CN104504710A - Moore stripe recognition method and device for X-ray grating phase-contrast imaging - Google Patents

Abstract

The invention discloses a Moore stripe recognition method and device for X-ray grating phase-contrast imaging. The method includes the steps of 1, subjecting a Moore stripe image to be recognized to illumination unevenness correction; 2, filtering the Moore stripe image subjected to illumination unevenness correction; 3, binarizing the Moore stripe image filtered to obtain a binary image; 4, detailing the binary image to extract center lines of Moore stripes in the binary image; and 5, recognizing precision positions of the Moore stripes according to the extracted center lines of the Moore stripes. The Moore stripe recognition method and device has the advantages that calculation of the angle and direction of Moore stripes is automated, errors caused by subjective factors of an instrument user are avoided, instrument adjusting speed is greatly increased, and precision is well guaranteed.

Description

The recognition methods of Morie fringe and device in X ray raster phase contrast imaging

Technical field

The present invention relates to a kind of angle of stripe pattern and the automatic testing method in cycle, especially, relate to the angle of Moire fringe in X ray raster phase contrast imaging device alignment procedures and the automatic identifying method in cycle.

Background technology

In Germany scientist roentgen Late Cambrian X ray in 1895 so far more than 100 year, X ray, because of its extremely strong penetrating power, is widely used in the imaging field of object.Traditional x-ray imaging method is mainly based on the absorption of object to X ray, and the quality of the final one-tenth image contrast obtained depends on the size of interior of articles each several part to X ray absorption characteristic difference to a great extent.At medical domain, because human body soft tissue part is little to the absorption of X ray, this just means that the soft tissue lesions diagnosis that X ray absorption imaging method applies to human body has significant limitation.

From last century the nineties, along with the development of third generation Synchrotron Radiation, Hard X-Ray Phase-Contrast Imaging technology is arisen at the historic moment.Multiple X-ray phase contrast imaging technology has been had to be developed at present.Its mechanism briefly, make use of exactly X ray penetrate object after its phase place occur movement carry out imaging.To compare absorption-contrast imaging, the advantage of phase contrast imaging is, the X ray of same dosage penetrates soft tissue, it is much bigger that the change that phase shifts produces absorbs than transmitted intensity the change produced, therefore obtained radioscopic image contrast will be greatly improved, see list of references [1].

X ray grating stepping phase contrast imaging method is a kind of X-ray phase contrast method that development is comparatively ripe at present, because its polychrome that general X-ray production apparatus can be utilized to produce, incoherence light carry out imaging, is widely adopted at present, see list of references [2].The X ray raster phase contrast imaging method generally adopted now, be that the people such as Pfeiffer F proposed first in 2006, the method adopts the grating of three pieces of difference in functionalitys, achieves and complete phase contrast imaging on general X-ray production apparatus.In experiment, light source grating Main Function common X-ray source is divided into a series of mutual incoherent line source.Before object sample is positioned over phase grating, single X ray line source is partial coherence, can produce Tabo effect with phase grating, finally by the analysis grating before being positioned over detector, obtains phase place change information, see list of references [3].

The key link adopting the method to obtain phase contrast image is phase grating and the aligning analyzing grating, and the precision of aligning has obvious impact to the picture quality obtained.Alignment methods is: grating is analyzed in first adjustment, makes its grid stroke level, afterwards by adjustment phase grating, the Taibo of phase grating is overlapped just completely with analysis grating from imaging.Judge to aim at whether complete be according to phase grating Taibo from imaging with analyze the Moire fringe angle that formed of grating and the cycle judges.Known according to the relevant knowledge of Morie fringe, when Morie fringe is vertical and the cycle is equal and just can be similar to when being tending towards infinity and think that aligning completes.And in current X ray optical grating contrast imaging experiment, the angle of punctual Morie fringe and cycle are all judged according to naked eyes, its drawback is band a guy subjective consciousness, and precision cannot ensure and speed is comparatively slow, see list of references [4].

