planeness detection system substantially all is based on the difference of elevation principle between the laser beam emitting device of fixing the measuring position and the high precision linear array photoelectric coupled device receiving target of movement.Though this method can record the flatness of each measuring point of tested plane accurately, when measuring, can't obtain the positional information of photoelectricity coupling receiving target.The position that at every turn needs manual record difference of elevation and receiving target correspondence to lay is unfavorable for the multimetering in batches of big plane.Said method does not relate to problems such as laser light scattering, hot spot mutation, benchmark inclination as yet in measuring process, and makes that measuring accuracy can not get guaranteeing.At the problems referred to above, the present invention proposes a kind of laser evenness measurement Error Compensation implementation method.

Summary of the invention

In order to overcome many technical matterss that traditional large scale surface smoothness detection system is brought, the invention provides laser evenness measurement Error Compensation implementation method.

The technical solution adopted for the present invention to solve the technical problems is:

A kind of laser evenness measurement Error Compensation method is characterized in that, compensates three parts by sub-pix image scatter compensation, benchmark pour angle compensation, hot spot mutation centering and forms, and improves the measuring accuracy of laser plane degree measuring system effectively.

Described sub-pix image scatter compensation is to utilize the pixel edge recognition methods to solve laser beam owing to the scattering problem that long Distance Transmission produces, and can make the vertical precision of plane surveying be guaranteed.

Described benchmark pour angle compensation utilizes conversion height method in inclination angle to solve the datum drift problem that laser target is placed, and can make the altitude datum of laser target be guaranteed.

The compensation of described hot spot mutation centering utilize spot size that reciprocal fine motion method solves laser along with the variation of target range the mutation problem, improve the projection precision of laser spots.

Further, described sub-pix image scattering compensating method is characterized in that, according to the recognition methods of laser facula pixel edge, and the gray scale f of pixel (n) _nGray scale f with pixel (n+1) _N+1Difference (f _N+1-f _n) greater than the absolute value of threshold values and gray scale also greater than threshold values, therefore can get the edge and be the centre position of pixel (n) and pixel (n+1), and then can get the central point of the line array CCD induction laser on the laser target at the point midway of high sensitization section.

The specific implementation step of sub-pix image scattering compensating method is: the gray-scale value output for a certain pixel can be expressed as:

$f(i,j)={\∫}_{j-0.5}^{j+0.5}{\∫}_{-0.5}^{+0.5}g(x,y)\mathrm{dxdy}---\left(1\right)$

In the formula (1), (x y) is the light distribution of consecutive image to g.In the formula as can be seen the gray scale function f (i j) is the result of each several part light intensity combined action on the pixel light-sensitive surface, and sampled result is to be the discrete matrix of numerical value with the gray-scale value.The expression formula of Gaussian curve is:

$y=\frac{1}{\sqrt{2\mathrm{\π\σ}}}\exp \left[\frac{{-(x-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}\right]---\left(2\right)$

In the formula (2), μ is average, and σ is standard deviation.The following formula both sides are taken the logarithm:

$\ln y=-\frac{{(x-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}+\mathrm{<figure-callout\; id="1"\; label="ln"\; filenames="HSA00000900164700011.png"\; state="\{\{state\}\}">ln</figure-callout>}\frac{1}{\sqrt{2\mathrm{\&pi;\&sigma;}}}---\left(3\right)$

Make y ^*=lny, then following formula is converted into:

y ^*＝Ax ²+Bx+C (4)

According to square footpath sampling thheorem, the pixel grey scale difference is:

$y^{*}\left(n\right)={\&Integral;}_{n-0.5}^{n+0.5}({\mathrm{Ax}}^2+\mathrm{Bx}+C)\mathrm{dx}---\left(5\right)$

Making the sequence number of the point of gray scale difference value maximum is n, and gray scale difference value is expressed as f _n, the sequence number of four pixels that the left and right sides is adjacent is respectively n-2, n-1, n+1 and n+2, and corresponding value is expressed as f respectively _N-2, f _N-1, f _N+1And f _N+2, the gray scale difference that can obtain five neighbor outputs according to following formula is respectively:

$f_{n-2}={\&Integral;}_{-2.5}^{-1.5}({\mathrm{Ax}}^2+\mathrm{Bx}+C)\mathrm{dx}=\frac{49}{12}A-2B+C---\left(6\right)$

$f_{n-1}=\frac{13}{12}A-B+C---\left(7\right)$

$f_n=\frac{1}{12}A+C---\left(8\right)$

$f_{n+1}=\frac{13}{12}A+B+C---\left(9\right)$

$f_{n+2}=\frac{49}{12}A+2B+C---\left(10\right)$

According to above-mentioned (6)～(10) five formula Simultaneous Equations, try to achieve A, B, C with least square method, again with the value substitution para-curve apex coordinate value x=-B/2A of A, B, C, obtain parabolical apex coordinate and be

