CN101276472B - Method and system for preprocessing sequence chart - Google Patents

Abstract

The invention provides a method and system for pre-processing sequence figures. The method comprises reading the number x of the figures of a first stage of sequence figures and a figure of a middle region with a sequence number s; calculating a correlation extent of the figure with the sequence number x and y figures in a second stage of sequence figures, determining the figure in the second stage of sequence figures, which is most correlated with the figure with the sequence number x, and determining the sequence number y of the figure in the second stage of sequence figures; if |x-y|<=delta, judging that the sequences of the two stage of sequence figures are same with each other; otherwise, judging that the sequences of the two stage of sequence figures are converse with each other and reversely-ordering one of the stages of sequence figures. Correspondingly, the system comprises an initializing module, a sampling module, a figure comparing module, a judging and processing module. The invention adopts a computer to judge whether the sequences of the two stages of sequence figures are same with each other, and re-order one of the stages of sequence figures when the sequences of the two stages of sequence figures are different so as to obtain two or more stages of sequence figures with same sequences, in this way, the invention has advantages of small error, high automatization extent and high efficiency.

Description

The method and system of preprocessing sequence chart

Technical field

The present invention relates to the pre-service of sequence chart, more specifically, relate to two phases or sequence chart of many phases are carried out pretreated method and system, described two phases or sequence chart of many phases are meant according to identical or opposite order carries out two series that image acquisition obtains or the picture of a plurality of series to related content.

Background technology

Sequence chart is meant ordinal relation, a series of pictures of related content is arranged.For example, according to carrying out Computerized chromatographic (being called for short CT) scanning from last (head) to the order of descending (pin) to the liver of human body, to obtain a series of CT figure, every width of cloth CT is corresponding with the cross section of liver, precedence relationship between each width of cloth CT figure is corresponding with the upper-lower position in each cross section of liver relation, therefore, this series of CT figure can be described as the CT sequence chart.

Medically, for the position of understanding focus, size and with relation of artery, vein etc., the sequence chart that the doctor can gather " region of interest " by MRI (nuclear magnetic resonance), CT modes such as (Computerized chromatographics).For example, relation for tissues such as artery, vein in the position, size and the focus that are well understood to the liver inner disease foci and the liver, the doctor gathers the CT picture of arteria hepatica phase and vena hepatica phase respectively according to the sequencing that contrast preparation flows in liver vessel, obtain arteria hepatica phase CT sequence chart and vena hepatica phase CT sequence chart.

In the CT scan owing to the operation and other etc. reason can cause the scanning sequency of above-mentioned two phase CT sequence chart inconsistent.For example, doctor's scanning liver artery has adopted the scanning sequency from last (head) to following (pin) during the phase, and the scanning vena hepatica has adopted during the phase from the scanning sequency of following (pin) to last (head), and so, resulting arteria hepatica phase CT sequence chart and resulting vena hepatica phase CT sequence chart are reversed in order.Because the order of this two phases CT sequence chart is inequality, cause to this two phases CT sequence chart carry out follow-up cutting apart, work such as reconstruction, information fusion are nonsensical, also are like this for the other types sequence chart.

Therefore, before two phases of " region of interest " or sequence chart of many phases being carried out follow-up cut apart, merging, reconstruction etc. handle, need carry out pre-service, guarantee that the order of this two phase or sequence chart of many phases is consistent this two phase or sequence chart of many phases.

At present, the artificial mode of general employing is carried out pre-service to sequence chart: judge artificially whether the order of two phase sequence chart is identical earlier, if find the reversed in order of two phase sequence chart, just artificially first phase sequence chart is wherein oppositely renamed, or artificially first phase sequence chart is wherein carried out sorting by reversals and the picture behind the sorting by reversals is stored as new first phase sequence chart separately.The artificial result who judges has judgement person's subjectivity, cause judged result can be along with judgement person's experience, ability is different and change to some extent, the judged result error is bigger, and, when the picture amount of sequence chart is big, this artificial preprocess method workload is big, and degree is low automatically, efficient is not high.