List of references:

[1]Chapman L D，Tomlinson W C，Johnston R E，Washburn D，Pisano E，Gmur N，Zhong Z，Menk R，Arfelli F，Sayers D 1997phys.med.biol.42 2015

[2]Atsushi MOMOSE，Recent Advances in X-ray Phase Imaging，Japanese Journal of Applied Physics，Vol.44，No.9A，2005，pp.6355-6367

[3]Franz Pfeiffer，TimmWeitkamp，Oliver Bunk，Christian David，Phaseretrieval and differentialphase-contrast imaging with low-brillianceX-raysources，nature physics VOL 2 APRIL 2006

[4]PavloBaturin，Mark Shafer，Optimization of grating-basedphase-contrast imaging setup，Medical Imaging 2014：Physics of MedicalImaging，Vol.9033，90334

Summary of the invention

In order to realize the accurately measuring and calculating fast in Morie fringe angle and cycle in X ray optical grating contrast imaging alignment procedures, thus improve instrument alignment precision and obtain high-quality phase contrast image.

The present invention proposes the recognition methods of the Morie fringe in a kind of X ray raster phase contrast imaging, it comprises:

Step 1: the even correction of uneven illumination is carried out to Morie fringe image to be identified;

Step 2: carry out filtering to through the even revised Morie fringe image of uneven illumination;

Step 3: carry out binaryzation to filtered Morie fringe image, obtains binary image;

Step 4: carry out refinement to described binary image, to extract the initial position message of each Morie fringe in binary image;

Step 5: according to the exact position of each Morie fringe of initial position message identification of extracted each Morie fringe.

The invention allows for the recognition device of the Morie fringe in a kind of X ray raster phase contrast imaging, it comprises:

Correcting module: the even correction of uneven illumination is carried out to Morie fringe image to be identified;

Filtration module: carry out filtering to through the even revised Morie fringe image of uneven illumination;

Binarization block: carry out binaryzation to filtered Morie fringe image, obtains binary image;

Refinement module: carry out refinement to described binary image, to extract the initial position message of each Morie fringe in binary image;

Identification module: according to the exact position of each Morie fringe of initial position message identification of extracted each Morie fringe.

Compared with prior art, scheme provided by the invention achieves the robotization that Morie fringe angle and direction calculates, the error that the subjective factor of the instrument user avoided causes, and instrument regulation speed improves greatly, and degree of accuracy have also been obtained good guarantee.

Accompanying drawing explanation

Fig. 1 is the formation schematic diagram of X ray grating stepping phase contrast imaging system;

Fig. 2 is several exemplary relative positions of two blocks of gratings and corresponding Moire fringe schematic diagram;

Fig. 3 is the process flow diagram of the recognition methods of Morie fringe image in the present invention in X ray raster phase contrast imaging;

Fig. 4 (a)-(e) is for realizing software interface and the treatment step schematic diagram of the recognition methods of the Morie fringe image in X ray raster phase contrast imaging of the present invention.

Embodiment

For making the object, technical solutions and advantages of the present invention clearly understand, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in further detail.

It is as follows that Morie fringe produces principle: produce Grating self-imaging by the known phase grating of Tabo effect at talbot distance, analysis grating is positioned over specific talbot distance by us, the striped of the two intersects formation Morie fringe, and receives acquisition original image by being positioned over the ccd detector analyzed after grating.

Fig. 1 shows the formation of X ray grating stepping phase contrast imaging system.As shown in Figure 1, system is followed successively by X-ray source from right to left, light source grating, test specimen, phase grating, analyzes grating and ccd detector.

Fig. 2 shows several exemplary relative positions and corresponding Moire fringe of analyzing grating and phase grating.As shown in Figure 2, the first row respectively schemes left side for analyzing grating schematic diagram, right side be the Taibo of phase grating from imaging, the second behavior they superpose the Moire fringe of rear formation.

As shown in Figure 3, the invention discloses the recognition methods of the Morie fringe image in a kind of X ray raster phase contrast imaging, it comprises:

Step 1: obtain stripe pattern, and the even correction of uneven illumination is carried out to stripe pattern; Wherein, described stripe pattern can be Morie fringe, also can be general stripe pattern;

X-ray source in actual experiment is pointolite, and the stripe pattern therefore obtained certainly exists the even situation of uneven illumination.Uneven illumination can make image quality decrease, and display effect variation the more important thing is the precision that can affect follow-up Morie fringe and calculate.Therefore to the original image obtained, the present invention will carry out the even correction of uneven illumination at first.