$x=-\frac{-0.2f_{n-2}-0.1f_{n-2}+0.1f_{n+1}-0.2f_{n+2}}{2(0.1429f_{n-2}-0.0714f_{n-2}-0.1429f_n-0.0714f_{n+1}+0.1429f_{n+2})}---\left(11\right)$

Because formula (3) y value in the conversion process of (4) has been got logarithm, so the pixel grey scale difference also should remove logarithm, so sub-pixel edge extracting formula is:

$x_g=-\frac{-0.2\ln f_{n-2}-0.1\ln f_{n-2}+0.1\ln f_{n+1}-0.2\ln f_{n+2}}{2(0.1429\ln f_{n-2}-0.0714\ln f_{n-2}-0.1429\ln f_n-0.0714\ln f_{n+1}+0.1429\ln f_{n+2})}---\left(12\right)$

Behind the pixel edge location, obtain the gray-scale value of the neighbor point of this gradient direction and do difference processing, in the substitution formula (12), can try to achieve the marginal point of sub-pixel precision.

Further, described benchmark pour angle compensation method is characterized in that, utilizes twin shaft high-precision tilt angle sensor to obtain the angle of inclination of current laser target, utilizes the trigonometric function relation to obtain actual laser target height.

Further, described multi-target radio laser target, it is characterized in that, adopt laser to patrol the edge variation problem that the hot spot centering method of sweeping can effectively solve laser facula repeatedly, its processing procedure is as follows: at first, after the line array CCD on the laser target perceives laser, send information by wireless network, notice generating laser startup laser patrols repeatedly sweeps operation.Generating laser carries out the scanning of fine motion back and forth with minimum angles with its laser scanning head in current location repeatedly after receiving the notice of laser target.On this basis, the diameter of the laser facula of gained is gathered in the control chip comparative analysis at every turn, and wherein the corresponding hot spot of Zui Da value is the sample frame that the center of laser overlaps with the center of CCD.

Beneficial effect of the present invention mainly shows:

1) the sub-pix image scatter compensation of the present invention's proposition is to utilize the pixel edge recognition methods to solve laser beam owing to the scattering problem that long Distance Transmission produces, and can make the vertical precision of plane surveying be guaranteed.

2) the benchmark pour angle compensation of the present invention's proposition utilizes conversion height method in inclination angle to solve the datum drift problem that laser target is placed, and can make the altitude datum of laser target be guaranteed.

3) the hot spot mutation centering compensation that proposes of the present invention utilize spot size that reciprocal fine motion method solves laser along with the variation of target range the mutation problem, improve the projection precision of laser spots.

Description of drawings

Fig. 1 is line array CCD integral principle synoptic diagram;

Fig. 2 is that pour angle compensation calculates synoptic diagram;

Fig. 3 is that hot spot mutation centering is handled synoptic diagram.

Embodiment

By reference to the accompanying drawings, the present invention is described in detail below.

1. agent technology thinking

A kind of laser evenness measurement Error Compensation method involved in the present invention, its agent technology thinking is, compensate three parts by sub-pix image scatter compensation, benchmark pour angle compensation, hot spot mutation centering and form, improve the measuring accuracy of laser plane degree measuring system effectively.

Related sub-pix image scatter compensation is to utilize the pixel edge recognition methods to solve laser beam owing to the scattering problem that long Distance Transmission produces in the above-mentioned technology path, can make the vertical precision of plane surveying be guaranteed.

Related benchmark pour angle compensation utilizes conversion height method in inclination angle to solve the datum drift problem that laser target is placed in the above-mentioned technology path, can make the altitude datum of laser target be guaranteed.

In the above-mentioned technology path compensation of related hot spot mutation centering utilize spot size that reciprocal fine motion method solves laser along with the variation of target range the mutation problem, improve the projection precision of laser spots.

2. concrete function structure and implementation method

1) the laser beam scattering error compensating method of handling based on sub-pix

Because scattering to a certain degree can take place in transmission course laser, and increases along with the increase of transmission range, cause a deviation for the measurement result of line array CCD.CCD is light integrators spare, and it carries out integration to the light intensity of throwing on its light-sensitive surface in the interval at a fixed time with the area of fixed size.The difference that the edge recognizer of general Pixel-level is based on neighbor pixel basically cooperates the feature gray scale identification of high light part to carry out, and the pixel physical separation on the CCD is its resolution.