Summary of the invention

On the one hand, bigger, the inefficient defective of the method error that the present invention is directed to above-mentioned existing preprocessing sequence chart provides a kind of method of preprocessing sequence chart of robotization.The method of preprocessing sequence chart of the present invention may further comprise the steps: the picture number X that reads first phase sequence chart, read the picture number Y of second phase sequence chart, according to | the magnitude relationship of the X-Y| and the first preset value α is judged the validity of the described first phase and second phase sequence chart; If the described first phase and second phase sequence chart are effective, just read a secondary picture of non-zone line in the first phase sequence chart, the serial number of the picture that is read in first phase sequence chart is x; Calculate the degree of correlation of the included Y width of cloth picture of picture that described serial number is x and second phase sequence chart according to the difference of pixel value, determine in the second phase sequence chart with described serial number to be the maximally related picture of picture of x according to described degree of correlation, and write down the serial number y of this maximally related picture in described second phase sequence chart; Judge | the magnitude relationship of x-y| and the 3rd preset value δ, if | x-y|≤δ then the described first phase and second phase sequence chart are considered as in proper order identical; Otherwise, the described first phase and second phase sequence chart are considered as reversed in order and the picture of described first phase sequence chart or second phase sequence chart are carried out sorting by reversals.

Preferably, described difference according to pixel value is calculated in the step of described degree of correlation, comprising:

Calculate $R=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{S}^{i,j}(m,n)\&times;T(m,n)}{\sqrt{\left(\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[S^{i,j}(m,n)\right]}^2\right)}\sqrt{\left(\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[T(m,n)\right]}^2\right)}},$ Wherein, (m is that coordinate is (m, the pixel value of some n), S in the picture of x for described serial number n) to T ^{I, j}(m n) is subgraph S in certain width of cloth picture of second phase sequence chart ^{I, j}With described serial number be (m, n) pixel value of the point of Chong Heing of the picture T of x; I, j are S ^{I, j}The coordinate of top left corner pixel point in certain width of cloth picture of second phase sequence chart, described serial number is that length and the width value of the picture T of x is M; R is that described serial number is the related coefficient of described certain width of cloth picture of the picture of x and second phase sequence chart, and the degree of correlation of picture is directly proportional with the value of coefficient R.

Preferably, described difference according to pixel value is calculated in the step of described degree of correlation, with described serial number be the picture integral body of x as template figure, adopt template matching method to calculate described degree of correlation.

Preferably: in the step of the degree of correlation of calculating picture: in default rotation angle range, be that the picture of x is rotated earlier to described serial number, and be that the picture of x carries out convergent-divergent to described serial number in default zoom ranges, again picture behind rotation and the convergent-divergent and the Y width of cloth picture in the second phase sequence chart are carried out described template matches computing, and write down resulting maximum correlation coefficient R, rotation angle value and the zoom factor corresponding with this maximum correlation coefficient R; The method of described preprocessing sequence chart also comprises the X width of cloth picture in the described first phase sequence chart is rotated and convergent-divergent, the anglec of rotation equals the corresponding rotation angle value with described maximum correlation coefficient R, and the convergent-divergent multiple equals the corresponding zoom factor with described maximum correlation coefficient R.

Preferably, before calculating described degree of correlation, also comprise: with the serial number of first phase sequence chart is that the picture of x and the Y width of cloth picture in the described second phase sequence chart are converted to gray-scale map.

Preferably, use difference shadow method to calculate described degree of correlation, may further comprise the steps: calculate the picture that described serial number is x and the poor shadow of the Y width of cloth picture in the described second phase sequence chart successively, obtain Y poor shadow value, the poor shadow value of the degree of correlation of two width of cloth pictures and this two width of cloth picture is inversely proportional to.

Preferably, described picture in described first phase sequence chart or the second phase sequence chart is carried out after the sorting by reversals, also comprise: with the picture-storage behind the described sorting by reversals is third phase sequence chart.

Preferably, to equal 1%, the three preset value δ of X be 2% of X to the described first preset value α.

On the other hand, the present invention is directed to the defective that method error is big, efficient is not high of above-mentioned existing artificial preprocessing sequence chart, a kind of system of preprocessing sequence chart of robotization is provided.The system of preprocessing sequence chart of the present invention comprises initialization module, decimation blocks, picture comparison module, judgement and processing module; Described initialization module be used to read first phase sequence chart picture number X, read the picture number Y of second phase sequence chart, and according to | the magnitude relationship of the X-Y| and the first preset value α is judged the validity of the described first phase and second phase sequence chart; Described decimation blocks is used for reading a secondary picture of the non-zone line of first phase sequence chart, and the serial number of the picture that is read in first phase sequence chart is x; Described picture comparison module is used for calculating according to the difference of pixel value the degree of correlation of the included Y width of cloth picture of picture that described serial number is x and second phase sequence chart, determine in the second phase sequence chart with described serial number to be the maximally related picture of picture of x according to described degree of correlation, and write down the serial number y of this maximally related picture in described second phase sequence chart; Described judgement and processing module be used for judging | the magnitude relationship of x-y| and the 3rd preset value δ, if | x-y|≤δ then the described first phase and second phase sequence chart are considered as in proper order identical; Otherwise, the described first phase and second phase sequence chart are considered as reversed in order and the picture of described first phase sequence chart or second phase sequence chart are carried out sorting by reversals.

