CN104501781A - Lamination imaging technology based on known prior information restriction - Google Patents

Abstract

The invention discloses a lamination imaging technology based on partial known prior information of sample to be detected. According to the invention, known information samples can be placed at surrounding of the sample to be detected, scanning is carried out in a measure plane by a lamination scanning mode in order, the corresponding series intensity images can be recorded by an image sensor; the recorded series intensity images are processed, a lamination scan iteration algorithm based on the known surrounding prior information is used, so that a complex amplitude image of the sample to be detected can be obtained in the computer. Through the surrounding unknown prior information restriction, the recovered image is more accurate, the convergence speed is faster, and the image has good translation and noise-resistance capability, problem of overhigh specification requirement on a translation bench in traditional lamination imaging can be effectively solved, and cost is reduced greatly. The lamination imaging technology has the advantages of high imaging efficiency and good transplantability, and is suitable for imaging of various reflective or transmission objects.

Description

Based on the lamination imaging technique that known prior imformation retrains

Technical field

The present invention relates to imaging technique, be specifically related to a kind ofly to place the object of some known prior imformations around testing sample, and the lamination imaging that known substance is scanned together with sample.

Background technology

By the basic implementation of Ptychography principle and the implication of Greece's root " Ptych ", Ptychography is translated into lamination imaging technique by the present invention temporarily.Lamination imaging technique is a kind of lensless scanning coherent diffraction imaging system, in order to obtain the complex amplitude information of testing sample, the small sircle hole (or sample itself) of a mobile full impregnated makes incident plane wave be irradiated to the different parts of sample, and utilizes a series of diffracted intensity patterns obtained thus to be reconstructed amplitude and the phase information of sample by computing machine lamination iterative algorithm.See (Ultramicroscopy 10 (3): 187 ~ 198,1987).Lamination iterative algorithm belongs to a kind of Phase Retrieve Algorithm in essence, but its Phase Retrieve Algorithm again from traditional is different, retrains when carrying out phase recovery to the Diffraction fringe distribution of each position, solves the ambiguity problem of rebuilding image.Meanwhile, the basic realization of lamination Imaging is the relative movement of illuminator probe and small sircle hole and sample, and belong to without lens imaging, thus the complexity of system significantly reduces.

But, be mainly limited to three factors when carrying out imaging with existing lamination imaging technique: (1) scan aperture cannot reach accurate aligning, inevitably has certain translation error, thus affects experimental result; (2) to carry the experimental result that noise also can make to recover undesirable for the noise such as extraneous light and system; (3) when scan aperture exists translation error, speed of convergence improves space.See (Acta Crystallogr:A 25,495,1969).Current lamination imaging mainly adopts the motorized precision translation stage of high standard to reduce translation error, but the program makes experimental cost significantly increase, and still there is certain system noise simultaneously, affects the quality of imaging.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, solving translation error and system noise in existing lamination imaging technique affects the problem of experimental result, makes speed of convergence speed further simultaneously.

Object of the present invention realizes by following technical measures:

First use laser illumination testing sample, and it is carried out translation by the mode of lamination scanning in object plane, record the oscillator intensity image corresponding to each scanning position successively with imageing sensor; On computers the lamination iterative algorithm process based on known prior imformation is carried out to the oscillator intensity image of record subsequently, reappear the complex amplitude image of each layer of testing sample.

The scanning of described lamination is so a kind of scan mode, the testing sample distance that translation is certain successively in object plane, need to ensure between adjacent flat pan position on testing sample by the region memory of beam lighting certain overlapping during translation.

The present invention's imaging algorithm used is the lamination scan iterations reconstruction algorithm based on known prior imformation, and its specific implementation process is:

Algorithm:

(1) doing initial guess to sample is on computers O _in(x, y).(time initial, i=1 is iterations, and n=1 is first probe).