As shown in Figure 1, horizontal ordinate is the pixel linear array of the sensitization section of line array CCD, and ordinate is the gray-scale value of pel array.Because the convolution effect of optical component and inevitably optical diffraction effect, become the form of gradual change through optical imagery at the gray-scale value of real space drastic change, the edge is characterized by a kind of intensity profile in image, according to the pixel edge recognition methods, and the gray scale f of pixel (n) _nGray scale f with pixel (n+1) _N+1Difference (f _N+1-f _n) greater than the absolute value of threshold values and gray scale also greater than threshold values, therefore can get the edge and be the centre position of pixel (n) and pixel (n+1), the pixel edge of the same other end is pixel (n+6) and the centre position of pixel (n+7).Therefore, use the pixel edge recognizer can get the central point of the line array CCD induction laser on the WLT target at the point midway of pixel (n+1) to the high sensitization section of pixel (n+6).

Therefore, the physical separation between the pixel of said method is the limit of the measuring accuracy of line array CCD.The sub-pixel method for identification of edge that this project proposes can be broken through the restriction of the physical separation between pixel, can make precision improve an order of magnitude in theory.Identification theoretical core in sub-pixel edge is that the gray scale difference value with neighbor fits to Gaussian distribution curve (the image border gray-value variation should be Gaussian distribution), the point of gray scale difference maximum is the true edge point, and namely the vertex position of Gaussian curve is true edge point position.

As previously mentioned, this method can aftereffect solves the Laser Transmission scattering and the CCD integration offset issue that causes, and its implementation procedure is as follows: the gray-scale value output for a certain pixel can be expressed as:

$f(i,j)={\&Integral;}_{j-0.5}^{j+0.5}{\&Integral;}_{-0.5}^{+0.5}g(x,y)\mathrm{dxdy}---\left(1\right)$

In the formula (1), (x y) is the light distribution of consecutive image to g.In the formula as can be seen the gray scale function f (i j) is the result of each several part light intensity combined action on the pixel light-sensitive surface, and sampled result is to be the discrete matrix of numerical value with the gray-scale value.The expression formula of Gaussian curve is:

$y=\frac{1}{\sqrt{2\mathrm{\&pi;\&sigma;}}}\exp \left[\frac{{-(x-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}\right]---\left(2\right)$

In the formula (2), μ is average, and σ is standard deviation.The following formula both sides are taken the logarithm:

$\ln y=-\frac{{(x-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}+\mathrm{<figure-callout\; id="1"\; label="ln"\; filenames="HSA00000900164700011.png"\; state="\{\{state\}\}">ln</figure-callout>}\frac{1}{\sqrt{2\mathrm{\&pi;\&sigma;}}}---\left(3\right)$

Make y ^*=lny, then following formula is converted into:

y ^*＝Ax ²+Bx+C (4)

According to square footpath sampling thheorem, the pixel grey scale difference is:

$y^{*}\left(n\right)={\&Integral;}_{n-0.5}^{n+0.5}({\mathrm{<figure-callout\; id="2"\; label="Ax"\; filenames="HSA00000900164700011.png"\; state="\{\{state\}\}">Ax</figure-callout>}}^2+\mathrm{Bx}+C)\mathrm{dx}---\left(5\right)$

Making the sequence number of the point of gray scale difference value maximum is n, and gray scale difference value is expressed as f _n, the sequence number of four pixels that the left and right sides is adjacent is respectively n-2, n-1, n+1 and n+2, and corresponding value is expressed as f respectively _N-2, f _N-1, f _N+1And f _N+2, the gray scale difference that can obtain five neighbor outputs according to following formula is respectively:

$f_{n-2}={\&Integral;}_{-2.5}^{-1.5}({\mathrm{<figure-callout\; id="2"\; label="Ax"\; filenames="HSA00000900164700011.png"\; state="\{\{state\}\}">Ax</figure-callout>}}^2+\mathrm{Bx}+C)\mathrm{dx}=\frac{49}{12}A-2B+C---\left(6\right)$

$f_{n-1}=\frac{13}{12}A-B+C---\left(7\right)$

$f_n=\frac{1}{12}A+C---\left(8\right)$

$f_{n+1}=\frac{13}{12}A+B+C---\left(9\right)$

$f_{n+2}=\frac{49}{12}A+2B+C---\left(10\right)$

According to above-mentioned (6)～(10) five formula Simultaneous Equations, try to achieve A, B, C with least square method, again with the value substitution para-curve apex coordinate value x=-B/2A of A, B, C, obtain parabolical apex coordinate and be