Feature for pointolite is known, picture centre brightness is maximum, each pixel brightness size is the function of this pixel to illumination central pixel point distance, decay gradually toward surrounding brightness, the square distance law of reciprocity of corresponding points light source irradiation degree and the square distance inverse ratio cosine law of pointolite irradiance.For this feature, the present invention proposes a kind of modification method of simple possible.

The square distance law of reciprocity of pointolite irradiance, namely the light intensity of postulated point light source is I _θ, the distance that pointolite irradiates center to detector is I, then irradiate center light illumination to be and the square distance inverse ratio cosine law of pointolite irradiance, i.e. postulated point light source irradiation position and shadow surface out of plumb, still the light intensity of postulated point light source is I _θ, pointolite is I to the distance of detector plane, and pointolite and point of irradiation normal direction angle are θ, then irradiation position illuminance is suppose that the gradation of image value matrix obtained is A=(a accordingly _ij) _{m × n}, illumination center pixel is a _pq, have here k is the conversion coefficient that illuminance arrives image intensity value, then to any position pixel gray-scale value be

$a_{\mathrm{ij}}=k\frac{I_{\mathrm{\θ}}}{I^{\′2}}\cos \mathrm{\θ}=k\frac{I_{\mathrm{\θ}}}{{(I/\cos \mathrm{\θ})}^2}\mathrm{COS\θ}=k\frac{I_{\mathrm{\θ}}}{I^2}\·{\cos }^3\mathrm{\θ}$

Wherein I ' is for light source is to the distance of current pixel point (i, j) to be revised, and therefore we can obtain the gray-scale value a of current pixel point (i, j) _ij=a _pqcos ³θ.Accordingly, for pixel original gray value a on the image obtained _ij, we revise its gray-scale value is a ' _ij=a _ij/ cos ³θ, wherein and I can calculate and records, if current pixel point (i, j) to be revised to the distance at illumination center is i, j are the pixel coordinate of current pixel point (i, j) on original image, and p, q are the pixel coordinate of illumination center pixel (p, q), namely $a_{\mathrm{ij}}^{\′}=a_{\mathrm{ij}}\·{(I^2/(I^2+{(i-p)}^2+{(j-q)}^2))}^{-\frac{3}{2}}.$

Therefore specify illumination centre coordinate (p, q) in calculating, measurement light source, to sensor distance I, just can complete the even correction of uneven illumination to the pixel (i, j) on image.

Step 2: filtering is carried out to the Morie fringe image through the even correction of uneven illumination.

Due to the restriction of actual environment condition, the factor impacts such as CCD self character, there is noise in the original image of acquisition unavoidably.The common noise of ccd image comprises spiced salt noise, pulse noise, gaussian noise etc., and in addition for the purpose of the present invention, the image of grid stroke is also noise information.Noise information can have a significant effect to successive image result, therefore needs to carry out filtering process to image.

Conventional image filtering method comprises filter in spatial domain and frequency domain filtering.Filter in spatial domain refers to and directly processes the pixel grey scale of image, and the grey value characteristics according to each pixel carries out filtering.Common filter in spatial domain method comprises histogram equalization, median filtering method, mean filter method etc.Filter in spatial domain method simple, intuitive, but filter effect is often not ideal enough, therefore preferential employing frequency domain filtering method in the present invention.

The usual way of frequency domain filtering method is: first image is carried out Fast Fourier Transform (FFT), chooses frequency domain filtering function and carries out filtering, filtered image is carried out Fourier inversion, obtain filtered image, thus filtering noise, improve picture quality.

