CN109724690A - Schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting - Google Patents
Schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting Download PDFInfo
- Publication number
- CN109724690A CN109724690A CN201811528861.1A CN201811528861A CN109724690A CN 109724690 A CN109724690 A CN 109724690A CN 201811528861 A CN201811528861 A CN 201811528861A CN 109724690 A CN109724690 A CN 109724690A
- Authority
- CN
- China
- Prior art keywords
- image
- schlieren
- cutzb
- spot
- reduced size
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The present invention proposes a kind of schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting, can be improved the reconstruction accuracy of far-field focus.Method includes the following steps: firstly, being fitted the schlieren center of pellet of other spot image using least square method；Secondly, being cut respectively to main lobe image and side lobe image；Deduction one in main lobe reduced size image and the image cutzb ' after the circle of schlieren bead size and secondary lobe reduced size image cutpb are converted into two column vectors respectively again；Then, the charting of spectrum angle (SAM) value between two vectors is calculated, when two vector similitudes reach maximum value, then the upper left position of cutzb ' image is optimal match point；Finally, being reference with optimal match point, it is reconstructed using image cutzb' and other spot reduced size image cutpb focal spot, and splicing boundary is merged using using weighted mean method during final image mosaic.
Description
Technical field
The present invention relates to a kind of schlieren method far-field measurement focal spot reconstructing methods.
Background technique
The purpose of God Light III host apparatus parameter measurement integrated diagnostic system is to use schlieren method to high power laser light
The far-field focus distribution that device generates measures [1].After the fast automatic standard of comprehensive diagnos, which sets system, completes the aiming of optical path,
It can be burnt for final high dynamic range far field in the secondary lobe that the target practice stage obtains main lobe hot spot respectively and center is accurately blocked
The measurement of spot provides important leverage.Laser parameter integrated diagnostic system is a multi-functional, high-precision laser parameter diagnosis
System, major function are to complete laser beam energy, near field, far field, burst length waveform and light during the transmitting of device operation
The accurate diagnosis of the parameters such as spectrum obtains comprehensive, accurate device operating parameter.High dynamic range far-field focus [2] is that synthesis is examined
System of breaking is high power object to the important measurement point of frequency tripled laser focal spot and the difficult point of large scale laser instrument parameter diagnosis
Key technology in the urgent need to address in reason experiment.
Since the required dynamic range of far-field focus measurement in integrated diagnostic system is up to 1000:1, current monolithic is visited
Survey the dynamic range requirement that device is difficult to reach so high.Because energy is up to 10KJ when practicing shooting, if not carrying out big times to optical path
Rate decaying, then CCD is fully saturated；If decaying is excessive, although far-field focus can be measured, focal spot bottom is all by noise
It floods.Therefore, the measurement of complete hot spot distribution is realized, it is necessary to using the detection method of dim light sampling, beam splitting amplification imaging, i.e.,
Schlieren method [3].Schlieren method is the method for being used to measure the distribution of light laser far-field focus most popular at present, i.e. main lobe, secondary lobe point
Measurement is opened, surveys secondary lobe, main lobe amplification, the method that the focal spot measurement of high dynamic range is realized in splicing with the small ball light-blocking of schlieren.
Investigation on focal spot measurement by schlieren method optical path is as shown in Figure 1.Light beam is divided into two beams after spectroscope, and schlieren bead is put
In the focal position of secondary lobe optical path to block spot center, the hot spot that secondary lobe CCD is obtained is the facula information around schlieren bead,
Referred to as " secondary lobe ".And by light beam by appropriate decaying, what main lobe CCD was obtained is the hot spot not blocked, referred to as " main lobe ".Again will
One telescopic system and bead of looking in the distance are placed in the optical path before spectroscope, for demarcating main lobe light intensity amplification coefficient K and hot spot
Amplification coefficient b obtains the complete distribution of focal spot relative intensity finally by the reconstruct to main lobe and secondary lobe hot spot.
Schlieren method measures far field coke valve, and there are two difficult points: first is that main lobe and secondary lobe optical path have drift, collected master
Valve and side lobe image are of poor quality, can not accurately splice；Second is that for using main lobe and secondary lobe separate measurement method acquisition quality compared with
Good image is spliced using manual method, and without automatic reconstructing method [4-6], conventional efficient is low.
Summary of the invention
In order to realize the measurement of high dynamic range laser focal spot by integrated diagnostic system, the reconstruct essence of far-field focus is improved
Degree, the present invention propose a kind of schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting.
Technical scheme is as follows:
Firstly, the schlieren center of pellet of other spot image is fitted using least square method, less than 1 pixel of error precision.
Secondly, cutting respectively to main lobe image and side lobe image, the new image that cuts is size 300 × 300, is remembered respectively
For cutzb and cutpb；Again by deduction one in main lobe reduced size image and the image cutzb ' after the circle of schlieren bead size and side
Valve reduced size image cutpb is converted into two column vectors respectively；
Then, the charting of spectrum angle (SAM) value between two vectors is calculated, when two vector similitudes reach maximum value
When, then the upper left position of cutzb ' image is optimal match point；
Finally, being reference with optimal match point, weight is carried out using image cutzb' and other spot reduced size image cutpb focal spot
Structure, and splicing boundary is merged using using weighted mean method during final image mosaic.