$x=-\frac{-0.2f_{n-2}-0.1f_{n-2}+0.1f_{n+1}-0.2f_{n+2}}{2(0.1429f_{n-2}-0.0714f_{n-2}-0.1429f_n-0.0714f_{n+1}+0.1429f_{n+2})}---\left(11\right)$

Because formula (3) y value in the conversion process of (4) has been got logarithm, so the pixel grey scale difference also should remove logarithm, so sub-pixel edge extracting formula is:

$x_g=-\frac{-0.2\ln f_{n-2}-0.1\ln f_{n-2}+0.1\ln f_{n+1}-0.2\ln f_{n+2}}{2(0.1429\ln f_{n-2}-0.0714\ln f_{n-2}-0.1429\ln f_n-0.0714\ln f_{n+1}+0.1429\ln f_{n+2})}---\left(12\right)$

Behind the pixel edge location, obtain the gray-scale value of the neighbor point of this gradient direction and do difference processing, in the substitution formula (12), can try to achieve the marginal point of sub-pixel precision.

2) pour angle compensation modification method

Because the result that system detects and the high-positive correlation of laser target are if therefore laser target places the precision that directly influence is measured.Inclination angle as laser target is α, and the laser elevation of measurement is h, and then Shi Ji laser elevation should be h=(h * cos α), as shown in Figure 2.

Though be equipped with air-bubble level and mechanical level(l)ing device on the base of laser target, following factor still can have influence on the angle of its installation: the systematic error of air-bubble level is ± 0.2 °; Mode of operation during manual shift water; The slight inclination of measured surface itself.

The correction of the inclination when therefore laser target being installed can directly improve the measuring accuracy value of system.SCA100T in the laser target is twin shaft high-precision tilt angle sensor, its measuring accuracy can reach ± and 0.001 °, when cooperating the laser target receiving target standard pole of 50mm model, its maximum error that induces one owing to the inclination angle is 10 ^-9Mm is far smaller than the whole Measurement Resolution of system.

3) hot spot mutation centering method

Owing to the variation along with target range of the spot size of laser changes, its shape is not the round of rule, but its shape can time to time change, and the shape of the hot spot of different distance also is similar shape, as shown in Figure 3.Generating laser scan mobile laser spots to the line array CCD of laser target in, the hot spot that CCD gathers might not be the center of laser, therefore CCD may collect the non-regular edge of laser facula, thereby cause the laser elevation calculated thus to produce little deviation, shown in the situation on the left side of figure below and right side.The center of having only the center of each laser all to drop on line array CCD could be eliminated because the error that the laser facula out-of-shape causes.

The hot spot centering method that the present invention adopts laser to patrol repeatedly to sweep can effectively solve the edge variation problem of laser facula, its processing procedure is as follows: at first, after the line array CCD on the laser target perceives laser, send information by wireless network, notice generating laser startup laser patrols repeatedly sweeps operation.Generating laser carries out the scanning of fine motion back and forth with minimum angles with its laser scanning head in current location repeatedly after receiving the notice of laser target.On this basis, the diameter of the laser facula of gained is gathered in the control chip comparative analysis at every turn, and wherein the corresponding hot spot of Zui Da value is the sample frame that the center of laser overlaps with the center of CCD.

At last, it is also to be noted that what more than enumerate only is example of the present invention.Obviously, the invention is not restricted to above example, many distortion can also be arranged.All distortion that those of ordinary skill in the art can directly derive or associate from content disclosed by the invention all should be thought protection scope of the present invention.

Claims (4)

1. a laser evenness measurement Error Compensation method is characterized in that, compensates three parts by sub-pix image scatter compensation, benchmark pour angle compensation, hot spot mutation centering and forms, and improves the measuring accuracy of laser plane degree measuring system effectively.

Described sub-pix image scatter compensation is to utilize the pixel edge recognition methods to solve laser beam owing to the scattering problem that long Distance Transmission produces, and can make the vertical precision of plane surveying be guaranteed.

Described benchmark pour angle compensation utilizes conversion height method in inclination angle to solve the datum drift problem that laser target is placed, and can make the altitude datum of laser target be guaranteed.

The compensation of described hot spot mutation centering utilize spot size that reciprocal fine motion method solves laser along with the variation of target range the mutation problem, improve the projection precision of laser spots.

2. sub-pix image scattering compensating method according to claim 1 is characterized in that, according to the recognition methods of laser facula pixel edge, and the gray scale f of pixel (n) _nGray scale f with pixel (n+1) _N+1Difference (f _N+1-f _n) greater than the absolute value of threshold values and gray scale also greater than threshold values, therefore can get the edge and be the centre position of pixel (n) and pixel (n+1), and then can get the central point of the line array CCD induction laser on the laser target at the point midway of high sensitization section.