(2) object is by illuminator probe P _nbe transferred on CCD face after (x, y) illumination:

Ψ _in(ξ, η)=FrT _{λ, d}[P _n(x, y) O _in(x, y)] (1) (3) are by the light distribution photographed by CCD be brought in Iterative restoration process as constraint condition:

${{\mathrm{\ψ}}_{\mathrm{in}}}^{\′}(\mathrm{\ξ},\mathrm{\η})=\sqrt{I}\·\left(\right|{\mathrm{\ψ}}_{\mathrm{in}}(\mathrm{\ξ},\mathrm{\η})|/|{\mathrm{\ψ}}_{\mathrm{in}}(\mathrm{\ξ},\mathrm{\η})\left|\right)---\left(2\right)$

(4) 2 formulas are returned object plane against diffraction

$O_{\mathrm{in}}(x,y)=\mathrm{Fr}{T}_{\mathrm{\λ},d}^{-1}\left[{{\mathrm{\ψ}}_{\mathrm{in}}}^{\′}(\mathrm{\ξ},\mathrm{\η})\right]---\left(3\right)$

Wherein FrT ¹for the operational symbol of inverse Fresnel transform.

(5) then upgrade initial sample conjecture O (x, y), more new formula is:

O _in(x，y)＝O _in(x，y)+U[O _in(x，y)-p _n(x，y)·O _in(xy)] (4)

Wherein U is for upgrading coefficient:

$U=\frac{\left|P_n(x,y)\right|}{\left|P_{n,\max }(x,y)\right|}\·\frac{P_n^{*}(x,y)}{{\left|P_n(x,y)\right|}^2+\mathrm{\δ}}---\left(5\right)$

Wherein P _{n, max}(x, y) represents P _nthe maximal value of (x, y) modulus value, δ is noise compressibility factor, and its value is the constant close to 0, is usually set to 0.01.

(6), after completing this renewal rewards theory, next illuminator probe P is moved to _n(x, y) (n=n+1), recurring formula (2)-(5) are a lighting position to the last.

(7) continue to carry out above iterative process, i.e. k=k+1, until the value of k reaches the threshold value of setting, k=100 can export reconstructed results preferably under normal circumstances.

The present invention compared with prior art has following advantage:

(1) utilize imaging technique of the present invention can recover experimental image when there is larger translation error scan aperture, improve experimental system opposing translation error ability.

(2) retrain because the present invention places prior imformation around testing sample, experimental system can be made when there being outside noise to recover good testing sample image, there is stronger noise resisting ability.

(3) because the present invention is by the constraint of prior imformation, the speed of convergence of testing is speeded further.

(4) because the lamination formation method that the present invention is used has very strong anti-translation error and noise resisting ability, the range of choice of the motorized precision translation stage used in experiment can be expanded, greatly reduce experimental cost.

Lamination formation method based on known prior imformation disclosed in this invention, is applicable to the imaging of reflection-type object or transmission-type object.

Accompanying drawing explanation

Fig. 1 a is that imaging technique of the present invention is imaged as the light channel structure figure of embodiment at the lamination of the known prior imformation constraint of throwing light on transmission-type Single wavelength.

Fig. 1 b is that imaging technique of the present invention is imaged as the light channel structure figure of embodiment at the lamination of the known prior imformation constraint of throwing light on reflective Single wavelength.

Fig. 2 is objective table in imaging technique of the present invention, and prior imformation and testing sample place schematic diagram.