The invention has the following advantages:
The present invention can be improved the accuracy of dynamic range far-field focus measurement and realize the automatic reconstruct of far-field focus.Cause
For optimal match point be method image procossing rather than it is traditional obtained by way of splicing by hand, this side
Method is insensitive for focal spot reconstruction accuracy, and can obtain higher experimental precision.
Detailed description of the invention
Fig. 1 is schlieren measure far-field focus schematic diagram.
Fig. 2 is process flow diagram of the invention.
Fig. 3 is original hot spot；Wherein, (a) main lobe image, (b) side lobe image.
Fig. 4 is to cut image for the first time；Wherein, (a) main lobe reduced size image, (b) secondary lobe reduced size image.
Fig. 5 is the schematic diagram for calculating side lobe image schlieren center of pellet；Wherein, (a) mathematical mor-phology method handles secondary lobe
Reduced size image (b) carries out edge detection to binary image using Sobel operator, and (c) first fit is as a result, (d) by fitting circle
Quadratic fit after outer marginal point gray value is set to 0 obtains real schlieren bead edge and circle fitting knot as a result, (e) detecting
Fruit will (f) be fitted the obtained schlieren bead center of circle and radius shown in side lobe image.
Fig. 6 is the schematic diagram for searching schlieren center of pellet optimal match point in main lobe image；Wherein, (a) secondary lobe is reduced
Image cutpb, (b) main lobe reduced size image cutzb, (c) main lobe deducts best match image after bead, and (d) secondary lobe cuts three-dimensional
Image, (e) the main lobe 3-D image cut, (f) main lobe deducts best match 3-D image after bead.
Fig. 7 is the schematic diagram for calculating the SAM value of cutzb and cutpb；Wherein, (a) secondary main lobe cuts the Crop Area of image
Domain, the main lobe for (b) finding optimal match point cut and deduct the figure of bead.
Fig. 8 is the SAM value schematic three dimensional views of 100 × 100 near-in sidelobe cutpb and all cutzb main lobe images.
Fig. 9 is to merge image 512 × 512；Wherein, (a) directly merges image stretch effect, (b) boundary fused image
Drawing effect, (c) final three-dimensionalreconstruction image.
Figure 10 is reconstruct focal spot horizontal direction two dimensional gray curve；Wherein, (a) primitive curve, the song after (b) taking log10
Line.
Figure 11 is splicing regions signature analysis；Wherein, (a) splices the position of ring, (b) splices characteristic area, (c) Three-dimensional Gravity
Composition picture.
Figure 12 be reconstruct each parameter of focal spot image between correlation schematic diagram, wherein (a) K ≈ 1, (b) K > > 1.
Figure 13 is 90DL calculated result.
Specific embodiment
The present invention is with schlieren measure far-field focus mathematical model [2] for theoretical foundation, and measures far field according to schlieren method
The principle [3] of burnt valve has selected two 12 science CCD to acquire the secondary lobe hot spot of main lobe hot spot and central shielding respectively.Firstly,
Main lobe image and side lobe image are cut respectively, it will be after the circle that one and schlieren bead size be deducted in main lobe reduced size image
Image cutzb ' and secondary lobe reduced size image cutpb be converted into two column vectors respectively；Secondly, calculating the light between two vectors
The charting of spectral corner degree (SAM) value, when two vector similitudes reach maximum value, then the upper left position of cutzb ' image is best
Match point；Finally, the schlieren center of pellet of other spot image is fitted using least square method, less than 1 pixel of error precision；And
Splicing boundary is merged using using weighted mean method during final image mosaic.The experimental results showed that the party
Method can be realized the precise measurement and automatic reconstruct of high dynamic range far-field focus, the splicing for the side lobe image of clean mark
Less than 1 pixel of error, meets the requirement for experimental precision and efficiency of Targeting, and integrated diagnostic system is obtained
Comprehensively, accurately device operating parameter has a very important significance.
Using autocorrelation matching method when conventional method calculates the optimal match point of main spot and other spot image, each image is to make
Operation is carried out with matrix-style, calculation amount is larger.The present invention proposes the automatic restructing algorithm mapped based on spectral modeling, and thought is
The image for carrying out relevant matches is needed to be converted into two column vectors respectively by two, then by calculating the light between two vectors
The charting of spectral corner degree (Spectral Angel Mapping, SAM) [7] value, i.e. angle value between two vectors judge two width
The similitude of image.Angle value between two vectors is calculated and is indicated with formula (1)
In formula: Vp--- secondary lobe reduced size image column vector generated, Vz--- deduct one together in main lobe reduced size image center
After the circle of schlieren bead size and the column vector that generates, the element that N --- the wide and high product of reduced size image and vector include
Number.
