CN101261176B - Sequence image correction based pneumatic optical transmission effect evaluation method - Google Patents

Abstract

The invention discloses a pneumatic optical transmission effect evaluation method and a device which are based on sequence image correction. The pneumatic optical transmission effect evaluation method calculates a correction image and a point spread function by three correction methods of space blocking concurrent correction, two-dimensional sampling concurrent correction and concurrent iterativecorrection, and gains the evaluation indexes by further calculation; the evaluation indexes are used for measuring, analysing and evaluating the pneumatic optical transmission effect. Aiming at the pneumatic optical transmission effect evaluation method, a calculation processing device based on a host computer, a multifunctional image collecting control card and a concurrent digital signal assistprocessing card is designed. The collecting, processing, memorizing and displaying of different passage image source data can be achieved by the calculation processing device; by virtue of the characteristics of high parallelism of the evaluation device calculation and the extensibility of the processing unit, the quantity of the calculation processing units is smartly adopted so as to achieve the real-time processing; the evaluation device of the invention has the advantages of small volume, light weight, and easy carry.

Description

Pneumatic optic transmission effect Measurement and analysis method based on sequence image correction

Technical field

The invention belongs to the interdisciplinary science technical field that Pneumatic optical, Flame Image Process combine with measuring technique, be specifically related to a kind of pneumatic optic transmission effect Measurement and analysis method and apparatus based on sequence image correction.

Background technology

Pneumatic optical is that a subject to the influence and the correction thereof of high-speed aircraft imaging detection is streamed in research at a high speed.When the high-speed aircraft that has an optical imagery detection system flies in the endoatmosphere, form complicated flow field between optics head-shield and the incoming flow, cause heat radiation and light transmission to disturb to the optical imagery detection system, cause scene image skew, shake, fuzzy, this effect is called aero-optical effect.When the optical imagery detection system that high-speed aircraft carries is flown in the endoatmosphere, the marked change that violent interaction will cause ambient atmosphere takes place between its optics head-shield and the incoming flow: the variation of the variation of the variation of composition, component density and density gradient, gas temperature and thermograde, even generation molecular ionization phenomenon etc., form complicated flow field, cause heat, heat radiation and image transmission to disturb to the optical imagery detection system, cause image shift, shake, aero-optical effect such as fuzzy.Aero-optical effect comprises: High Speed Flow Field pneumatic optic transmission effect, pneumatic thermoradiation efficiency and the pneumatic thermal effect of optics head-shield.

Pneumatic optic transmission effect is meant behind the pneumatic flow field of light wave process that scenery sends, because the random variation in pneumatic flow field makes transmitting beam produce deviation, inhomogeneous medium has phenomenons such as scattering, absorption to luminous energy, finally be reflected as image blurring, effects such as kine bias is moved, the decay of focus energy and picture shake in the aircraft imaging system.Pneumatic optic transmission effect generally with kine bias move, as fuzzy, weigh as parameters such as shake and energy attenuation degree.

Above-mentioned Pneumatic optical application problem is present in high speed real gas flow field, and the mutual relationship in light wave and flow field is very complicated.At present, these complicated aero-optical effect problems there is not mature theory, this means that all progress applicatory must depend on the ground wind tunnel experiment, on the ground the flight High Speed Flow Field is simulated, from the wind-tunnel confirmatory experiment is measured, gather complete reliable data.Obtaining the aero-optical effect sequential image data in wind tunnel test, is a kind of convenient and practical, economic, effective test method, for the kine bias that the research aero-optical effect causes move, as fuzzy, provide reliable test and appraisal means as the analyses and prediction of shake.

Yet, at present domestic and international open source literature only has the report of microscopic measurement means, these classic methods of measuring based on flow field parameter are directly measured from microcosmic angle stream field parameter, be difficult to set up the mathematical model of these flow field parameter, also be difficult to set up the relation of high-speed aircraft Mach number and pneumatic optic transmission effect and the relation of high-speed aircraft optical window and pneumatic optic transmission effect the target imaging quality influence.Therefore, can not estimate the influence of pneumatic optic transmission effect quantitatively to image quality.The quantitative measurment of Pneumatic optical transmission macroeffect (comprising point spread function PSF, optical-modulation transfer function MTF, Strehl ratio, jitter range, chattering frequency etc.) and evaluation also lack means and measure at present.

At present, be used for the method that Pneumatic optical measures in the open source literature and mainly contain following several form.

1. the schlieren method and the shadow method that show of flow field;

Can the show transparency subtle change of refractive index in the medium of schlieren and shadow method.The schlieren method is summed up as the first order derivative of measuring variations in refractive index; And shadow method is measured the second derivative of variations in refractive index.Its image recording that obtains is the integral process of space, time.

(seeing (R.W.Ladenburg) Princeton univ Press of document 1:Beams.J W.Shadow and Schlieren Methods.In " Physical Measurements inGas Dynamics and Combustion ", New Jersey, 1954)

2. the flow field shows the laser mating plate scattering method of measuring;

With the optical scattering method in laser mating plate cutting flow field, the plane that instantaneous demonstration is cut obtains flow field information.Laser mating plate scattering method has two kinds of plane Michaelis particulate scattering method and plane Rayleigh molecular scattering methods.(seeing document 2:Dyer T.M.Rayleigh Scattering Measurements of Time-Resolved Concentration in a TurbulentPropane Jet.AIAA J.17 (8), 912-914,1979)

3. the interference technique of flow field density distribution measuring;

The interference technique of flow field density distribution measuring is the measuring technique based on the optical interference principle.Interference fringe in the interferogram is the information carrier that interferometry obtains, and has reflected the optical path difference of the beam Propagation process in the different paths of two-way.Interference technique has double beam interferometry, point-diffraction interference method, holographic interferometry and shearing interference method.(see document 3:BachaloW D.Optical Interferometer in Fluid Dynamics Research.Optical Engineering, Vol.24, No.3, May 1985)

4. wavefont sensor method;

Wavefont sensor is a kind of optic probe on the corrugated (or phase place) that is used to measure incident beam, when the light with corrugated, plane passes through the flow field, to influence total light path and phase of light wave by flow field density or gas flow temperature variation, be out of shape the variation of analyzing the flow field from measuring the corrugated.Measure two parameters of light wave with wavefont sensor: irradiation profile and corrugated PHASE DISTRIBUTION.With the data computation point spread function that wavefont sensor is measured, be a kind of non-contact measurement method of flow field characteristic.Wavefont sensor can accurately be measured the corrugated for primary aberration, thereby the calculation level spread function is different with primary aberration for the aberration that the high transient flow field of Pneumatic optical produces, and causes very big scattering angle, and point spread function is descended greatly.(see document 4:Neal D.R., O ' Hern T.J., Torczyanski J.R., Warren M.E.and Shul R., " Wavefront Sensorsfor Optical Diagnostics in Fluid Mechanics:Applications to Heated Flows, Turbulence andDroplet Evaporation; " Optical Diagnostics in Fluid and Thermal Flows, SPIE, Vol.2005,1993)

Said method is mainly measured flow field parameter, is difficult to set up the mathematical model of these parameters to the influence of scenery image quality.The measurement computing method of disclosed point spread function utilize wavefont sensor to measure the aberration quantity that introduce in the flow field in the document at present.