Fig. 3 a is the amplitude image picture of the testing sample that in embodiment, Numerical Experiment is used, and placing prior imformation around testing sample is grating, and grating constant is 12 μm;

Fig. 3 b is the phase image of the testing sample that in embodiment, Numerical Experiment is used, and placing prior imformation around testing sample is grating, and grating constant is 12 μm;

Fig. 4 a is the lamination scanning schematic diagram of Numerical Experiment in embodiment;

Fig. 4 b is the schematic diagram of Numerical Experiment scanning position central point position in embodiment;

Fig. 5 a be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, translation error is the amplitude image picture with the testing sample of traditional lamination imaging algorithm reconstruction when offseting 120 μm to the right;

Fig. 5 b be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, translation error is the amplitude convergence curve rebuilding testing sample image when offseting 120 μm to the right with traditional lamination imaging algorithm;

Fig. 5 c be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, translation error is the phase image with the testing sample of traditional lamination imaging algorithm reconstruction when offseting 120 μm to the right;

Fig. 5 d be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, translation error is the phase place convergence curve rebuilding testing sample image when offseting 120 μm to the right with traditional lamination imaging algorithm;

Fig. 6 a be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, the amplitude image picture that translation error is the testing sample rebuild based on the lamination imaging algorithm of known prior imformation that adopts the present invention to propose when offseting 120 μm to the right;

Fig. 6 b be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, translation error rebuilds by the lamination imaging algorithm based on known prior imformation that adopts the present invention to propose when offseting 120 μm to the right the amplitude convergence curve of testing sample image;

Fig. 6 c be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, the phase image that translation error is the testing sample rebuild based on the lamination imaging algorithm of known prior imformation that adopts the present invention to propose when offseting 120 μm to the right;

Fig. 6 d be in embodiment in Numerical Experiment as scan aperture P ₁₁, P ₁₄, P ₂₄, P ₃₄there is translation error, translation error rebuilds by the lamination imaging algorithm based on known prior imformation that adopts the present invention to propose when offseting 120 μm to the right the phase place convergence curve of testing sample image;

Wherein, 1. laser instrument, 2. collimating and beam expanding system, 3. two-dimension translational platform, 4. CCD image sensor, 5. computing machine, 6. beam splitter, 7. catoptron, 8. objective table, 9. prior imformation object, 10. testing sample.

Embodiment

Below in conjunction with accompanying drawing and embodiment, the present invention will be further described.

Fig. 1 a and Fig. 1 b is formation method of the present invention at the light channel structure figure of two kinds of exemplary embodiment of transmission-type and the reflective lamination imaging based on known prior imformation respectively.This structure comprises laser instrument 1, collimating and beam expanding system 2, two-dimension translational platform and testing sample 3, CCD image sensor 4, computing machine 5, beam splitter 6, catoptron 7.Translate stage 3 is d to the distance of imageing sensor 4.Translate stage 3 and imageing sensor 4 have controlled phase-shift phase interpolation by computing machine 5 respectively, lamination scans and image record.First the object of known amplitude phase information will be placed around testing sample, amplitude grating is adopted in this example, grating constant is 12 μm, open laser instrument 1 subsequently, the position of scan aperture is adjusted by motorized precision translation stage 2, and with the strength information of imageing sensor and computer recording scan image, the complex amplitude information finally extracting testing sample from recorded intensity image is completed by the computer program designed according to the inventive method.

Fig. 2 is that in the present invention, schematic diagram put by sample, and this structure comprises objective table 8, prior imformation object 9, testing sample 10.Wherein have in this example of object of prior imformation and adopt grating, by shown in Fig. 2, grating and testing sample are placed successively in experiment.

Fig. 3 a Fig. 3 b is the initial setup data used during the Computer Numerical Simulation of carrying out for above-described embodiment is tested.Fig. 2 a and Fig. 2 b is amplitude and the phase information of testing sample used in experiment, and they are 256 × 256 pixels, amplitude by naturalization to [0,1], position by naturalization to [0,2 π].

Fig. 4 a is lamination scanning schematic diagram, and the circular port of illuminator probe to be radius be 70 pixels, the overlapping ratio in adjacent illumination region is 0.64.Fig. 4 b is the lamination scanning position figure used in experiment, and have 4 × 4 scanning positions in figure, the central point of each scanning position uses "+" to mark, and uses symbol P _mn(m, n are respectively the row and column ordinal number of scan matrix) carries out marking to show difference.In experiment, laser wavelength lambda is 632.8nm, and the spacing d of imageing sensor and translate stage is 30mm, and the pixel size of imageing sensor is 6.45 μm.