The specific implementation step of sub-pix image scattering compensating method is: the gray-scale value output for a certain pixel can be expressed as:

$f(i,j)={\&Integral;}_{j-0.5}^{j+0.5}{\&Integral;}_{-0.5}^{+0.5}g(x,y)\mathrm{dxdy}---\left(1\right)$

In the formula (1), (x y) is the light distribution of consecutive image to g.In the formula as can be seen the gray scale function f (i j) is the result of each several part light intensity combined action on the pixel light-sensitive surface, and sampled result is to be the discrete matrix of numerical value with the gray-scale value.The expression formula of Gaussian curve is:

$y=\frac{1}{\sqrt{2\mathrm{\&pi;\&sigma;}}}\exp \left[\frac{{-(x-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}\right]---\left(2\right)$

In the formula (2), μ is average, and σ is standard deviation.The following formula both sides are taken the logarithm:

$\ln y=-\frac{{(x-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}+\ln \frac{1}{\sqrt{2\mathrm{\&pi;\&sigma;}}}---\left(3\right)$

Make y ^*=lny, then following formula is converted into:

y ^*＝Ax ²+Bx+C (4)

According to square footpath sampling thheorem, the pixel grey scale difference is:

$y^{*}\left(n\right)={\&Integral;}_{n-0.5}^{n+0.5}({\mathrm{Ax}}^2+\mathrm{Bx}+C)\mathrm{dx}---\left(5\right)$

Making the sequence number of the point of gray scale difference value maximum is n, and gray scale difference value is expressed as f _n, the sequence number of four pixels that the left and right sides is adjacent is respectively n-2, n-1, n+1 and n+2, and corresponding value is expressed as f respectively _N-2, f _N-1, f _N+1And f _N+2, the gray scale difference that can obtain five neighbor outputs according to following formula is respectively:

$f_{n-2}={\&Integral;}_{-2.5}^{-1.5}({\mathrm{Ax}}^2+\mathrm{Bx}+C)\mathrm{dx}=\frac{49}{12}A-2B+C---\left(6\right)$

$f_{n-1}=\frac{13}{12}A-B+C---\left(7\right)$

$f_n=\frac{1}{12}A+C---\left(8\right)$

$f_{n+1}=\frac{13}{12}A+B+C---\left(9\right)$

$f_{n+2}=\frac{49}{12}A+2B+C---\left(10\right)$

According to above-mentioned (6)～(10) five formula Simultaneous Equations, try to achieve A, B, C with least square method, again with the value substitution para-curve apex coordinate value x=-B/2A of A, B, C, obtain parabolical apex coordinate and be

$x=-\frac{-0.2f_{n-2}-0.1f_{n-2}+0.1f_{n+1}-0.2f_{n+2}}{2(0.1429f_{n-2}-0.0714f_{n-2}-0.1429f_n-0.0714f_{n+1}+0.1429f_{n+2})}---\left(11\right)$

Because formula (3) y value in the conversion process of (4) has been got logarithm, so the pixel grey scale difference also should remove logarithm, so sub-pixel edge extracting formula is:

$x_g=-\frac{-0.2\ln f_{n-2}-0.1\ln f_{n-2}+0.1\ln f_{n+1}-0.2\ln f_{n+2}}{2(0.1429\ln f_{n-2}-0.0714\ln f_{n-2}-0.1429\ln f_n-0.0714\ln f_{n+1}+0.1429\ln f_{n+2})}---\left(12\right)$

Behind the pixel edge location, obtain the gray-scale value of the neighbor point of this gradient direction and do difference processing, in the substitution formula (12), can try to achieve the marginal point of sub-pixel precision.

3. benchmark pour angle compensation method according to claim 1 is characterized in that, utilizes twin shaft high-precision tilt angle sensor to obtain the angle of inclination of current laser target, utilizes the trigonometric function relation to obtain actual laser target height.

4. multi-target radio laser target according to claim 1, it is characterized in that, adopt laser to patrol the edge variation problem that the hot spot centering method of sweeping can effectively solve laser facula repeatedly, its processing procedure is as follows: at first, after the line array CCD on the laser target perceives laser, send information by wireless network, notice generating laser startup laser patrols repeatedly sweeps operation.Generating laser carries out the scanning of fine motion back and forth with minimum angles with its laser scanning head in current location repeatedly after receiving the notice of laser target.On this basis, the diameter of the laser facula of gained is gathered in the control chip comparative analysis at every turn, and wherein the corresponding hot spot of Zui Da value is the sample frame that the center of laser overlaps with the center of CCD.

Images