At present, the computing system that is used for the authentication image bearing calibration in the open source literature mainly contains following several form.

1. high-performance standalone computer systems;

The high-performance standalone computer systems is that the image rectification algorithm simulating is realized the most frequently used platform, and advantage is that one-of-a-kind system is simple and easy to usefulness, and is with low cost, and shortcoming is that calculated performance is limited, finishes the complicated length consuming time of proofreading and correct, and the emulation cycle is long.

2. the microcomputer multi-node system of Network Environment;

The multi-computer system of Network Environment overcomes the limited shortcoming of one-of-a-kind system calculated performance, use many high-performance computers to form the The Realization of Simulation that multi-computer system is carried out correcting algorithm, as document 5: " Parallel Image Restoration withDomain Decomposition " mention in (" Real-time Imaging " magazine the 7th phase of calendar year 2001) by 9 sun stations (3 Ultra30,3 Sparc20 and 3 Sparc10) multi-computer system formed, workstation is by the 10Mbps ethernet interconnect; And Cray T3E/1200 system, used 128 DEC Alpha 21164 processors to connect into three-dimensional network.Multimachine is used for image rectification by the 10/100Mbps ethernet interconnect.Advantage is that calculated performance is improved, and shortcoming is that system complex is huge, and integrated level is not high, because Ethernet data communication bandwidth bottleneck, the multimachine parallel efficiency is not high, and can not receive the external camera data and directly import, and can not proofread and correct assessment in real time.

The aforementioned calculation method is applicable to that the off-line of High Speed Flow Field degraded image restores and correction, carries inconvenience, can not realize real-time correction, and inconvenience is carried on the experiment place.

Summary of the invention

The invention provides a kind of pneumatic optic transmission effect Measurement and analysis method based on the sequence image figure adjustment, be used for online and off-line aero-optical effect numeral test and appraisal test platform, both can be used for on-the-spot aero-optical effect wind tunnel test test and appraisal, also can be to the test and appraisal of aero-optical effect off-line and the laboratory aero-optical effect simulation study of high-speed aircraft practical flight process.

Pneumatic optic transmission effect Measurement and analysis method based on the sequence image figure adjustment provided by the invention, its step comprises:

(1) gather Pneumatic optical trial image g (x, y);

(2) (x y) utilizes the image rectification algorithm to estimate that unknown benchmark image f (x, y) (x, y), use respectively with point spread function h by the correcting image and the point spread function that estimate to the Pneumatic optical trial image g that collects With

Expression;

(3) utilize the correcting image of estimating , the point spread function estimated

(x y) calculates the Measurement and analysis index, and the Measurement and analysis index comprises a kind of or several arbitrarily than in, side-play amount and the jitter range of image blur, encircled energy, point spread function supporting domain, Si Telieer with the trial image g that collects;

(4) the Measurement and analysis index of utilizing step (three) to obtain is analyzed, and wherein, image blur is used for the order of severity of pneumatic optic transmission effect is analyzed; Encircled energy is used for pneumatic optic transmission effect is analyzed the influence of scenery target imaging; The supporting domain of point spread function is used for the order of severity of pneumatic optic transmission effect is analyzed; Si Telieer analyzes than the order of severity that is used for pneumatic optic transmission effect; The side-play amount of scenery target is used for pneumatic optic transmission effect is analyzed the influence of scenery target imaging; Jitter range is used for pneumatic optic transmission effect is analyzed the influence of scenery target sequence image imaging.

A kind of assessment device of pneumatic optic transmission effect based on the sequence image figure adjustment provided by the invention is characterized in that: device comprises host computer, multifunctional image acquisition and control card, parallel digital signal association transaction card; Wherein host computer and multifunctional image acquisition and control card are interconnected by high speed external apparatus interface bus, and multifunctional image acquisition and control card and parallel digital signal association transaction card are by externally expansion interface is interconnected between plate;

Host computer comprises control module, and control module control assessment device is finished following task: 1. realize man-machine interaction, receive the various control commands and the required configuration parameter of algorithm operation of user's input; 2. transmit control command and configuration parameter to the multifunctional image acquisition and control card; 3. send the image source data to the multifunctional image acquisition and control card; 4. receive the evaluating result that the multifunctional image acquisition and control card passes back and give display module; 5. real time monitoring multifunctional image acquisition and control card and parallel digital signal are assisted the duty of transaction card and it are fed back to user interface;

The multifunctional image acquisition and control card comprises that two are realized interconnected main control node by high speed bus interface and from Control Node; The main control node is finished following task: 1. receive control command, algorithm parameter, image source data from host computer; 2. finish self command interpretation decoding; 3. to transmitting control command and image source data from Control Node and other computing node; Finish following task from Control Node: 1. receive control command and view data that the main control node is transmitted; 2. finish self command interpretation decoding; 3. finish the calculating and the assessment of assessment indicator;

Parallel digital signal association transaction card comprises that the N sheet is by the interconnected computing node of high-speed bus realization, N 〉=1; Wherein a slice is the host computer node, and all the other are from computing node; The host computer node is finished following task: 1. receive on the multifunctional image acquisition and control card algorithm parameter that sent by the main control node and configuration order and transmit from computing node to other; 2. finish the unloading and the renewal of this computing node self algorithm solidification software bag; 3. finish the computing of this computing node self; 4. receive each result of calculation passed back from computing node and the merging of carrying out result images; Finish following task from computing node: 1. receive algorithm parameter and image source data that the host computer node is transmitted; 2. finish the unloading and the renewal of this computing node algorithm solidification software bag; 3. finish the computing of this computing node self; 4. to host computer node passback computing result.

The present invention can proofread and correct and calculate the corresponding point spread function of estimation to the caused degraded image of pneumatic optic transmission effect.Utilize correcting image and point spread function to calculate a series of assessment indicator, be used for pneumatic optic transmission effect composite measurement and analysis.The present invention not only is used for on-the-spot aero-optical effect wind tunnel test test and appraisal, and can be to the test and appraisal of aero-optical effect off-line and the laboratory aero-optical effect simulation study of high-speed aircraft practical flight process.Therefore, this assessment device both can be measured from sequence image and estimate the influence of aero-optical effect to imaging, but also off-line measurement is estimated the aero-optical effect of high-speed aircraft.

Taked structure at the assessment device of assessment method design based on " host computer+multifunctional image acquisition and control card+parallel digital signal association transaction card ", can expand hardware unit according to the complexity of algorithm and the requirement of real-time, have that the multi-channel data source obtains, extensibility is strong and the real-time advantages of higher.

Traditional in the past background technology is compared, and innovative point of the present invention is:

1. a kind of assessment method of the pneumatic optic transmission effect based on sequence image correction has been proposed.Be mainly used to of the influence of the aero-optical effect of online and the generation of off-line test and appraisal High Speed Flow Field to image quality.

The present invention estimates correcting image and point spread function by method for correcting image, and the aero-optical effect of quantitatively testing and assessing is to the influence of image quality, and then the aero-optical effect order of severity is measured.The test and appraisal of validity are carried out in the image blur that this device can cause aero-optical effect, skew, shake etc.The aero-optical effect assessment indicator mainly comprises following several: image blur, encircled energy, point spread function (PSF) supporting domain, Si Telieer are than Strehl, side-play amount, shake index.