Assuming that original secondary lobe is expressed as Orgp, center is expressed as (Opx, Opy), and original main lobe image is Orgz, center indicates
For (Ozx, Ozy), the size of original main lobe image and side lobe image is all 512 × 512.Side lobe image Cut after reductionpTable
Show, the center of schlieren bead is exactly in secondary lobe reduced size image CutpCenter, indicate that actual value is with (Cpx, Cpy)
(151,151), the main lobe reduced size image Cut after reductionzIndicate, main lobe and side lobe image after reduction having a size of 300 ×
300.If the objective function of optimal match point are as follows:
F (m, l)=min (Sam (Vz,Vp)) (2)
Automatic restructing algorithm based on spectrum angle charting realizes that steps are as follows:
1) secondary lobe reduced size image Cut is obtainedp, in OrgpIn image, reduced centered on (Opx, Opy) having a size of 300*300
Image Cutp=Orgp(Opy-150:Opy+150,Opx-150:Opx+150-1)。
2) schlieren bead mapping matrix P is obtainedc
It then is column vector, P by the matrix conversionc=Pc(:), r are the small radius of a ball of schlieren.
3) from OrgzThe main lobe reduced size image Cut having a size of 300 × 300 is obtained in imagez=Orgz(Ozy-150+m:Ozy
+ 150+m, Ozx-150+l:Ozx+150-1+l), central point is located at (Ozx+l, Ozy+m), and wherein the value range of l value is -50
The value range of≤l < 50, m value is the coordinate that -50≤m < 50, l and m are relative to 100 × 100 rectangular area centers.
4) Cut is obtainedzThe transposed matrix V of image arrayz'=Cutz T, transposed matrix is expressed as column vector form Vz'=
Vz' (:), the image column vector representation V after obtaining the big roundlet of main lobe image deduction schlieren beadz=Vz'.*Pc T；It obtains
CutpThe transposed matrix V of image arrayp=Cutp T, transposed matrix is expressed as column vector form Vp=Vp(:)。
5) vector V is calculatedpAnd VzSAM value, i.e. Sam (Vz,Vp)。
6) 3-4 step is repeated, until the value of 100 × 100 all m and l of rectangular area has been calculated.
7) optimal match point is obtained by objective function Equation (2).
The automatic restructing algorithm of schlieren includes several important steps: 1) pre-processing；2) schlieren bead is calculated using schlieren method
Center；3) optimal match point of the schlieren center of pellet on main lobe hot spot is found；4) image merges.It is most important in these steps
The step of be to find optimal match point of the schlieren center of pellet on main lobe hot spot.The flow chart of data processing figure of automatic restructing algorithm
As shown in Figure 2.
1 pretreatment
Before processing is reconstructed to image, first have to subtract background respectively to original main lobe and side lobe image.Secondly, wanting
Image is cut, because the picture size of different type CCD acquisition is different, wherein there is size in ICF Targeting
The image of 1024 × 1024,12 scientific grade CCD acquisitions and the image of 2048 × 2048,16 scientific grade CCDs acquisition.Figure
3 be original main lobe and side lobe image.
Because main lobe and secondary lobe hot spot are relatively small, schlieren center of pellet is searched on main lobe by cutting to reduce
Calculation amount during optimal match point.The unified image that main lobe and side lobe image are cut to 512 × 512 sizes, remembers respectively
For orgzb and orgpb, main lobe and secondary lobe hot spot is allowed to be located at image center location as far as possible.It is cut shown in result figure 4 for the first time.
2 calculate schlieren center of pellet
In order to calculate schlieren center of pellet, schlieren bead edge is obtained using Sobel operator [8] first, then using most
Small square law carries out round fitting, to improve fitting precision.
If objective function indicates are as follows:
In formula: N --- participate in the number of the characteristic point of the Fitting Calculation.This is a non-linear least square problem, as point O
(x0,y0) near coordinate origin when, often with formula (2) replacement:
Assuming that the point number on all boundaries is N, (xi,yi) for the boundary coordinate of image, the central coordinate of circle (x fitted0,
y0) and radius r expression formula be [9]:
In formula: N --- marginal point sum, (xi,yi) --- the edge coordinate of side lobe image, r --- radius, (x0,
y0) --- schlieren center of pellet coordinate；Schlieren center of pellet and radius are in the side lobe image obtained using circle fitting formula
(Opx, Opy) and okr, the pixel value in side lobe image within schlieren bead are all set to 0.In order to calculate schlieren bead half
Diameter it may first have to the real edges of schlieren bead are detected, herein using the method detection schlieren bead for the fitting that iterates
The step of edge, detection schlieren bead real edges and calculating schlieren center of pellet, is as follows:
1) using binaryzation and mathematical mor-phology method processing secondary lobe reduced size image [10], as a result as shown in Fig. 5 (a).
2) using the edge [8] of Sobel operator detection side lobe image, as a result as shown in Fig. 5 (b).
3) the least square method fitting center of circle and radius, first fit result such as Fig. 5 are used using the edge image of Fig. 5 (b)
(c) shown in, and the pixel number that gray value is equal to 255 is counted, is denoted as allcount.
4) the marginal point gray value outside fitting circle is set to 0, as a result as shown in Fig. 5 (d), gray scale in statistical fit circle
Value is equal to 255 number of pixels, is denoted as count.
5) xs=count/allcount is calculated, if xs less than 0.995, continues to execute step 3) and 4).Otherwise, really
Schlieren bead edge be then detected, the center of circle of the schlieren bead of final schlieren bead edge detection results and fitting and
Shown in radius such as Fig. 5 (e).
6) the obtained schlieren bead center of circle will be fitted and radius is shown in side lobe image, shown in result figure 5 (f).