Can form Moire fringe when phase grating and analysis grating interfere, suppose that coordinate points is expressed as (x, y) in X, Y plane, the cycle of two blocks of gratings is respectively d ₁and d ₂, make first block of grating and phase grating or analyze the grid line of grating parallel with Y-axis, second block of grating namely analyze the grid line of grating or phase grating and Y-axis into θ angle clockwise, assuming that the Fourier transform that two grating gaps are the penetration function of zero, first block of grating is f ₁(T), the Fourier transform of the penetration function of second block of grating is f ₂(T) the interference field light distribution that, they are formed in space is:

$\begin{array}{c}f\left(T\right)=f_1\left(T\right)\·f_2\left(T\right)=\frac{a_{01}{a}_{02}}{4}+\frac{a_{02}}{2}\×\underset{n=1}{\overset{\∞}{\mathrm{\Σ}}}{a}_n\cos \left(n\frac{2\mathrm{\π}}{d_1}x\right)+\frac{a_{01}}{2}\×\underset{m=1}{\overset{\∞}{\mathrm{\Σ}}}{a}_m\cos \left(m\frac{2\mathrm{\π}}{d_1}\right(x\cos \mathrm{\θ}\\ -y\sin \mathrm{\θ}\left)\right)+\left\{\underset{n=1}{\overset{\∞}{\mathrm{\Σ}}}{a}_n\cos (n-\frac{2\mathrm{\π}}{d_1}x)\right\}\×\left\{\underset{m=1}{\overset{\∞}{\mathrm{\Σ}}}{a}_m\cos \left(m\frac{2\mathrm{\π}}{d_2}(x\cos \mathrm{\θ}-y\sin \mathrm{\θ})\right)\right\}\end{array}$

Wherein, x, y are the coordinate of arbitrfary point on Morie fringe, and T is the Fourier transform cycle; a ₀₁, a _nbe the Fourier Transform Coefficients of first piece of grating penetration function, a ₀₂, a _mbe second piece of grating penetration function Fourier Transform Coefficients, analyze known, on the right of above formula equation, Section 1 is not containing phase factor, and it represents bias light, and Section 2 contains frequency content it comprises the structural information of first block of grating and phase grating, Section 3 contains frequency content it comprises the structural information that namely second block of grating analyze grating, and Section 4 contain two gratings with frequently and difference frequency component, belong to Morie fringe information.Therefore Section 1 and Section 4 are the useful information that we need to retain, and Section 2 and Section 3 need filtering.Certainly, the actual image obtained contains more complicated noise frequency composition, all should filtering in addition in filtering.

According to the above picture frequency information characteristics analyzed, present invention employs a kind of low pass junction filter of combining logical with band, low pass is used for leaching background component, is with general in leach Moire fringe information.Wave filter kind, can select ideal filter in the present invention, exponential filter, and Bart irrigates husband's wave filter etc., and wherein ideal filter form is comparatively simple and effect reaches requirement, therefore the method preferentially adopted for the present invention.Namely filter function is H (u, v), has

$H(u,v)=\left\{\begin{array}{c}1,u^2+v^2\≤r_1\\ 1,{(u-u_0)}^2+{(v-v_0)}^2\≤r_2\\ 0,\mathrm{others}\end{array}\right.$

Wherein r ₁for low-pass filter filter radius, r ₂for band-pass filter radius, u, v are the coordinate that filtering level is put, u ₀, v ₀for bandpass filtering centre coordinate.

Therefore hypothesis filter wavefront image information Fourier transform is F (u, v), filtered image Fourier transform should be G (u, v)=F (u, v) H (u, v), Fourier inversion is carried out to it, obtains filtered image.

Step 3: carry out binaryzation to having carried out filtered original image.

Binaryzation is as the term suggests be exactly image is divided into area-of-interest and region two parts of loseing interest in.In the present invention, image bright rays is area-of-interest, represented by described area-of-interest, and other regions represents with 0 after binaryzation with 1.Image binaryzation method based on Image Segmentation Theory is very ripe, and the method usually adopted now is threshold division, and threshold division algorithm mainly contains two steps: determine the segmentation threshold of needs and compared to divide pixel with pixel value by segmentation threshold.Threshold division method comprises local thresholding method, Global thresholding, manual threshold method etc., wherein determine that suitable threshold value is the key of Iamge Segmentation, and the extracting method of threshold value is also varied.Different threshold segmentation methods is suitable for the image of different characteristic, and the present invention needs image information to be processed relatively simple, and through pre-service, picture quality is also relatively good, and therefore most of threshold segmentation method is all applicable.

Step 4: refinement is carried out, to extract the initial position message of image Morie fringe to the area-of-interest of image after having carried out binaryzation.