2. at apparatus structure, proposed the parallel bearing calibration of aero-optical effect numeral and comprised that the space block parallel is proofreaied and correct, two-dimensional sampling walks abreast correction and parallel iteration is proofreaied and correct.

Space piecemeal bearing calibration can be divided into image overlapping plurality of sub piece mutually, simultaneously to each height piece fast processing that walks abreast, and the boundary effect of having avoided traditional piecemeal to handle.The parallel bearing calibration of two-dimensional sampling utilizes the mode of space sampling, entire image is sampled, thereby entire image is divided into the experimental process piece, so that walk abreast fast processing.Conventional flow ability of swimming iterative process in the correcting algorithm is become the point spread function estimation in the parallel iteration bearing calibration and correcting image is estimated the parallel iteration process, to improve treatment effeciency.

3. aero-optical effect assessment device Efficient and Flexible is controlled and Processing Structure

Assessment device adopts the structure based on " host computer+multifunctional image acquisition and control card+parallel digital signal association transaction card ".The simulating, verifying function that Embedded Real-Time proofreaies and correct and the problem of algorithm update functions are carried and do not had to being not easy to of having solved that the multi-computer system of existing high-performance standalone computer systems and Network Environment exists in the Pneumatic optical process of the test, and the net result of its checking can be directly used in the research of Pneumatic optical degeneration image correcting algorithm.

Utilize high performance parallel digital signal to assist transaction card to handle, it is limited to have overcome existing aero-optical effect demo plant calculated performance, finish correction calculation length consuming time, the shortcoming that the emulation cycle is long has shortened the time of doing needed analog simulation function in aero-optical effect test and appraisal early stage greatly.Under the situation of non-wind-tunnel collection, can use existing degraded image that the function and the effect of designed algorithm are verified fast.

This structure is compared with traditional multi-computer system based on many high-performance computers compositions, its volume is compared greatly with multi-computer system with power consumption and is reduced, the multi-computer system that structure is also formed based on many high-performance computers is simplified, when carrying out wind tunnel test, be easy to carry about with one and Installation and Debugging, shortened the time of setting up testing equipment.Because it uses special-purpose embedded high-performance digital signal processor to carry out treatment for correcting, achievement in research also can be directly used in the embedded corrective system of development.

Assessment device has the function of obtaining dissimilar image source data.Comprise image source data from the local hard drive storer, from network environment image source data; Utilize image source data that analogue camera collects and the image source data of utilizing digital camera to collect.The user can select control to data source by the software control platform based on this device.Make this device can adapt to the image acquisition and the processing in several data source.

Description of drawings

Fig. 1 is the processing flow chart of assessment method among the present invention;

Fig. 2 is the processing flow chart that the space block parallel is proofreaied and correct;

Fig. 3 is space inverted pleat piecemeal (a four) schematic diagram;

Fig. 4 is based on the processing flow chart of the loop iteration recovery algorithm of maximum likelihood estimation;

Fig. 5 is the parallel processing flow chart of proofreading and correct of two-dimensional sampling;

Fig. 6 is two-dimensional sampling correction principle figure;

Fig. 7 is the processing flow chart that parallel iteration is proofreaied and correct;

Fig. 8 is the wind-tunnel degeneration image sequence.Wherein, Fig. 8 (a) wind-tunnel degenerate series image (sequence number 1); Fig. 8 (b) wind-tunnel degenerate series image (sequence number 2); Fig. 8 (c) wind-tunnel degenerate series image (sequence number 3); Fig. 8 (d) wind-tunnel degenerate series image (sequence number 4); Fig. 8 (e) wind-tunnel degenerate series image (sequence number 5); Fig. 8 (f) correcting image; The pneumatic optic transmission effect PSF (sequence number 1) that Fig. 8 (g) device estimates; The pneumatic optic transmission effect PSF (sequence number 2) that Fig. 8 (h) device estimates; The pneumatic optic transmission effect PSF (sequence number 3) that Fig. 8 (i) device estimates; The pneumatic optic transmission effect PSF (sequence number 4) that Fig. 8 (j) device estimates; The pneumatic optic transmission effect PSF (sequence number 5) that Fig. 8 (k) device estimates; Fig. 8 (l) PSF transverse center section curve (sequence number 1); Fig. 8 (m) PSF longitudinal center's section curve (sequence number 1); Fig. 8 (n) PSF transverse center section curve (sequence number 2); Fig. 8 (o) PSF longitudinal center's section curve (sequence number 2); Fig. 8 (p) PSF transverse center section curve (sequence number 3); Fig. 8 (q) PSF longitudinal center's section curve (sequence number 3); Fig. 8 (r) PSF transverse center section curve (sequence number 4); Fig. 8 (s) PSF longitudinal center's section curve (sequence number 4); Fig. 8 (t) PSF transverse center section curve (sequence number 5); Fig. 8 (u) PSF longitudinal center's section curve (sequence number 5);

Fig. 9 is the data flowchart of assessment method of the present invention.

Figure 10 is the structured flowchart of assessment device of the present invention.

Figure 11 is an assessment device detailed block diagram of the present invention.

Figure 12 is that assessment device internal module of the present invention calls hierarchical relationship figure.

Figure 13 is that device carries out continuous-flow type iterative computation data flow diagram.

Figure 14 is that device carries out parallel type iterative computation data flow diagram.

Figure 15 is an assessment device workflow diagram of the present invention.

Embodiment

The present invention is further detailed explanation below in conjunction with accompanying drawing and example.

As shown in Figure 1, assessment method of the present invention may further comprise the steps:

(1) gather Pneumatic optical trial image g (x, y), (x, y) pixel coordinates on presentation video height and the Width respectively;

The image that collects can be the aero-optical effect sequential image data of obtaining in the wind tunnel test, or utilizes high-speed aircraft to carry the pneumatic optic transmission effect view data that imaging detector obtains in the high-speed flight process.

(2) (x y) proofreaies and correct to the Pneumatic optical trial image g that collects;

To the Pneumatic optical trial image g that collects (x, y) utilize the image rectification algorithm estimate unknown benchmark image f (x, y) and point spread function h (x, y).The correcting image and the point spread function that estimate are used respectively

With

Expression.

The present invention utilizes following three kinds of parallel bearing calibrations:

(a) space block parallel bearing calibration

In conjunction with the accompanying drawings 2, the space divides the fast parallel detailed step of proofreading and correct to be:

A.1 (x y) is divided into four sub-pieces, and sub-piece is of a size of 2 power power, and makes overlapping areas is arranged mutually between the sub-piece with Pneumatic optical trial image g to be corrected;

Because employed Fourier transform is based on full figure in the Flame Image Process, exist contact closely between the adjacent image point, therefore, the boundary effect that causes in order to reduce sub-piece splicing place pixel gray scale sudden change, image is divided into four of inverted pleat, and making has overlapping areas mutually between the sub-piece.Simultaneously, in order to utilize fast fourier transform, the size of sub-piece should be 2 power power, and therefore, each height piece should remove the border of same widths with respect to the overlapping region.Guaranteeing that sub-piece is of a size of under the prerequisite of 2 power power, makes little that overlapping areas is tried one's best between the sub-piece.With image space inverted pleat piecemeal is that four schematic diagram is seen accompanying drawing 3.