In in detection schlieren bead real edges above and the step of calculate schlieren center of pellet, from step 3) to 5) passing through
It iterates and is fitted center of circle radius, the condition that iteration terminates is less than this fitting center of circle and the gray scale pixel equal to 255
The number of pixels that number/all gray scales are 255, i.e. xs=count/allcount are greater than 0.995.By iterating, ratio system
Number increases to 0.997 from 0.509.Shown in obtained schlieren bead real edges such as Fig. 5 (e), the center of circle being finally fitted is
(250.55,256.17), radius 88.53.
Table 1, which iterates, is fitted the center of circle and radius result
3 find optimal match point of the schlieren center of pellet on main lobe
In order to quickly find optimal match point of the schlieren center of pellet on main lobe, main lobe image and side on main lobe image
Valve image has to pass through secondary reduction, and the image size reduced is 300 × 300, is indicated respectively with cutzb and cutpb, such as schemes
Shown in 6 (a) and 6 (b).
The step of identifying optimal match point is as follows:
1) center of gravity for obtaining orgzb image, is expressed as (Ozx, Ozy), obtains line in orgpb image using least square method
Shadow center of pellet and radius are expressed as (Opx, Opy) and okr..
2) the secondary lobe reduced size image for obtaining size 300 × 300, is expressed as cutpb, cutpb image is cut out from orgpb image
Subtract to obtain, center is located at (Opx, Opy).The center of secondary lobe reduced size image cutpb is expressed as (Cpx, Cpy), true value be (150,
150), shown in secondary lobe reduced size image such as Fig. 6 (a).
3) rectangular matrix that size is 300 × 300 is obtained in cutpb image to identify schlieren bead region, table
It is shown as Pcir, 0, i.e. Pcir (i, j)=0 are set by (i, j) gray value for being located at pixel in schlieren bead region, by (i, j)
Gray value positioned at schlieren bead region exterior pixel is set as 1, i.e. Pcir (i, j)=1.
4) the main lobe reduced size image cutzb image having a size of 300 × 300 is obtained from orgzb image, central point is located at
(Ozx+l, Ozy+m), wherein the value range of l value is -50≤l < 50, and it is phase that the value range of m value, which is -50≤m < 50, l and m,
For the coordinate at 100 × 100 rectangular area centers.
If the white rectangle region of Fig. 7 (a) is real estate of the cutzb image on orgzb, the white rectangle of Fig. 7 (b)
Region is real estate of the cutpb image on orgpb, and size dimension is all 300 × 300.It is deducted in new cutzb image
One border circular areas.When Pcir (i, j) is equal to 0, the pixel value in cutzb (i, j) is also configured as 0.
Cutzb and cutpb are respectively converted into column vector, are expressed as CutzAnd Cutp, calculated using formula (1)
CutzAnd CutpSpectrum angle charting value SAM, change l and m value (- 50≤l < 50, -50≤m < 50), when SAM minimum,
Optimal match point relative coordinate on orgzb image is (Ozx+l, Ozy+m), and minimum SAM value is 0.5631.Therefore, in main lobe
The image surrounded in image by white rectangle is exactly best match image cutzb, relative coordinate m, l, as shown in Fig. 7 (b).Most
Whole main lobe reduced size image size is 300 × 300, is represented as cutzb', in cutzb' schlieren center of pellet be (cxzb,
cyzb)。
Cutpb image and all cutzb image SAM values centered on (Ozx+m, Ozy+m) are generated one 100
× 100 matrix, the 3-D image that 100 × 100 matrixes of all SAM values composition are generated.When SAM value is minimized, phase
Value to coordinate position m and l is respectively -16 and 1, as shown in Figure 8.
4 images merge
The step of image merges is as follows:
1) the schlieren bead region of cutpb image is filled, cutpb image is located at the data in schlieren bead region, that is, works as
The region of Pcir (i, j)=1 is replaced after amplifying K multiple with cutzb ' data, is named as cutpb '.
2) region cutpb in orgpb image is replaced using final data cutpb ', therefore new orgpb image
For final merging image, it is named as Imerge.
3) splicing boundary is merged using weighted mean method [13], as a result as shown in Fig. 9 (b).
Because schlieren bead edge be not it is very regular, it is true to splice radius and be increased 5 pixels relative to okr, be used for
Eliminate the splicing trace merged in image.Shown in merging image such as Fig. 9 (a) of direct splicing, after Fusion Edges [14]
Merge image such as Fig. 9 (b), shown in final three-dimensionalreconstruction image such as Fig. 9 (c).
The horizontal direction two dimensional gray curve of reconstructed image is shown as shown in Figure 10.
In schlieren restructuring procedure, since focal spot is distributed very irregular, side lobe image can not accurately be blocked by schlieren bead occur
The case where center, the marginal portion for allowing for side lobe image in this way are saturated, because CCD has saturation spills-over effects, if
In the case that center is saturated, the image data for being saturated overflow area is also inaccurate, so by the saturation of schlieren bead marginal portion
The main lobe image data of region and peripheral region corresponding position after light intensity amplification coefficient multiplied by being replaced, to protect as far as possible
Demonstrate,prove the authenticity of stitching image.
Effect of the invention is analyzed below.