Preferably, described picture comparison module is used for:

Calculate $R=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{S}^{i,j}(m,n)\&times;T(m,n)}{\sqrt{\left(\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[S^{i,j}(m,n)\right]}^2\right)}\sqrt{\left(\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[T(m,n)\right]}^2\right)}},$ Wherein, (m is that coordinate is (m, the pixel value of some n), S in the picture of x for described serial number n) to T ^{I, j}(m n) is subgraph S in certain width of cloth picture of second phase sequence chart ^{I, j}With described serial number be (m, n) pixel value of the point of Chong Heing of the picture T of x; I, j are S ^{I, j}The coordinate of top left corner pixel point in certain width of cloth picture of second phase sequence chart, described serial number is that length and the width value of the picture T of x is M; R is that described serial number is the related coefficient of described certain width of cloth picture of the picture of x and second phase sequence chart, and the degree of correlation of picture is directly proportional with the value of coefficient R.

Preferably, described picture comparison module is used for: calculate the poor shadow of the Y width of cloth picture of picture that described serial number is x and described second phase sequence chart successively, obtain Y poor shadow value, the poor shadow value of the degree of correlation of two width of cloth pictures and this two width of cloth picture is inversely proportional to.

Preferably, also to be used for the picture-storage behind the described sorting by reversals be third phase sequence chart for described judgement and processing module.

Compare with the method for existing preprocessing sequence chart, the present invention adopts computing machine to judge two phases or the maximally related picture of sequence chart of many phases, and whether the order of judging sequence chart in view of the above is identical, when the order of two phase sequence chart is inequality,, finally obtain order identical two phases or sequence chart of many phases to the rearrangement of the picture in the first phase sequence chart wherein.The method and system of this preprocessing sequence chart provided by the invention does not rely on people's subjectivity, and the pre-service result is objective, error is little, and the automaticity height can be efficiently, apace two phases or sequence chart of many phases are adjusted into same sequence.

Description of drawings

Fig. 1 is the process flow diagram of preprocessing sequence chart of the present invention;

Fig. 2 is the system framework figure of preprocessing sequence chart of the present invention;

Fig. 3, Fig. 4, Fig. 5 be respectively template matching method first by than figure, second by than figure and template figure;

Fig. 6 is the synoptic diagram of template matching method;

Fig. 7, Fig. 8, Fig. 9 be respectively first in the difference shadow method by than figure, second by than figure, difference shadow picture;

Figure 10 and Figure 11 related coefficient curve map for obtaining in an alternative embodiment of the invention.

Embodiment

Fig. 1 is the process flow diagram of preprocessing sequence chart of the present invention.As shown in Figure 1, among the step S102, read the picture number X of first phase sequence chart, read the picture number Y of second phase sequence chart.Those skilled in the art should recognize, can read quantity of documents in certain sequence chart or certain file easily according to prior art, and this does not give unnecessary details.

Then, among the step S104, the magnitude relationship of X and Y relatively, this is in order to check the validity of this two phases sequence chart, more specifically, the picture number of two phase sequence chart want identical or situation about being more or less the same under, be only effective.Therefore, if | the X-Y|≤first preset value α, just illustrate that this two phases sequence chart is effective, flow process enters step S106, and the first preset value α can choose according to the requirement of precision, for example, high for accuracy requirement, α can get 0.1%～10% of X, and perhaps 0.5%～5% of X, perhaps 0.5%～2% or the like.If two phases | the X-Y|＜first preset value α, flow process enters step S105, reports an error to the user.This shows, the present invention also is applicable to and handles two phases or the sequence chart of many phases that picture number has certain difference applicable to handling identical two phases or the sequence chart of many phases (for example the arterial phase CT sequence chart at the same position of human body and venous phase CT sequence chart) of picture number.