Through above-mentioned process, image information simplifies greatly, but target of the present invention is the exact position obtaining Morie fringe, therefore needs the positional information extracting Moire fringe further, i.e. image thinning.Image thinning is exactly the trunk information extracting image, is extracted as example, obtains center point coordinate and the angle of Morie fringe exactly with Moire fringe.

Step 5: according to the exact position of the initial position message identification Morie fringe of extracted Morie fringe.

The striped initial position message extracted through thinning processing roughly meets precision needs, but in order to improve precision further, the present invention proposes a kind of iterative calculation method based on gray variance weight, to identify the exact position of Morie fringe.The method is specially:

Step 51: the initial position message obtaining current Morie fringe to be identified, if its central point is (x ₀, y ₀), length is l, and angle is θ ₀;

Step 52: setting iterative computation number of times, calculates angle of eccentricity α and line quantity n, with (x ₀, y ₀) centered by, l is length, and each tilt alpha draws n bar straight line to the left and right;

Step 53: establish the variance of pixel gray-scale value on each straight line to be followed successively by σ ₁, σ ₂, σ ₃..., σ _2n.Simultaneously from describing before, the straight line drift angle β of each line correspondences _ialso be not difficult to calculate, as ${\mathrm{\β}}_1={\mathrm{\θ}}_0+\frac{1}{n}\mathrm{\α},{\mathrm{\β}}_2={\mathrm{\θ}}_0+\frac{2}{n}\mathrm{\α},$ The like.

Step 54: according to the size of pixel gray variance on each bar straight line, gives weight to each bar linear angle of inclination $Q_i=1-\frac{{\mathrm{\σ}}_i}{{\mathrm{\Σ}}_{k=1}^{2n}{\mathrm{\σ}}_k},$ According to formula

${\mathrm{\θ}}_1=\underset{i=1}{\overset{2n}{\mathrm{\Σ}}}{\mathrm{\β}}_i\×Q_i=\underset{i=1}{\overset{2n}{\mathrm{\Σ}}}{\mathrm{\β}}_i\×(1-\frac{{\mathrm{\σ}}_i}{{\mathrm{\Σ}}_{k=1}^{2n}{\mathrm{\σ}}_k})$

The new angle of the Morie fringe current to be identified calculated;

Step 55: the iterative computation number of times set according to step 52, repeat step 52 to step 54 iterative process, finally obtain the exact position of current Morie fringe to be identified, the exact position of described current Morie fringe to be identified comprises center point coordinate and the angle of current Morie fringe to be identified, described center point coordinate is still the center point coordinate in the initial position message obtained in step 4, and angle is the angle after step 51-55 optimizes.

The precise position information of each Morie fringe can be obtained by repeated execution of steps 5, and the cycle of Morie fringe can be obtained according to the center point coordinate of each Morie fringe.

Step 6: instrument calibration judgement up to standard.

For each bar Morie fringe calculated, then with its central point for benchmark, subtract each other with the horizontal ordinate of adjacent center point, calculate their cycle.According to actual needs, to angle and cycle set allowable error, when instrument is adjusted to a certain state, if the angle calculated and cycle are all up to standard, just can think that instrument is aimed at and complete.

The recognition methods of above-mentioned Morie fringe of the present invention is by LabVIEW programming realization.

The present invention utilizes LabVIEW to write software, realizes above-mentioned algorithm (note: LabVIEW is upper should be provided with " vision and motion " module).All image processing process all carry out based on image intensity value, so " IMAQ ImageToArray " function first should be used to convert image to gray-scale value two-dimensional matrix.

Image irradiation retouch.Light source is input quantity to sensor distance and illumination center, and pixel position coordinates and pixel gray-scale value read in from image, described in before, bring formula into $a_{\mathrm{ij}}^{\′}=a_{\mathrm{ij}}\·{(I^2/({I^2+(i-p)}^2+{(j-q)}^2))}^{-\frac{3}{2}},$ Revised gradation of image matrix can be obtained, utilize " IMAQ ArrayToImage " function can obtain revised image.We with the 3-D view of " curved surface " function drawing image gray matrix, can observe correction effect intuitively in addition.