A, the texture region with feature is analyzed
For the automatic reconstruction result of main lobe and side lobe image, divided first according to the textural characteristics of splicing edge
Analysis is the splicing ring of reconstructed image in Figure 11 (a), and in annulus obtained by main lobe image completion, annulus is outer to be filled out by side lobe image
Fill gained.As can be seen that the texture of 6 selected areas is very identical from Figure 11 (b), wherein 1 or so stitching error of region is small
In 1 pixel, less than 1 pixel of about 3 region 2 and stitching error, 5-6 texture trend in region is complied fully with, and illustrates that reconstruct is spelled
It connects precision and meets requirement of experiment.
For the accuracy of authentication image reconstruct splicing, in k=1, between the merging image after reconstruct and main lobe image
Related coefficient be 0.997, illustrate that the accuracy rate of reconstructed image is up to 99%, it is possible thereby to determine, which is true
It is real reliable.Mean value between reconstructed image and main lobe image is respectively 34.62,35.73, and variance is respectively 65.81,65.10,
Illustrate that there is great similitude between two images.
B, schlieren center of pellet precision is improved using circle fitting algorithm
During table 2 is 5 Targeting, when calculating the schlieren bead center of circle using circle fitting process and gravity model appoach respectively, two
Application condition between kind algorithm and the practical calibration center of circle, wherein the true center of circle and radius are calibrated by the method for artificial interpretation
's.From Table 2, it can be seen that calculating the error in the center of circle and the true center of circle by circle fitting process is (0.3,0.36), and gravity model appoach
Error with the true center of circle is that (1.7,1.5) are calculated it can thus be seen that circle fitting algorithm is more excellent than gravity model appoach algorithm
Side lobe image center it is more acurrate, more can guarantee the authenticity of reconstructed image.As long as and this method carries out opposite side when circle fitting
Boundary's dot cycle can once calculate the center of circle and radius, and time complexity is O (n), and it is whole that entire algorithm time-consuming, which is 0.8 second,
A restructuring procedure saves the time.
Application condition (unit: pixel) between the circle fitting process of table 2 and gravity model appoach and the true center of circle
Integrated diagnostic system obtains that there are certain deficiencies for schlieren center of pellet, i.e., is hidden every time using schlieren bead
It is all that schlieren bead is allowed to be in a fixed position when keeping off secondary lobe spot center, such schlieren bead is all unable to entirely accurate and blocks
Schlieren center of pellet, this just needs to design an automatic-aligning system and schlieren bead is made accurately to block the center of secondary lobe hot spot,
The positioning accuracy of schlieren bead is improved, blocks intact side lobe image to obtain.
C, corresponding points of the analysis schlieren center of pellet on main lobe image
It is to guarantee the deciding factor of schlieren reconstruction accuracy that schlieren center of pellet, which is calculated, in the corresponding points of main lobe image, usually
The method for calculating corresponding points has gravity model appoach [11], geometrical center method [12], standardization [5], Spectral angle mapper method.Four kinds of algorithms
Reconstruct accidentally parameter it is as shown in table 3, in this Targeting data the theoretical value of corresponding points be (249.38 272.68), four kinds
Method corresponding points obtained and theoretical value error, wherein gravity model appoach, geometrical center method error are maximum, and vertical direction is greater than 10
Pixel；Standardization is smaller, but is greater than 1 pixel, and autocorrelation matching method is minimum, less than 1 pixel.
For the accuracy of authentication image reconstruct splicing, in k=1, schlieren bead first is obtained using four kinds of methods respectively
Then image reconstruction is completed in the corresponding points of main lobe image in center, be finally calculated between reconstructed image and main lobe image
Related coefficient, respectively 0.629,0.615,0.806,0.994, wherein autocorrelation matching method related coefficient reaches 0.994, explanation
The accuracy rate of reconstructed image is up to 99%, it is possible thereby to determine, the automatic restructing algorithm based on autocorrelation matching is really may be used
It leans on.Mean value between reconstructed image and main lobe image is respectively 34.62,35.74, and variance is respectively 65.81,65.10, explanation
There is great similitude between two images.
The reconstructed error of 3 four kinds of methods of table compares
In previous Targeting, when finding the corresponding optimal match point of secondary lobe center of pellet in main lobe image, one
As use standardization [5] and gravity model appoach, both methods has apparent defect.It is past when due to practicing shooting for standardization
Toward having very big vibration, there are atmospheric perturbations to use so the experiment condition of Targeting optical path and calibration optical path is entirely different
The splice point obtained in calibration optical path replaces the splice point actually practiced shooting, and there is a certain error.And for gravity model appoach, it is main
The distribution of valve hot spot is more uneven, then the error of center of gravity and best splice point is bigger.And being spliced based on optimal match point for this paper is calculated
Method has the advantage that 1) being capable of Automatic-searching stitching position.2) reduce stitching error.3) boundary blending algorithm eliminates spelling
Connect the splicing trace of position.