A.2 four sub-pieces of image are used the loop iteration recovery algorithm of estimating based on maximum likelihood respectively and handle, obtain the correcting image and the point spread function of four sub-pieces;

In conjunction with the accompanying drawings 4, the treatment step that the loop iteration of estimating based on maximum likelihood restores algorithm is:

1. set iterations, this algorithm comprises in two circulation and an outer circulation, and circulation is respectively the calculation level spread function and circulates that (iterations is set at M in two _Psf) and the calculation correction image circulates, and (iterations is set at M _Obj), the outer circulation iterations is set at N ₀M _PsfGenerally get 1-50, M _ObjGenerally get 1-10, N ₀Generally get 10-50.Set the initial value of correcting image and point spread function, the initial value of correcting image and point spread function can be set at complete 1 matrix.Make n=0, m=0, k=0, n represent the number of times that the outer circulation iteration has been carried out, m represents the number of times that the point spread function loop iteration has been carried out.K represents the number of times that the point spread function loop iteration has been carried out.

2. loop iteration calculation level spread function.According to following formula iterative computation point spread function:

${\hat{h}}_{n,m+1}(x,y)=\{\left(\frac{g(x,y)}{{\hat{f}}_{n,m}(x,y)\⊗{\hat{h}}_{n,m}(x,y)}\right)\·{\hat{f}}_{n,m}(x,y)\}*{\hat{h}}_{n,m}(x,y)$

Wherein, g (x, y) the Pneumatic optical trial image that collects of expression,

The point spread function of representing this iteration, With

Represent correcting image and point spread function that last iteration calculates respectively,

The expression convolution algorithm, the expression related operation, * represents dot product.

3. judge that whether m is less than M _Psf, if, make m=m+1, change step then over to 2.; Otherwise enter step 4.

4. loop iteration calculation correction image.According to following formula iterative computation correcting image:

${\hat{f}}_{n,k+1}(x,y)=\{\left(\frac{g(x,y)}{{\hat{f}}_{n,k}(x,y)\⊗{\hat{h}}_{n,k}(x,y)}\right)\·{\hat{h}}_{n,k}(x,y)\}*{\hat{f}}_{n,k}(x,y)$

Wherein, g (x, y) the Pneumatic optical trial image that collects of expression,

The correcting image of representing this iteration,

With

Represent correcting image and point spread function that last iteration calculates respectively,

5. judge that whether k is less than M _Obj, if, make k=k+1, change step then over to 4.; Otherwise enter step 6.

6. make n=n+1, judge that whether n is less than N ₀, if change step then over to 2.; Finish otherwise calculate.

The loop iteration recovery algorithm of using among the present invention of estimating based on maximum likelihood more describes visible document 6: Hong Hanyu in detail, Zhang Tianxu etc. aero-optical effect degraded image loop iteration restores algorithm research, Central China University of Science and Technology's journal (natural science edition)), 2005,33 (9): 15-18.]

A.3 the correcting image and the point spread function of four sub-pieces that a.2 step is estimated splice, and form and original image space to be corrected big or small identical correcting image and point spread function.

A.4 a.3 step is spliced the correcting image and the point spread function that and carry out smoothing processing, obtain correcting image

And point spread function

Smoothing processing can be utilized the method for low-pass filtering.

(b) the parallel bearing calibration of two-dimensional sampling

In conjunction with the accompanying drawings 5, the parallel detailed step of proofreading and correct of two-dimensional sampling is:

B.1 utilize the mode of space sampling, (x is to sample in the sampling interval with 2 y), obtains 4 sub-pieces with the Pneumatic optical trial image g that collects.

Concrete sample mode is: fetch bit is formed first sub-piece in the pixel of even number line and even column respectively, fetch bit is formed second sub-piece in the pixel of even number line and odd column, fetch bit is formed the 3rd sub-piece in the pixel of odd-numbered line and even column, and fetch bit is formed the 4th sub-piece in the pixel of odd-numbered line and odd column.

B.2 4 sub-pieces of image being used the loop iteration recovery algorithm of estimating based on maximum likelihood respectively handles;

B.3 the correcting image and the point spread function of 4 sub-pieces that a.2 step is estimated splice, and obtain correcting image

And point spread function

Splicing is a step inverse process b.1, and concrete mode is: 4 sub-pieces that b.2 step estimates are filled back whole figure by the position of each pixel in whole figure, form the image after proofreading and correct.

The two-dimensional sampling correction principle is seen accompanying drawing 6.Among Fig. 6, numeral 1 is represented odd-numbered line odd column pixel, and numeral 2 is represented odd-numbered line even column pixel, and numeral 3 is represented even number line odd column pixel, and numeral 4 is represented even number line even column pixel.

(c) parallel iteration bearing calibration

As shown in Figure 7, the parallel iteration bearing calibration may further comprise the steps:

(c1) set iterations M _Psf, M _Obj, N ₀, M _PsfBe the iterations of calculation level spread function, M _ObjBe the iterations of calculation correction image, N ₀Iterations for outer circulation; And the initial value of setting correcting image and point spread function; Make n=0, m=0, k=0, n represent the number of times that the outer circulation iteration has been carried out, m represents the number of times that the point spread function loop iteration has been carried out.K represents the number of times that the correcting image loop iteration has been carried out.

(c2) carry out process (F) and process (G) simultaneously;

Process (F): according to following step iterative computation point spread function;

(F1) according to following formula iterative computation point spread function:

${\hat{h}}_{n,m+1}(x,y)=\{\left(\frac{g(x,y)}{{\hat{f}}_{n,m}(x,y)\⊗{\hat{h}}_{n,m}(x,y)}\right)\·{\hat{f}}_{n,m}(x,y)\}*{\hat{h}}_{n,m}(x,y)$

Wherein, g (x, y) the Pneumatic optical trial image that collects of expression,

The point spread function of representing this iteration,

With

Represent correcting image and point spread function that last iteration calculates respectively;

(F2) judge that whether m is less than M _Psf, if, make m=m+1, change step (F1) then over to; Otherwise enter step (c3)

Process (G): according to following step iterative computation correcting image:

(G1) according to following formula iterative computation correcting image:

${\hat{f}}_{n,k+1}(x,y)=\{\left(\frac{g(x,y)}{{\hat{f}}_{n,k}(x,y)\⊗{\hat{h}}_{n,k}(x,y)}\right)\·{\hat{h}}_{n,k}(x,y)\}*{\hat{f}}_{n,k}(x,y)$

Wherein, g (x, y) the Pneumatic optical trial image that collects of expression,

The correcting image of representing this iteration,

With

Represent correcting image and point spread function that last iteration calculates respectively,

(G2) judge that whether k is less than M _Obj, if, make k=k+1, change step (G1) then over to; Otherwise enter step (c3)

(c3) deterministic process (F) and (G) whether all finish, if, enter step (C4), otherwise wait for, all finish until process (F) with (G), enter step (c4) again;

(c4) make n=n+1, judge that whether n is less than N ₀, if change step (c2) then over to; Finish otherwise calculate.