Image filtering this part used the Mixed-Programming Technology of LabVIEW and MATLAB.LabVIEW and MATLAB programmes the advantage having oneself original, the graphic programming mode of LabVIEW, developer can be allowed more to be absorbed in algorithm itself, the graphics process tool box of MATLAB is then integrated with many image processing function, bring great convenience to image procossing, therefore LabVIEW and MATLAB is combined the programming efficiency that programming can improve us greatly.The hybrid programming implementation method of LabVIEW and MATLAB is varied, the method adopted in the present invention for using " MATLAB script " node Calling MATLAB in LabVIEW, it is positioned at " mathematics > script and formula > script node > MATLAB script ", on the left of it, input is added after inserting this node, output is added on right side, in node, input MATLAB process code, can use, simple and convenient.

Therefore in this part of filtering, we use MATLAB script node, revised gradation of image matrix utilizes " fft2 " function in MATLAB to realize fast two-dimensional Fourier transform, from front panel input filter centre coordinate and filter radius size, completes filter function and builds.Image uses " ifft2 " to realize Fourier inversion after filtering afterwards, obtains filtered image.

Binarizing portion.Filtered image realizes binaryzation in this part, this part we be integrated with " local thresholding method " " automatic threshold method " and " manual threshold method " three kinds of methods.Wherein " local thresholding method " use " IMAQ Local Threshold " function realizes, and " automatic threshold method " use " IMAQAutoBThreshold 2 " function realizes, and " manual threshold method " use " IMAQ Threshold " function realizes.Concrete input parameter needs according to each function and picture situation is inputted by operator.

Image thinning is made up of refinement and striped initial position determination two parts in fact.The refinement of image uses " IMAQ Skeleton " function to realize, and stripe pattern just can obtain the train of thought of striped after utilizing this function process.But bending may appear in train of thought image, the problems such as burr is many, and its more specific location information also cannot be determined, therefore and then we use " IMAQ Find Straight Edges 2 " function to realize tentatively determining of fringe position.This function is originally for searching graph line edge, but we are applied to the image after refinement, by arranging appropriate parameter, it just can realize searching image cathetus, and return the function of linear position and angle, therefore may be used for tentatively determining of fringe position.

Fringe position accurate Calculation part mainly uses " IMAQ Line Profile " function, and this function can return pixel gray-scale value variance size on this straight line to straight line of specifying a certain on image.After obtaining variance, concrete calculating removes programming realization according to previously described.

The allowable error of striped exact position front panel input angle and variance again after determining, each fringe position and standard are contrasted, can judge angle and the cycle whether up to standard, and two boolean's lamps be set at front panel indicate angle and cycle result of determination respectively, represent up to standard so that lamp is bright.

Fig. 4 (a)-(e) shows the angle of Moire fringe and the automatic identifying method in cycle and software simulating in a kind of X ray optical grating contrast imaging alignment procedures proposed in the embodiment of the present invention, and the said method that the present invention proposes also is applicable to arbitrarily containing the fringe counting method of stripe pattern.Five steps is divided into complete when concrete enforcement.

As shown in Fig. 4 (a) step one, the first step is the even correction of uneven illumination.First measurement light source to detector distance and input.Read in original Morie fringe image, be converted into Three-Dimensional Gray figure and observe its gray feature, according to the result observed, set gradually horizontal ordinate and the ordinate at illumination center, operating software, the uneven illumination of image is even can be revised.Revised image and its Three-Dimensional Gray figure are also shown in software, help us to observe correction effect.We can preserve current correction setting by clicking " storage corrected parameter " button in addition.

As shown in Fig. 4 (b) step 2, second step is image filtering.The revised image of illumination is input, and input picture first carries out two-dimensional fast fourier transform, and the frequency domain figure picture after output transform is convenient to the frequency content present position that we observe needs.According to the image result observed, determine horizontal ordinate ordinate and the filter radius size of wave filter three filtering points successively, frequency domain figure picture after being provided with after operating software output filtering and its inverse transformation, can see the filter effect of image, optimum configurations can be finely tuned again.We can preserve current correction setting by clicking " storage filtering parameter " button in addition.