D, Range Analysis
The dynamic range [14] that stitching algorithm can measure is not only related with the dynamic range of two CCD itself, and and it
Relative energy decay coefficient K (K > 1) it is related, K is smaller, and dynamic range is lower, boundary fusion part it is more smooth, such as Figure 12
(a) shown in；K is bigger, and dynamic range is also bigger, and boundary fusion part will appear tomography, and the fractional error is bigger than normal, and stitching image is got over
Unsmooth, as shown in Figure 12 (b), with the increase of K, two CCD linear superposition regions start to become smaller, the dynamic range after reconstruct
It is bigger, but linear superposition region is mobile to the linear lower limit of main lobe CCD, therefore the error of boundary fusion part becomes larger, spliced map
As more rough.It can thus be appreciated that, as K=1, two linear regions CCD are completely overlapped in the case where b=1, do not having
The image that main secondary lobe obtains in the case where baffle is consistent, and the Noise Background distribution in each region of stitching image is completely the same, this
When the image that splices it is most smooth；As K > 1, with the increase of K, two CCD linear superposition regions start to become smaller, the dynamic after reconstruct
Range is also bigger, but linear superposition region is mobile to the linear lower limit of main lobe CCD, therefore the error of boundary fusion part becomes larger,
Stitching image is more rough, and image is possible to occur significantly splicing trace.
When being measured using schlieren method to far-field spot, the error of restructing algorithm is mainly by main lobe and the opposite spelling of secondary lobe
Connect the influence of position.In order to improve image mosaic precision, it is identical with secondary lobe optical path horizontal magnification factor and decay to devise main lobe
The different optical system of coefficient.This requirement is acquired using the CCD of same model, the focal spot horizontal magnification of main lobe and secondary lobe
Coefficient is identical, and main lobe and secondary lobe light intensity amplification coefficient are then different, i.e. b=1 and K > 1.In this way, main lobe hot spot and secondary lobe light
It is corresponded between each pixel of spot, specific scaling method bibliography [2].
It, need to be to the calibration value of related coefficient in order to reduce the influence of the intensity, beam quality of Calibrating source to measurement result
Carry out statistical average.The deviation of any corresponding points in two CCD detection faces can be controlled 1 using the method for on-line proving γ
In a pixel.In addition, the accurate of schlieren center of pellet calculates the precision that can also be improved reconstructed image, because only that determining secondary lobe
The corresponding relationship of each pixel between schlieren center of pellet and radius and main lobe and secondary lobe hot spot, just can determine that in secondary lobe hot spot
Middle schlieren bead region needs the main lobe hot spot data replaced, and calculates schlieren center of pellet coordinate by circle fitting algorithm and makes partially
Less than 1 pixel of difference.
The dynamic range of reconstructed image is the ratio between the maximum value of reconstructed image and side lobe image minimum signal, wherein
The minimum signal of side lobe image is defined as 90DL (1DL=6.7 μm of target position, corresponding 600 μm of target spots), is between schlieren bead
The average value of 1.5 times of radius and 1.5 extraordinarily 5 pixel time gray scales, as shown in figure 13.The dynamic range of three groups of experimental datas is surveyed
Test result is as shown in table 4:
4 dynamic range test result of table compares
As can be seen from the table, 4 experiments all realize the measurement of high dynamic range far-field focus, and surveyed dynamic range is all
Greater than 1000:1, but isotopic number CCD light intensity attenuation coefficient k difference is not huge, and the light intensity attenuation coefficient of experiment 1 and experiment 2 is
10 times for testing 3 light intensity attenuation coefficients, this is because using schlieren method to the measurement of far-field spot mainly by scientific CCD dynamic
The limitation of range.When being measured using 12 science CCD, maximum gradation value 4095, dynamic range is only 100:1.It wants
The far-field spot that maximum gray scale is 100,000 gray levels or hundreds of thousands gray level is measured, the decaying of main lobe is up to 25~250
Times.In splicing, main lobe light intensity will amplify 25~250, and such reconstructed image will be distorted serious.Experimental study shows main lobe light
Between strong 1~10 times of amplification, distortion is minimum, and splicing effect is best.And when being measured using 16 science CCD, maximum ash
Degree is 65535, to measure the far-field spot that maximum gray scale is 100,000 gray levels or hundreds of thousands gray level, the decaying of main lobe will
Reach only 1~10 times, the distortion very little of reconstructed image, splicing effect is best.But 16 scientific CCD are from foreign countries
Import, it is expensive, and also Targeting energy is very big, if decaying is improper, 16 scientific CCD will be destroyed, experiment
Cost is too big.
The present invention for schlieren theoretical foundation, reduces figure by calculating main lobe with schlieren measure far-field focus mathematical model [2]
The spectrum of image and secondary lobe reduced size image after deducting the circle of one and schlieren bead size as in is sought compared with charting (SAM) value [7]
The smallest point of spectrum charting value looked between two images is optimal match point, is realized to oneself of main lobe image and side lobe image
Dynamic splicing, then splicing boundary is merged using weighted mean method.In addition, being searched herein using based on spectrum angle charting algorithm
Optimal match point between main lobe image and side lobe image is sought, the search precision of optimal match point is improved.
By the way that experimental result, this method can be realized the automatic weight of focal spot in the laser focal spot measurement of high dynamic range
Structure, reconstructed image have the splendid goodness of fit, less than 1 pixel of stitching error on splicing regions texture, and measurement dynamic range connects
Nearly 4 quantity is not only able to obtain complete far-field focus image, and meets Targeting and conventional efficient is wanted
It asks, obtaining comprehensive, accurate laser parameter for integrated diagnostic system has highly important directive significance.