(3) correcting image that step (2) is obtained

And point spread function

Calculate, obtain assessment indicator.

Utilize the correcting image of estimating The point spread function of estimating

(x y) calculates assessment indicator with the trial image g that collects.Assessment indicator comprises: image blur, encircled energy, point spread function supporting domain, Si Telieer are than (Strehl ratio), side-play amount and jitter range.The definition and the computing method of each assessment indicator are as follows:

(a) computed image blur level

For a given gray level image, as follows to the computation process of image blur:

A.1 view picture figure being carried out 1 layer of Ha Er (Harr) discrete wavelet decomposes;

Treating evaluation map looks like to carry out to have obtained low frequency component matrix L L, level detail component matrix H L, vertical detail component matrix L H and diagonal detail component matrix H H behind one deck Haar wavelet transformation.

A.2 according to following formula diagonal detail component matrix H H is calculated | HH| _Gray

${\left|\mathrm{HH}\right|}_{\mathrm{gray}}(m,n)=\frac{\left|\mathrm{HH}\right|(m,n)-\min \left(\right|\mathrm{HH}\left|\right)}{\max \left(\right|\mathrm{HH}\left|\right)-\min \left(\right|\mathrm{HH}\left|\right)}\×255$

A.3 travel through | HH| _GrayIn each pixel, judge whether it is the fuzzy edge unique point, the number N of record fuzzy edge unique point _cRecord simultaneously | HH| _GrayMiddle gray-scale value is greater than the number N of zero picture element _Nonzero

The fuzzy edge unique point is meant the picture element that the value of the picture element that eight neighborhoods of a certain picture element are interior all equates with the value of this picture element.

A.4 the blur level of determining full figure is D _b:

(b) encircled energy

The encircled energy Y of definition original image and benchmark image _MoFor:

$Y_{\mathrm{mo}}=\frac{h_m}{h_o}$

h _oBe the pixel peak value on the benchmark image appointed area; h _mBe the pixel peak value on the original image appointed area.

(c) point spread function (PSF) supporting domain

Point spread function (PSF) supporting domain (being effective range) is meant that point spread function is the size in the zone of nonzero value.

(d) Si Telieer is than (Strehl ratio)

The definition of Strehl ratio is as follows:

$\mathrm{Strehl}=\frac{p_{\max }}{y_{\max }}$

Wherein, p _MaxThe expression correcting image (or acquired original image) with respect to the PSF energy normalized of benchmark image after peak value, y _MaxPeak value after the expression imaging system diffraction PSF energy normalized.

(e) side-play amount

Side-play amount be in the correcting image scenery target with respect to the gray scale centroid motion pixel of scenery target in the benchmark image.

The gray scale barycenter of scenery target is defined as:

$X_c=\frac{\underset{(i,j)\∈T_k}{\mathrm{\Σ}}I(i,j)*i}{S_{\mathrm{sum}}}$ $Y_c=\frac{\underset{(i,j)\∈T_k}{\mathrm{\Σ}}I(i,j)*j}{S_{\mathrm{sum}}}$

In the formula, the pixel coordinates of i and j presentation video, I (i, j) (i, the j) gray-scale value at pixel place, S in the presentation video _SumExpression scenery target pixel gray-scale value sum.X _cBe the barycenter on the picture traverse direction, Y _cBe the barycenter on the picture altitude direction.T _kExpression entire image zone.

Correcting image

Barycenter and benchmark image f (x, the difference of barycenter y) is side-play amount.Side-play amount δ on the picture altitude direction _xExpression, the side-play amount δ on the picture traverse direction _yExpression.

(f) shake index

The calculating object of shake index be the multiframe original image that collects and proofread and correct after multiple image.The shake index comprises: the average of side-play amount and jitter range.

The average of side-play amount:

${\mathrm{\μ}}_x=\frac{1}{N}\mathrm{\Σ}{\mathrm{\δ}}_x$ ${\mathrm{\μ}}_y=\frac{1}{N}\mathrm{\Σ}{\mathrm{\δ}}_y$

Wherein, δ _x, δ _yBe the side-play amount of every two field picture on x (width) and y (highly) direction, N is total frame number.

Jitter range:

The expression sequence image makes it poor with respect to maximal value and the minimum that benchmark image is offset on x (width) and y (highly) direction.(maximum magnitude)

(4) pneumatic optic transmission effect comprehensive test and analysis.

A series of assessment indicators that step (3) is calculated are used for pneumatic optic transmission effect is carried out comprehensive test and analysis.

(a) utilize image blur that the order of severity of pneumatic optic transmission effect is analyzed;

Blur level has characterized the fog-level of image, and blur level is big more to show that the original image that collects is fuzzy more, illustrates that pneumatic optic transmission effect is serious more to the influence of image quality.Generally can think: when blur level less than 0.1 the time, image is more clear, a little less than the influence of pneumatic optic transmission effect to image quality; When blur level greater than 0.1 less than 0.2 the time, image exists to a certain degree fuzzy, pneumatic optic transmission effect has certain influence to image quality; When blur level greater than 0.2 the time, the fog-level of image is bigger, pneumatic optic transmission effect is more serious to the influence of image quality.

(b) utilize encircled energy that pneumatic optic transmission effect is analyzed the influence of scenery target imaging;

The encircled energy of image has reflected the concentration of energy degree at designated pixel point place, in scenery target pixel point position, the encircled energy of image is more little, shows the energy diffusion more of scenery target, image quality is poor more, illustrates that pneumatic optic transmission effect is serious more to the influence of image quality.Generally can think: when the encircled energy of image less than 0.5 the time, the energy of scenery target has bigger diffusion, pneumatic optic transmission effect is very serious to the influence of scenery target imaging; When the encircled energy of image greater than 0.5 less than 0.85 the time, the energy of scenery target has certain diffusion, pneumatic optic transmission effect has certain influence to the scenery target imaging; When blur level greater than 0.85 the time, the energy of scenery target has slight spread, pneumatic optic transmission effect to the influence of scenery target imaging a little less than.

(c) utilize the supporting domain of point spread function (PSF) that the order of severity of pneumatic optic transmission effect is analyzed;

The supporting domain of point spread function (PSF) is big more, illustrates that pneumatic optic transmission effect is serious more to the influence of image quality.Generally can think: when the effective range of point spread function (PSF) supporting domain during less than 7 * 7 pixels, a little less than the influence of pneumatic optic transmission effect to image quality; When the effective range of point spread function (PSF) supporting domain greater than 7 * 7 less than 21 * 21 the time, pneumatic optic transmission effect has certain influence to image quality; When the effective range of point spread function (PSF) supporting domain greater than 21 * 21 the time, pneumatic optic transmission effect is more serious to the influence of image quality.

With wind-tunnel degeneration image sequence in the accompanying drawing 8 and to proofread and correct the result is that example describes: the effective range corresponding to the supporting domain of point spread function Fig. 8 (g) of Fig. 8 (a) is approximately 15 * 15 pixels, can think that pneumatic optic transmission effect has certain influence to the imaging of Fig. 8 (a).Effective range corresponding to the supporting domain of Fig. 8 (h)～(k) of the point spread function of Fig. 8 (b)～(e) all is approximately 80 * 80 pixels, and pneumatic optic transmission effect is more serious to the imaging influence of Fig. 8 (b)～(e).