As shown in Fig. 4 (c) step 3, the 3rd step is image binaryzation.Image after input filter, selects a kind of binarization method.Here the method that we select is the Niblack method in local thresholding method, and design parameter is set to, and Niblack coefficient of deviation is 0.2, and calculation window size is that 32 pixels are multiplied by 32 pixels.In practical operation, we also can select other binarization method to obtain best binaryzation effect according to picture quality.

As shown in Fig. 4 (d) step 4, the 4th step is image thinning.Image after binaryzation, as input, can delimit the scope needing refinement and measurement in image.The option that straight line is searched is more, " refinement direction " determines according to stripe direction, can left and right directions, also can above-below direction, " kernel size " inputs minimum value, in line options, " number of lines " is as far as possible a little greatly to comprise all straight lines, and other parameter default values are just passable.Also other optimum configurations can be adjusted as required to obtain better effect in practical operation.The striped number detected exports in the lower right corner of software interface.

As shown in Fig. 4 (e) step 5, the 5th step is that striped calculates and qualification determination.The straight line information that step 4 finds such as fringe center dot position information, angle and line segment length information etc. are input to this step, the image obtained carries out line calculate according to straight line information after step 2 filtering.Angle and the number of line are inputted by software operation person, and after having calculated, the streak line finally obtained will be presented in filtered image with red line form.We export the cycle information between the angle information of every stripe and each striped, and by it with the display of the form of statistical graph in a program, conveniently observe.We arrange the allowable error δ (within angle ± δ ° for qualified) of angle and the allowable error Δ in cycle again (if the mean value in cycle is cycle be qualified within pixel), if the angle of each stripe and cycle qualifiedly will light boolean's pilot lamp.Finally gather each stripe information, the angle that decision-making system is overall again and the cycle whether up to standard, boolean's lamp is lighted if up to standard, two lamps all light i.e. system alignment and complete, if not, select suitable calibration steps according to stripe angle and the distribution situation in cycle, continue adjustment stop position, repeat each step work.

Above-described specific embodiment; object of the present invention, technical scheme and beneficial effect are further described; be understood that; the foregoing is only specific embodiments of the invention; be not limited to the present invention; within the spirit and principles in the present invention all, any amendment made, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. a recognition methods for the Morie fringe in X ray raster phase contrast imaging, it comprises:

Step 1: the even correction of uneven illumination is carried out to Morie fringe image to be identified;

Step 2: carry out filtering to through the even revised Morie fringe image of uneven illumination;

Step 3: carry out binaryzation to filtered Morie fringe image, obtains binary image;

Step 4: carry out refinement to described binary image, to extract the initial position message of each Morie fringe in binary image;

Step 5: according to the exact position of each Morie fringe of initial position message identification of extracted each Morie fringe.

2. the method for claim 1, wherein in described step 1 according to the square distance law of reciprocity of pointolite irradiance and the square distance inverse ratio cosine law of electric light source irradiance, the even correction of uneven illumination is carried out to described Morie fringe image.

3. method as claimed in claim 2, wherein, the optional position pixel gray-scale value in described step 1 on revised Morie fringe image calculates as follows:

$a_{\mathrm{ij}}^{\′}=a_{\mathrm{ij}}\·{(I^2/(I^2+{(i-p)}^2+{(j-q)}^2))}^{-\frac{3}{2}}$ Wherein, a _ijwith a ' _ijbe respectively any pixel (i on described Morie fringe image, j) with revised grey scale pixel value before revising, I is the distance of pointolite to Morie fringe image detector irradiation position, p and q is respectively the pixel coordinate of illumination center pixel (p, q).

4. the method as described in any one of claim 1-3, wherein, step 2 specifically comprises:

Step 21: carry out Fast Fourier Transform (FFT) through the even revised Morie fringe image of uneven illumination to described;

Step 22: choose frequency domain filtering function from the result after Fast Fourier Transform (FFT) and carry out filtering;

Step 23: the image after frequency domain filtering is carried out Fourier inversion, obtains filtered Morie fringe image.