Bibliography:
The overall assembly of the Shenguang-Ⅲ laser aid such as [1] Liu Weibao, Zheng Wanguo, Zhu Qihua and exploration [J] national defence section
Skill, 2014,34 (12): 30-36.
[2] Wang Zhengzhou, Wang Wei, Xia Yanwen high dynamic range laser focal spot measure Study on Mathematic Model [J] photonics journal,
2014,43 (10), 1010002-1~10100002-7.
Investigation on focal spot measurement by schlieren method experimental study [J] the light laser such as [3] Cheng Juan, Qin Xingwu, Chen Bo and the particle beams,
2006, (18), 4,612-614.
[4]Haynam CA,Wegner PJ,Auerbach JM,Bowers MW,Dixit SN,Erbert GV,et
al.National Ignition Facility laser performance status.Applied Optics.2007,46
(16):3276-3303.
[5]Du XW.Factor for evaluating beam guality of a real high power
laser on the target surface in far field.Chinese journal of lasers.1997,24:
327-332.
[6] Zhang Zheng, Wang Yan equality Digital Image Processing and machine vision -- Visual C++ and Matlab realize the north [M]
Capital: People's Telecon Publishing House .2013:428-433.
[7] Zhang Bing, Gao Lianru classification hyperspectral imagery and the Beijing target acquisition [M]: Science Press, 2011,179-
185.
[8] Beijing Zhao Chunjiang Digital Image Processing algorithm Classical examples [M]: People's Telecon Publishing House, 2009.
[9] Algorithm of laser spot detection [J] of Kong Bing, Wang Zhao, Tan Yushan based on circle fitting is infrared with laser work
Journey, 2001,31 (3): 275-279.
[10] image analysis of Zhang Yujin Image Engineering (the middle volume second edition) Beijing [M]: publishing house, Tsinghua University, 2005:
222-223.
[11]Williams WH,Auerbach JM,Henesian MA,Jancaitis KS,Manes KR,Mehta
NC,et al.Optical propagation modeling for the national ignition facility.Proc
of SPIE.2004,5341:66-72.
[12]Ma Xiaoyu,Rao Changhui,Zheng Hanqing.Error analysis of CCD-based
point source centroid computation under the background light[J].Optics
Express.2009,17(10):8525-8541.
[13] the Shao Xiangxin digital picture splicing core algorithm research Jilin [D]: Jilin University, 2010.
[14]Yang PL,Feng GB,Wang ZB,Wang QS,Feng G,Zhang TQ.Detector Array
for Measuring Far-Field Power Density Distribution of Mid-Infrared
Laser.Chinese J.Lasers.2010,37(2):521-525.
[15] Zhang Zheng, Wang Yan equality Digital Image Processing and machine vision -- Visual C++ and Matlab realize [M]
Beijing: People's Telecon Publishing House .2013:428-433.
Claims (2)
1. the schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting, which is characterized in that including following step
It is rapid:
Step 1) is fitted the schlieren center of pellet of other spot image, less than 1 pixel of error precision using least square method；
The main spot image orgzb of acquisition and side spot image orgpb are cut to an equal amount of cutting image by step 2), are remembered respectively
For main spot reduced size image cutzb and other spot reduced size image cutpb；It is one small with schlieren by being deducted in main spot reduced size image cutzb
The comparable circle of ball size, obtains image cutzb '；Image cutzb ' and side spot reduced size image cutpb are converted into two respectively again
A column vector VzAnd Vp, and calculate the SAM value of the two column vectors；
Step 3) calculates the charting of spectrum angle (SAM) value between two column vectors, when two vector similitudes reach maximum value
When, then the upper left position of cutzb ' image is optimal match point；
Step 4) is reference with optimal match point, is reconstructed using image cutzb' and other spot reduced size image cutpb focal spot,
And splicing boundary is merged using using weighted mean method during final image mosaic.