(d) utilize the order of severity of Si Telieer comparison pneumatic optic transmission effect to analyze;

The value of Si Telieer ratio has reflected the departure degree of original image (or correcting image) with respect to the point spread function of the point spread function of benchmark image and imaging system diffraction, and its value is big more, and both are approaching more, otherwise both depart from more.When Si Telieer than the value of (Strehl than) hour, illustrate that pneumatic optic transmission effect is more serious to the influence of image quality.Generally can think: when Si Telieer than less than 0.4 the time, pneumatic optic transmission effect is more serious to the influence of image quality; When Si Telieer than greater than 0.4 less than 0.7 the time, pneumatic optic transmission effect has certain influence to image quality; When Si Telieer than greater than 0.7 the time, a little less than the influence of pneumatic optic transmission effect to image quality.

(e) utilize the side-play amount of scenery target that pneumatic optic transmission effect is analyzed the influence of scenery target imaging;

Side-play amount is big more to show that the scenery target in the trial image that collects departs from the scenery target in the benchmark image more, illustrates that pneumatic optic transmission effect is more serious to the influence of scenery target imaging.Generally can think: when the side-play amount of image during less than 2 pixels, the scenery target imaging departs from less, and pneumatic optic transmission effect is fainter to the influence of scenery target imaging; When the side-play amount of image greater than 2 pixels during less than 5 pixels, the scenery target imaging has necessarily and departs from, pneumatic optic transmission effect has certain influence to the scenery target imaging; When the side-play amount of image during greater than 5 pixels, the scenery target imaging has more greatly and departs from, and pneumatic optic transmission effect is slight to the influence of scenery target imaging.

(f) utilize the shake index that pneumatic optic transmission effect is analyzed the influence of scenery target sequence image imaging;

The shake index is the statistical measurement to scenery target in the image sequence, and its value is big more to show that the scenery target in the original sequence departs from the scenery target in the benchmark image more, illustrates that pneumatic optic transmission effect is serious more to the influence of sequence image image quality.The shake index is a unit with side-play amount (pixel).Generally can think: when the average of the side-play amount of image sequence less than 1 pixel, jitter range is during less than 3 pixels, pneumatic optic transmission effect is very faint to the influence of scenery target imaging; When the average of the side-play amount of image sequence greater than 1 pixel during less than 3 pixels, jitter range is greater than 3 during less than 5 pixels, pneumatic optic transmission effect has certain influence to the scenery target imaging; When the average of the side-play amount of image sequence greater than 3 pixels, jitter range is during greater than 5 pixels, pneumatic optic transmission effect is more serious to the influence of scenery target imaging.

Data flowchart based on the assessment method of the pneumatic optic transmission effect of sequence image correction is seen accompanying drawing 9.Collect the original image and the point spread function that obtain estimating after the calibrated processing of Pneumatic optical trial image, original image of estimating and point spread function are used to calculate a series of assessment indicator with the trial image of collection, and the assessment indicator that obtains is used for pneumatic optic transmission effect is carried out comprehensive test and analysis.

The device of realizing above-mentioned assessment method comprises host computer 1, multifunctional image acquisition and control card 2 (hereinafter to be referred as control card 2), parallel digital signal association transaction card 3 (hereinafter to be referred as association's transaction card 3) as shown in Figure 10.Wherein host computer 1 is interconnected by high speed external apparatus interface bus with control card 2, and control card 2 and association's transaction card 3 are by externally expansion interface is interconnected between plate; Association's transaction card 3 can be interconnected with the mode of cascade and more associations transaction card, thereby the computation capability that can control assessment device flexibly is to adapt to complexity computing time of algorithms of different, to reach the real-time requirement of assessment device computing.

Host computer 1 is generally the server that stable performance, memory capacity are big, Peripheral Interface is abundant.Host computer 1 mainly comprises control module 4, and control module 4 control assessment devices are finished following task: (a) realize man-machine interaction, receive the various control commands and the required configuration parameter of algorithm operation of user's input; (b) transmit control command and configuration parameter to control card 2; (c) send the image source data to control card 2; (d) receive the evaluating result that control card 2 passes back and give display module; (e) duty of real time monitoring control card 2 and association's transaction card 3 and with feedback of status to user interface.

Control card 2 comprises two by interconnected Control Node 5A, the 5B of high speed bus interface realization.Wherein Control Node 5A is the main control node, and Control Node 5B is from Control Node.Main control node 5A finishes following task: (a) receive control command, algorithm parameter, image source data from host computer 1; (b) finish self command interpretation decoding; (c) to transmitting control command and image source data from Control Node 5B and computing node 6.1～6.N; Finish following task from Control Node 5B: (a) receive control command and view data that main control node 5A transmits; (b) finish self command interpretation decoding; (c) finish the calculating and the assessment of assessment indicator;

Association's transaction card 3 comprises that the N sheet is by interconnected computing node 6.1, the 6.2...6.N of high-speed bus realization, N 〉=1.Reach the characteristics high to the Floating-point Computation processing requirements greatly at Measurement Algorithm complexity computing time, computing node 6.1,6.2...6.N can be selected the high performance float-point digital signal processor TS201 of ADI company for use.In actual applications, can select the quantity N of computing node flexibly according to demands such as the complicacy of calculation task, real-times.Wherein computing node 6.1 is the host computer node, and 6.2～6.N is from computing node.Host computer node 6.1 is finished following task: (a) receive the algorithm parameter that sent by main control node 5A on the control card 2 and configuration order and to transmitting from computing node 6.2～6.N; (b) finish unloading and the renewal that this computing node self algorithm solidifies bag; (c) finish the computing of this computing node self; (d) receive each result of calculation passed back from computing node 6.2～6.N and the merging of carrying out result images; Finish following task from computing node 6.2～6.N: (a) receive algorithm parameter and image source data that host computer node 6.1 is transmitted; (b) finish this unloading and renewal from computing node self algorithm solidification software bag; (c) finish the computing of this computing node self; (d) to host computer node 6.1 passback computing results.

In conjunction with the accompanying drawings 12, inner each module invokes relation of apparatus of the present invention is described.

Control module 4 sends to main control node 5A with the various algorithm parameters and the control command of user's input by the high-speed peripheral interface, main control node 5A promptly explains decoding to control command, simultaneously to transmitting this control command, also finish explanation decoding simultaneously to control command from Control Node 5B from Control Node 5B.According to the difference of control command explanation decode results, each Control Node is finished following different task respectively with computing node: if (a) order is " images acquired " order, then main control node 5A carries out image is gathered; (b) if order is " solidification software renewal ", then computing node 6.1～6.N begins to be cured the unloading and the renewal process of software package; (c) if order is " entering algorithm process ", then control module 4 begins to send the image source data, and the computing module that computing node 6.1～6.N calls separately begins to carry out the image calculation processing; (d) if order is " calculating assessment indicator ", then the result images of finishing 6.1 passbacks of host computer node from Control Node 5B carries out the calculating of assessment indicator.