Among the step S106, read a secondary picture of non-zone line in the first phase sequence chart.In present patent application, " picture of zone line " of described certain phase sequence chart is meant the picture in centre position in this phase sequence chart and the picture that is adjacent; And being exactly this, " picture of non-zone line " of certain phase sequence chart play other pictures outside " picture of zone line " in the sequence chart.According to accuracy requirement, picture adds up in the first phase sequence chart of X, the picture of non-zone line can be (X 49.5%) width of cloth picture and last (X 49.5%) width of cloth picture at first, or (X 45%) width of cloth picture at first and last (X 45%) width of cloth picture, or (X 30%) width of cloth picture at first and last (X 30%) width of cloth picture.If the serial number of the picture that is read among the step S106 in first phase sequence chart is x, hereinafter and in the accompanying drawing be that the picture of x is called " x width of cloth picture " with this serial number, obviously, i.e. x 〉=1 and x≤X.Because this x width of cloth picture is at the zone line (reason that should not read the picture that is in the sequence chart zone line will be set forth at further part integrating step S112) of first phase sequence chart, therefore, $1\&le;x\&le;\frac{X}{2}-\mathrm{\&beta;}$ Perhaps $\frac{X}{2}+\mathrm{\&beta;}\&le;x\&le;X,$ β is second preset value.As required, natural numbers such as β desirable 1,2,3, perhaps 1% of β value X, 2% etc.According to the programming language environment provide get random number functions (for example the ram () function of C language) can be easily from 1 to

Scope and

Select the x that meets the requirements to the scope of X, this does not give unnecessary details.

Then, among the step S108, calculate the degree of correlation of the included picture of picture that above-mentioned serial number is x and second phase sequence chart successively, degree of correlation between the picture can be judged by the value differences between the picture, the method that is suitable for comprises template matching method or difference shadow method etc., and template matching method will combine Fig. 3～Fig. 9 at further part with the related content of difference shadow method and set forth.Step S108 also comprises according to the degree of correlation that is calculated and determining in the second phase sequence chart and the maximally related picture of above-mentioned x width of cloth picture, and writes down the serial number y of this maximally related picture in this second phase sequence chart, obviously, and 1≤y≤Y.

For example, step S108 adopts template matching method to calculate successively after the degree of correlation of the included picture of above-mentioned x width of cloth picture and second phase sequence chart, can obtain Y related coefficient, the related coefficient of two width of cloth pictures is big more just represents that the degree of correlation of this two width of cloth picture or similarity degree are high more.Because a resulting Y related coefficient be with second phase sequence chart in Y width of cloth picture one to one, promptly, the maximum serial number of related coefficient in an above-mentioned Y related coefficient equals the serial number y of this maximally related picture in above-mentioned second phase sequence chart, therefore, when specific implementation, read the serial number y of this peaked related coefficient in an above-mentioned Y related coefficient and get final product.

Then, among the step S112, judge the relation of x and y, and carry out different processing with the relation of y according to x.

Those skilled in the art will recognize, because first phase sequence chart all is at identical or relevant object with second phase sequence chart, so if the order of this two phases sequence chart is identical, so, above-mentioned x should be identical with y, or approaching.For example, the picture number of supposing first phase sequence chart and second phase sequence chart all is 99 width of cloth (being equivalent to X=Y=99), if the order of this two phases sequence chart is identical, so, the 10th width of cloth picture (being equivalent to x=10) of first phase sequence chart should be the most relevant or the most similar with the 10th width of cloth picture (being equivalent to y=10) in the second phase sequence chart, and promptly the value of x should equal the value of y; If the sequence chart reversed in order of this two phase, so, the 10th width of cloth picture (being equivalent to x=10) of first phase sequence chart should be the most relevant or the most similar with the 90th width of cloth picture (being equivalent to y=90) in the second phase sequence chart, and promptly the value of x should equal the value of y.

Under general situation, above-mentioned conclusion can instead push away, for example, if the 10th width of cloth picture (being equivalent to x=10) of first phase sequence chart equate with the 10th secondary picture (being equivalent to y=10) of second phase sequence chart, illustrate that then this first phase sequence chart is identical with the order of second phase sequence chart.But, because no matter the order of two phase sequence chart is identical or opposite, the picture that mediates all is the most relevant or the most similar, therefore, can not be according to the 50th width of cloth picture of the 50th width of cloth picture of first phase sequence chart and second phase sequence chart the most relevant or the most similar be identical with regard to the order of judging this two phases sequence chart.So, when reading samples pictures (as the x width of cloth width of cloth sheet among the above-mentioned step S106), avoid reading the picture in sequence chart centre position.In view of the picture number of first phase sequence chart may be not equal to the picture number of second phase sequence chart, and the error of considering picture, therefore, x will satisfy $1\&le;x\&le;\frac{X}{2}-\mathrm{\&beta;}$ Perhaps $\frac{X}{2}+\mathrm{\&beta;}\&le;x\&le;X,$ Wherein β is a preset value.