2. the schlieren method far-field measurement focal spot automatic reconfiguration method according to claim 1 based on spectrum angle charting,
It is characterized in that, detailed process is as follows with step 3) for step 2):
1.1) schlieren center of pellet and radius in other spot image orgpb are obtained using least square method, be expressed as (Opx,
) and okr Opy；The other spot reduced size image having a size of 300*300 is reduced centered on (Opx, Opy), is denoted as cutpb；
1.2) schlieren bead mapping matrix P is obtainedc:
It is column vector form P by the matrix conversionc=Pc(:), wherein (Cpy, Cpx) and r are respectively schlieren center of pellet and half
Diameter；
1.3) main spot reduced size image cutzb=orgzb (Ozy-150+m:Ozy+150+m, Ozx-150+l:Ozx+150-1+l),
Central point is located at (Ozx+l, Ozy+m), and wherein the value range of l value is -50≤l < 50, the value range of m value be -50≤m <
50, l and m is the coordinate relative to 100 × 100 rectangular area centers；
1.4) the transposed matrix V' of cutzb image array is obtainedz=cutzbT, transposed matrix is expressed as column vector form V'z=
V'z(:), the image after a schlieren bead region is deducted in cutzb image is cutzb ', and is expressed as column vector form Vz
=V'z.*Pc T；Obtain the transposed matrix V' of cutpb image arrayp=cutpbT, transposed matrix is expressed as column vector form Vp
=V'p(:)；
1.5) vector V is calculatedpAnd VzSAM value, i.e. Sam (Vz,Vp):
In formula: Vp--- secondary lobe reduced size image column vector generated, Vz--- deduct a same schlieren in main lobe reduced size image center
After the circle of bead size and the column vector that generates, the element that N --- the wide and high product of reduced size image and vector include are a
Number；
1.6) 1.3) -1.4 are repeated) step, until the SAM value of 100 × 100 all m and l of rectangular area has been calculated；
1.7) pass through objective function Sam (Vz,Vp) obtain optimal match point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811528861.1A CN109724690A (en) | 2018-12-13 | 2018-12-13 | Schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811528861.1A CN109724690A (en) | 2018-12-13 | 2018-12-13 | Schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109724690A true CN109724690A (en) | 2019-05-07 |
Family
ID=66295918
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811528861.1A Pending CN109724690A (en) | 2018-12-13 | 2018-12-13 | Schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109724690A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112001956A (en) * | 2020-07-30 | 2020-11-27 | 中国科学院西安光学精密机械研究所 | CNN-based schlieren strong laser far-field focal spot measurement image denoising method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201897502U (en) * | 2010-12-01 | 2011-07-13 | 中国科学院西安光学精密机械研究所 | Schlieren device for testing far-field focus faculae of high-power laser |
| CN105807441A (en) * | 2016-05-27 | 2016-07-27 | 武汉理工大学 | Transmission constant combustion bomb high-speed photographing schlieren system capable of correcting chromatic aberration |
-
2018
- 2018-12-13 CN CN201811528861.1A patent/CN109724690A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201897502U (en) * | 2010-12-01 | 2011-07-13 | 中国科学院西安光学精密机械研究所 | Schlieren device for testing far-field focus faculae of high-power laser |
| CN105807441A (en) * | 2016-05-27 | 2016-07-27 | 武汉理工大学 | Transmission constant combustion bomb high-speed photographing schlieren system capable of correcting chromatic aberration |
Non-Patent Citations (2)
| Title |
|---|
| ZHENGZHOU WANG 等: "A fast method for searching the best matching point of the automatic reconstruction of focal spot measurement using the steepest descent method", 《PROCEEDINGS OF SPIE》 * |
| ZHENGZHOU WANG 等: "An improved schlieren method for measurement and automatic reconstruction of the far-field focal spot", 《PLOS ONE》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112001956A (en) * | 2020-07-30 | 2020-11-27 | 中国科学院西安光学精密机械研究所 | CNN-based schlieren strong laser far-field focal spot measurement image denoising method |
| CN112001956B (en) * | 2020-07-30 | 2024-04-09 | 中国科学院西安光学精密机械研究所 | CNN-based image denoising method for strong laser far-field focal spot measurement by using schlieren method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103592108A (en) | CCD chip modulation transfer function test device and method | |
| CN103503022B (en) | The method and apparatus that the view data of the image in the region for building target object is provided | |
| CN101292255A (en) | Artifact mitigation in three-dimensional imaging | |
| CN105675455B (en) | A kind of method and device reducing stochastic system noise in Particle Size Analyzer | |
| CN106092158B (en) | Physical parameter method of estimation, device and electronic equipment | |
| CN102692394B (en) | Two-dimensional imaging method and device based on thermal lens effect | |
| Lindroos et al. | Stacking of large interferometric data sets in the image-and uv-domain–a comparative study | |
| Stovall et al. | TLSLeAF: automatic leaf angle estimates from single‐scan terrestrial laser scanning | |
| CN112597428B (en) | Flutter detection correction method based on beam adjustment and image resampling of RFM model | |
| JP5595463B2 (en) | Wavefront optical measuring device | |
| CN111561877B (en) | Variable resolution phase unwrapping method based on point diffraction interferometer | |
| Liu et al. | Analysis and correction of distortions in a spatial heterodyne spectrometer system | |
| CN109724690A (en) | Schlieren method far-field measurement focal spot automatic reconfiguration method based on spectrum angle charting | |
| CN104280120B (en) | A kind of spectral bandwidth measuring method and device | |
| JP4340625B2 (en) | Optical inspection method and apparatus | |
| JP6044052B2 (en) | Fisheye image data creation program and LAI calculation program | |
| CN113298700A (en) | High-resolution image reconstruction method in scattering scene | |
| Tanaka et al. | Measurement of forest canopy structure with a laser plane range-finding method–development of a measurement system and applications to real forests | |
| Perraut et al. | Benchmarking the fundamental parameters of Ap stars with optical long-baseline interferometric measurements | |
| CN110308117A (en) | Diffraction range calibration method and system and image rebuilding method and system | |
| Campi et al. | Photogrammetric processes and augmented reality applications using mobile devices | |
| CN211669359U (en) | Long-focus laser three-dimensional imager distortion testing device based on angle measurement method | |
| CN108007387A (en) | Surface shape measurement apparatus and method based on Structured Illumination | |
| CN110806572B (en) | Device and method for testing distortion of long-focus laser three-dimensional imager based on angle measurement method | |
| CN109785426A (en) | The method that optimal match point fast searching based on steepest descent method realizes focal spot reconstruct |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190507 |
|
| WD01 | Invention patent application deemed withdrawn after publication |