5. method as claimed in claim 4, wherein, the frequency domain filtering function adopted in step 22 represents as follows:

$H(u,v)=\left\{\begin{array}{c}1,u^2+v^2\≤r_1\\ 1,{(u-u_0)}^2+{(v-v_0)}^2\≤r_2\\ 0,\mathrm{others}\end{array}\right.$

Wherein r ₁for low-pass filter filter radius, r ₂for band-pass filter radius, u, v are the coordinate that filtering level is put, u ₀, v ₀for bandpass filtering centre coordinate.

6. the method as described in claim 1-3,5 any one, wherein step 5 specifically comprises:

Step 51: the initial position message obtaining current Morie fringe to be identified, if the central point of current Morie fringe to be identified is (x ₀, y ₀), length is l, and angle is θ ₀;

Step 52: setup algorithm angle of eccentricity α and line quantity n, with (x ₀, y ₀) centered by, l is length, and each tilt alpha draws n bar straight line to the left and right;

Step 53: establish the variance of pixel on each straight line to be followed successively by σ ₁, σ ₂, σ ₃..., σ _2n, then the straight line drift angle β of each line correspondences is calculated _i;

Step 54: according to the size of pixel gray variance on each bar straight line, gives weight to each bar linear angle of inclination and according to formula

${\mathrm{\θ}}_1=\underset{i=1}{\overset{2n}{\mathrm{\Σ}}}{\mathrm{\β}}_i\×Q_i=\underset{i=1}{\overset{2n}{\mathrm{\Σ}}}{\mathrm{\β}}_i\×(1-\frac{{\mathrm{\σ}}_i}{{\mathrm{\Σ}}_{k=1}^{2n}{\mathrm{\σ}}_k})$

The new angle of the Morie fringe current to be identified calculated;

Step 55: the iterative computation number of times set according to step 52, repeats step 52 to step 54 iterative process, finally obtains the exact position of current Morie fringe to be identified.

7. the method for claim 1, wherein the exact position of current Morie fringe to be identified described in step 5 comprises central point and the angle of current Morie fringe to be identified.

8. method as claimed in claim 7, wherein, the method also comprises:

Step 6: the standard judging whether to reach instrument calibration according to the angle of each Morie fringe calculated and Morie fringe cycle.

9. a recognition device for the Morie fringe in X ray raster phase contrast imaging, it comprises:

Correcting module: the even correction of uneven illumination is carried out to Morie fringe image to be identified;

Filtration module: carry out filtering to through the even revised Morie fringe image of uneven illumination;

Binarization block: carry out binaryzation to filtered Morie fringe image, obtains binary image;

Refinement module: carry out refinement to described binary image, to extract the initial position message of each Morie fringe in binary image;

Identification module: according to the exact position of each Morie fringe of initial position message identification of extracted each Morie fringe.

10. device as claimed in claim 9, wherein, described identification module identifies the exact position of each Morie fringe as follows:

First the initial position message of current Morie fringe to be identified is obtained, if the central point of current Morie fringe to be identified is (x ₀, y ₀), length is l, and angle is θ ₀;

Setting iterative computation number of times, calculates angle of eccentricity α and line quantity n, with (x ₀, y ₀) centered by, l is length, and each tilt alpha draws n bar straight line to the left and right; If on each straight line, the variance of pixel is followed successively by, σ ₁, σ ₂, σ ₃..., σ _2n, then the straight line drift angle β of each line correspondences is calculated _i;

According to the size of pixel gray variance on each bar straight line, give weight to each bar linear angle of inclination $Q_i=1-\frac{{\mathrm{\σ}}_i}{{\mathrm{\Σ}}_{k=1}^{2n}{\mathrm{\σ}}_k},$ And according to formula

${\mathrm{\θ}}_1=\underset{i=1}{\overset{2n}{\mathrm{\Σ}}}{\mathrm{\β}}_i\×Q_i=\underset{i=1}{\overset{2n}{\mathrm{\Σ}}}{\mathrm{\β}}_i\×(1-\frac{{\mathrm{\σ}}_i}{{\mathrm{\Σ}}_{k=1}^{2n}{\mathrm{\σ}}_k})$

The new angle of the Morie fringe current to be identified calculated;

According to the iterative computation number of times of above-mentioned setting, repeat above-mentioned iterative process, finally obtain the exact position of current Morie fringe to be identified.