Below to count N=4 with computing node be example, each part mentioned above is described in further detail.

Control module 4 comprises command parameter configuration module 4.1, image source data transmission blocks 4.2 and display module 4.3.

Command parameter configuration module 4.1 is used to realize man-machine interaction, receives various algorithm parameters and control command that the user sends, and sends to control card 2 by high speed external apparatus interface various parameters and control command that algorithm is required.

Image source data transmission blocks 4.2 is used for and will is stored in the image source data of host computer local hard drive storer and sends to control card 2 from the image source data stream in the network environment by interconnected with it external apparatus interface.

Display module 4.3 is used for showing in real time the image that camera collects, image, network image, assessment indicator result of calculation and the assessment device work state information of local hard drive storer.

Main control node 5A comprises command interpretation decoding module 5A.1, multi-path image source data acquisition module 5A.2.

Command interpretation decoding module 5A.1 finishes the explanation of the control command of host computer 1 decoding, will order simultaneously and controlled variable sends to from Control Node 5B and computing node 6.1～6.4.Its task is as follows: (a) receive various algorithm parameters and control command that host computer parameter and command configuration module 4.1 send, (b) to the control command decoding that makes an explanation; (c) algorithm parameter and control command are transmitted to from Control Node 5.B and computing node 6.1～6.4; (d) carry out the scheduling and the execution of algorithm task according to decode results.

With the transmitting control commands is example, and the principle of work of command interpretation decoding module is described.

(a) command interpretation decoding module 5A.1 receives the control command that host computer 1 sends, and reads the control command word.

(b) look-up command mapping table finds the task jump address identical with the order that receives from table.

(c) should order and be distributed to simultaneously from Control Node 5B and computing node 6.1～6.4.

(d) carry out the task redirect of this Control Node, carry out corresponding task.

With the transmission algorithm parameter is example, and the principle of work of command interpretation decoding module is described.

(a) command interpretation decoding module 5A.1 receives the algorithm configuration parameter that host computer 1 sends.

(b) the external expansion interface of configuration parameter by control card sent to computing node on association's transaction card 3.The task that module 5A.1 will carry out redirect may be one of following several situations:

(a) start multi-path image source data acquisition task, begin to carry out image acquisition, but do not carry out computing.

(b) start multi-path image source data acquisition task, begin to carry out image acquisition, start calculation task simultaneously and begin computing.

(c) each computing node transmitting control commands on association's transaction card 3 notifies it to start the solidification software updating task, promptly carries out the curing again of algorithm solidification software bag and upgrades.

(d) stop calculation task, assessment device is transferred to idle condition.

Multi-path image source data acquisition module 5A.2 provides the image input function for the computing of assessment device, to obtain the image source data of separate sources, comprises following several: (a) from the image source data of host computer harddisk memory; (b) from the image source data of the network environment of host computer; (c) the image source data that collect of control card analogue camera interface; (d) the image source data that collect of control card digital camera interface; Wherein (a) and (b) are the data that are stored on this locality or other server hard disc storer, send control card 2 by host computer to by the high speed external bus; (c), (d) be realtime image data, by the image capture interface input of control card self.In the actual motion, can utilize the software control platform to finish data source and select and switching, in the time of need not handling when the needs collection, this module is only finished the image acquisition task, and the assessment device of this moment can be used as an image pick-up card; When needs collection and processing in real time, this module handles for association's transaction card 3 when gathering and with image data transmission, and this moment, this device can be finished real-time collection, processing and pneumatic optical transmission effect evaluation.

Comprise that from Control Node 5B command interpretation decoding module 5B.1, assessment indicator calculate evaluation module 5B.2.

The decoding principle of command interpretation decoding module 5B.1 is identical with 5A.1, mainly finishes the command interpretation decoding task from Control Node 5B self, and just its parameter and control command are by main control node 5A input rather than by host computer 1 input.Its principle of work is as follows:

(a) command interpretation decoding module 5B.1 receives the control command that main control node 5A transmits, and reads the control command word.

(b) look-up command mapping table finds the task jump address identical with the order that receives from table.

(c) carry out the task redirect of this Control Node, carry out corresponding task.

The task that module 5B.1 will carry out redirect may be one of following several situations:

(a) start the assessment indicator calculation task, finish the calculating of assessment indicators such as shake, skew.

(b) to the result images of display module 4.3 passback processing and the result of calculation of assessment indicator.

The main task that assessment indicator calculates evaluation module 5B.2 is as follows: the result of calculation image that (a) receives association's transaction card 3 passbacks; (b) result images is carried out the further calculating of assessment indicator, draw the various efficiency indexs that propose in the measuring method, as the centroid motion of correcting image and degraded image, blur level, encircled energy etc.; (c) to host computer 1 display module 4.3 passback assessment indicator result of calculations.

Association's transaction card 3 comprises that host computer node 6.1 reaches from computing node 6.2,6.3,6.4.Each host computer node reaches and includes a solidification software update module and computing module respectively from computing node.

Solidification software update module 6.1.1,6.2.1,6.3.1,6.4.1 provide the unloading and the update functions of algorithm solidification software bag for computing node 6.1,6.2,6.3,6.4.The aero-optical effect assessment device allow with different assessment methods with the mode of software package be cured to the corresponding nonvolatile memory of each computing node in, guiding and load the algorithm software that upgrades when powering in system.

Computing module 6.1.2,6.2.2,6.3.2,6.4.2 handle the degeneration image sequence of input and the output calibration image sequence.It is input as the degenerate series image, is output as the correction result images sequence after handling as calculated, and this correcting image sequence is carried out the foundation that assessment indicator calculates as next step.Wherein computing node 6.1 is the host computer node, and computing node 6.2,6.3,6.4 is from computing module.Host computer node 6.1 receives degeneration image sequences and view data is distributed to from computing node 6.2,6.3,6.4, result of calculation is passed after the computing node computing back host computer node 6.1 and carries out the result and merge.With computing node number N=4 is example, and the principle of work of computing module is described.

Accompanying drawing 13 has represented to restore according to the loop iteration of estimating based on maximum likelihood the continuous-flow type iterative scheme of algorithm computation.Computing node 6.1～6.4 restores algorithm to a frame blurred picture according to the loop iteration of estimating based on maximum likelihood respectively separately and carries out the sequential iterative computation, and iterative computation is passed the result back control card 2 after finishing successively.Concrete steps are as follows: (1) host computer node 6.1 receives blurred picture sequence S01, S02, S03, the S04 of control card 2; (2) the host computer node is temporarily preserved S01, simultaneously S02, S03, S04 is distributed to from computing node 6.2,6.3,6.4; (3) computing node 6.1～6.4 restores algorithm according to the loop iteration of estimating based on maximum likelihood simultaneously; (4) iteration finishes back host computer node 6.1 output calibration image R01, from computing node 6.2,6.3,6.4 output calibration image sequence R02, R03, R04 and pass to host computer node 6.1; (5) host computer node 6.1 returns correcting image R01, R02, R03, R04 to control card 2 successively; (6) repeating step (1)～(5).