The amplitude image picture of the testing sample that Fig. 5 a rebuilds for use tradition lamination imaging algorithm computer program, iterations k=200.Fig. 5 b is its iteration convergence curve.

The phase image of the testing sample that Fig. 5 c rebuilds for use tradition lamination imaging algorithm computer program, iterations k=200.Fig. 5 d is its iteration convergence curve.

Fig. 6 a is for using the amplitude image picture of the testing sample rebuild based on the lamination imaging algorithm computer program of known prior imformation designed by the present invention, iterations k=200.Fig. 6 b is its iteration convergence curve.

Fig. 6 c is for using the amplitude image picture of the testing sample rebuild based on the lamination imaging algorithm computer program of known prior imformation designed by the present invention, iterations k=200.Fig. 6 d is its iteration convergence curve.

From the contrast of Fig. 5 a and Fig. 6 a and Fig. 5 c and Fig. 6 c, we can visually see, and use the formation method under the known prior imformation condition designed by the present invention to be significantly increased than traditional lamination imaging results.

As can be seen from the contrast of Fig. 5 b and Fig. 6 b, we, rebuild under using known prior imformation condition designed by the present invention to restrain when iteration is less than 20 times in testing sample amplitude information process and just more level off to 0.95; And traditional lamination imaging not only speed of convergence is comparatively slow but also can only restrain and level off to 0.9.

As can be seen from the contrast of Fig. 5 d and Fig. 6 d, we, rebuild under using known prior imformation condition designed by the present invention to restrain when iteration is less than 20 times in testing sample phase information process and just more level off to 1; And traditional lamination imaging not only speed of convergence more slowly and the highest can only convergence levels off to 0.8, the increase along with iterations becomes and does not restrain gradually.

Said method and embodiment are all the amplitude image pictures by recording testing sample different parts in known prior imformation situation by the mode of lamination scanning, and carry out lamination iterative approximation for the purpose of the complex amplitude information recovering more high-quality testing sample to amplitude image picture.Enforcement of the present invention is not limited to above-mentioned specific embodiments.As long as carry out formation method, device and system by the scanning of the lamination of placing known prior imformation around testing sample and lamination iterative approximation to object, all belong to protection scope of the present invention.

Claims (6)

1. the lamination imaging technique in known testing sample peripheral part prior imformation situation, its imaging process comprises the following steps:

The first step, places the sample of Given information around testing sample;

Second step, carries out translation in the mode of lamination scanning by testing sample, and records the intensity image corresponding to each scanning position successively with imageing sensor in object plane;

3rd step, uses the lamination scan iterations algorithm based on known surrounding prior imformation to rebuild the complex amplitude image of testing sample.

2. as in claim 1 the lamination scan iterations reconstruction algorithm of known surrounding prior imformation that uses, it is characterized in that, iterative approximation is carried out to testing sample in the basis of the image information around known testing sample, by the constraint of Given information, make anti-translation error ability in scanning process stronger, also have better noise resisting ability simultaneously.

3. as in claim 1 the lamination scan iterations reconstruction algorithm of known surrounding prior imformation that uses, it is characterized in that, by the constraint of Given information, make the speed of convergence recovering sample information image faster.

4. the lamination scanning as described in claim 1,2 and 3, it is characterized in that, the testing sample distance that translation is certain successively in object plane two-dimensional coordinate completes scanning, and to ensure between adjacent flat pan position on testing sample by the region memory of beam lighting certain overlapping.

5. the lamination scanning as described in claim 1,2 and 3, is characterized in that, can be optimized design according to the relative size of actual illumination light beam and testing sample.

6. the lamination imaging technique around known testing sample as described in claim 1 in a part of prior imformation situation, is applicable to the imaging of various reflection-type or transmission-type object.