Accompanying drawing 14 has been represented the parallel type iterative scheme that calculates according to the parallel iteration bearing calibration.Its principle is that computing node is divided into two groups, and first group is that 6.1,6.4, the second groups of computing nodes are computing node 6.2,6.3.With first group is example, computing node 6.1 only carries out PSF and calculates, computing node 6.4 only carries out OBJ and calculates, computing node 6.1,6.4 carries out the parallel type iteration simultaneously according to the parallel iteration bearing calibration, second group identical with first group of principle of work, after iteration finishes, pass result of calculation back control card 2 by computing node 6.3,6.4.Concrete steps are as follows: (1) computing node 6.1 receives the blurred picture sequence that control card 2 sends, and establishing frame number is S01, and computing node 6.2 receives the blurred picture sequence that control card 2 sends simultaneously, and establishing frame number is S02; (2) computing node 6.1, computing node 6.4 carry out the parallel type iterative computation as one group according to the parallel iteration bearing calibration, and computing node 6.2,6.3 carries out the parallel type iterative computation as another group according to the parallel iteration bearing calibration simultaneously; (3) iteration finishes, and computing node 6.3,6.4 is returned to control card 2 with correcting image R01, R02; (4) repeat (1)～(3).

Be the process flow diagram of assessment device of pneumatic optic transmission effect work as shown in Figure 15.Constitute and the module hierarchical relationship in conjunction with aforesaid hardware, the device work principle is described.

(a) behind the device electrification reset, enter " free time " state, wait for receiving parameter and control command.

(b) in a single day receive parameter and control command, then enter the command interpretation decoding module and carry out command interpretation decoding.If, then enter step (f), otherwise enter next step for " entering processing " order.

(c) if order for " entering software upgrading ", then enters next step, otherwise enters step (m).

(d) be cured the software upgrading process.

(e) if the solidification software renewal process finishes, then enter step (m), otherwise repeating step (d)～(e).

(f) if order for " entering algorithm process ", then enters step (i), otherwise enters next step.

(g) if order for " images acquired ", then enters next step, otherwise enters step (m).

(h) carry out image acquisition process.

(i) carry out the algorithm computation processing procedure.

(j) carry out the calculating evaluation process of assessment indicator.

(k) show images acquired, correction result and assessment indicator

(l) judge whether all images sequence disposes, if then enter next step, otherwise repeating step (b)～(l).

(m) enter " free time " state, the wait device resets and restarts.

Table 1 is that many DSP handle computing time, handle frame frequency and speed-up ratio table

Claims (1)

1. pneumatic optic transmission effect Measurement and analysis method based on the sequence image figure adjustment, its step comprises:

(1) gather Pneumatic optical trial image g (x, y);

(2) (x y) utilizes the image rectification algorithm to estimate that unknown benchmark image f (x, y) (x, y), use respectively with point spread function h by the correcting image and the point spread function that estimate to the Pneumatic optical trial image g that collects

With Expression;

(3) utilize the correcting image of estimating The point spread function of estimating

(x y) calculates the Measurement and analysis index, and the Measurement and analysis index comprises a kind of or several arbitrarily than in, side-play amount and the jitter range of image blur, encircled energy, point spread function supporting domain, Si Telieer with the trial image g that collects;

(4) the Measurement and analysis index of utilizing step (three) to obtain is analyzed, and wherein, image blur is used for the order of severity of pneumatic optic transmission effect is analyzed; Encircled energy is used for pneumatic optic transmission effect is analyzed the influence of scenery target imaging; The supporting domain of point spread function is used for the order of severity of pneumatic optic transmission effect is analyzed; Si Telieer analyzes than the order of severity that is used for pneumatic optic transmission effect; The side-play amount of scenery target is used for pneumatic optic transmission effect is analyzed the influence of scenery target imaging; Jitter range is used for pneumatic optic transmission effect is analyzed the influence of scenery target sequence image imaging;

Step (two) estimate according to one of following three kinds of image rectification algorithms:

The process of first kind of image rectification algorithm is:

(a1) (x y) is divided into four sub-pieces, and sub-piece is of a size of 2 power power, and makes overlapping areas is arranged mutually between the sub-piece with Pneumatic optical trial image g to be corrected;

(a2) four sub-pieces of image are used the loop iteration recovery algorithm of estimating based on maximum likelihood respectively and handle, obtain the correcting image and the point spread function of four sub-pieces;

(a3) correcting image and the point spread function of four sub-pieces that step (a2) is obtained splice, and form and original image space to be corrected big or small identical correcting image and point spread function;

(a4) step (a3) is spliced the correcting image and the point spread function that and carry out smoothing processing, obtain correcting image

And point spread function

The process of second kind of image rectification algorithm is:

(b1) (x y) carries out interval sampling, obtains a plurality of sub-pieces with the Pneumatic optical trial image g that collects;

(b2) each height piece of image is used the loop iteration recovery algorithm of estimating based on maximum likelihood respectively and handle, obtain the correcting image and the point spread function of each height piece;

(b3) correcting image and the point spread function of each height piece that step (b2) is obtained splice, and obtain correcting image

And point spread function

The process of the third image rectification algorithm is:

(c1) set iterations M _Psf, M _Obj, N ₀, M _PsfBe the iterations of calculation level spread function, M _ObjBe the iterations of calculation correction image, N ₀Iterations for outer circulation; And the initial value of setting correcting image and point spread function; Make n=0, m=0, k=0, n represent the number of times that the outer circulation iteration has been carried out, and m represents the number of times that the point spread function loop iteration has been carried out, and k represents the number of times that the correcting image loop iteration has been carried out;

(c2) carry out process (F) and process (G) simultaneously;

Process (F): according to following step iterative computation point spread function;

(F1) according to following formula iterative computation point spread function:

${\hat{h}}_{n,m+1}(x,y)=\{\left(\frac{g(x,y)}{{\hat{f}}_{n,m}(x,y)\⊗{\hat{h}}_{n,m}(x,y)}\right)\·{\hat{f}}_{n,m}(x,y)\}*{\hat{h}}_{n,m}(x,y)$

Wherein, g (x, y) the Pneumatic optical trial image that collects of expression,

The point spread function of representing this iteration,

With Represent correcting image and point spread function that last iteration calculates respectively;

(F2) judge that whether m is less than M _Psf, if, make m=m+1, change step (F1) then over to; Otherwise enter step (c3)

Process (G): according to following step iterative computation correcting image:

(G1) according to following formula iterative computation correcting image:

${\hat{f}}_{n,k+1}(x,y)=\{\left(\frac{g(x,y)}{{\hat{f}}_{n,k}(x,y)\⊗{\hat{h}}_{n,k}(x,y)}\right)\·{\hat{h}}_{n,k}(x,y)\}*{\hat{f}}_{n,k}(x,y)$

Wherein, g (x, y) the Pneumatic optical trial image that collects of expression, The correcting image of representing this iteration,

With

Represent correcting image and point spread function that last iteration calculates respectively,

(G2) judge that whether k is less than M _Obj, if, make k=k+1, change step (G1) then over to; Otherwise enter step (c3)

(c3) deterministic process (F) and (G) whether all finish, if, enter step (c4), otherwise wait for, all finish until process (F) with (G), enter step (c4) again;

(c4) make n=n+1, judge that whether n is less than N ₀, if change step (c2) then over to; Finish otherwise calculate.

Images