Similarly, in view of the picture number of first phase sequence chart may be not equal to the picture number of second phase sequence chart, and consider error when picture is gathered, can think, if | x-y| just illustrates that less than certain threshold value above-mentioned first phase sequence chart is identical with the order of above-mentioned second phase sequence chart.Particularly, among the step S112, judge | the magnitude relationship of x-y| and the 3rd preset value δ, as the case may be, integers such as δ desirable 0,1,2,3, also according to circumstances 1% of value X, 2% etc.If | x-y|≤δ then with described first phase sequence chart and second phase sequence chart be considered as the order identical, enter step S114 then; Otherwise just described first phase sequence chart and second phase sequence chart are considered as reversed in order and enter step S113.

Among the step S113, first phase sequence chart or second phase sequence chart are carried out sorting by reversals.So-called sorting by reversals is adjusted into last width of cloth picture with the 1st width of cloth picture in the sequence chart exactly, the 2nd width of cloth picture is adjusted into the 2nd width of cloth picture reciprocal, by that analogy.When the picture amount is big, by the computing machine batch processing picture is carried out sorting by reversals, will be quicker than existing artificial sorting by reversals, efficiently, more accurate.

Obviously, after the execution of step S113, the order of resulting two phase sequence chart is identical, has finished pretreated target, can enter step S114, finishes pretreated process.

As an improvement project, other follow-up processing flow can also be continued in step S114 back.For example, if among the step S113 be the Y width of cloth picture in the second phase sequence chart is carried out sorting by reversals, so, can also be in subsequent step be third phase sequence chart with this Y width of cloth picture-storage behind the sorting by reversals, obviously the picture sequence in the third phase sequence chart be with first phase sequence chart identical, with sorting by reversals before the reversed in order of second phase sequence chart.Again for example, can this Y width of cloth picture behind the sorting by reversals be renamed, the sequence chart after this is renamed is identical with the order of first phase sequence chart.

Fig. 2 is the framework synoptic diagram of the system of preprocessing sequence chart of the present invention.As shown in Figure 2, the system of preprocessing sequence chart of the present invention comprises initialization module 1, decimation blocks 3, picture comparison module 5, judgement and processing module 7.Wherein, decimation blocks 1 is used to carry out the flow process of above-mentioned steps S102, step S104, specifically comprises the picture number X that reads first phase sequence chart, reads the picture number Y of second phase sequence chart, and the validity of checking this two phases sequence chart.If this two phases sequence chart is effectively, just forward treatment scheme to decimation blocks; Otherwise, to user report an error (refer step S105).

Decimation blocks 3 is used to carry out above-mentioned step S106, specifically comprises the x width of cloth picture that reads in this first phase sequence chart, and wherein, x is the serial number of picture in first phase sequence chart, $1\&le;x\&le;\frac{X}{2}-\mathrm{\&beta;}$ Perhaps $\frac{X}{2}+\mathrm{\&beta;}\&le;x\&le;X,$ β is second preset value.

Picture comparison module 5 is used to carry out the flow process of above-mentioned steps S108, specifically comprise the degree of correlation of calculating the included picture of above-mentioned x width of cloth picture and second phase sequence chart successively, determine in the second phase sequence chart and the maximally related picture of above-mentioned x width of cloth picture according to degree of correlation, and write down the serial number y of this maximally related picture in described second phase sequence chart.Calculating the process of the degree of correlation between the picture will set forth in conjunction with Fig. 3～Fig. 9 at further part.

Judge the flow process that is used to carry out above-mentioned steps S112 to S114 with processing module 7, specifically comprise judgement | the magnitude relationship of x-y| and the 3rd preset value δ, if | x-y|≤δ then described first phase sequence chart and second phase sequence chart are considered as in proper order identical; Otherwise just described first phase sequence chart and second phase sequence chart are considered as reversed in order, and to first phase sequence chart or second phase sequence chart are carried out sorting by reversals.

Similarly, can also in the system of preprocessing sequence chart of the present invention, increase other functional modules, for example rename module, memory module etc. again.But these variations do not depart from the scope and spirit of the present invention.

Set forth the method for the degree of correlation between the comparison picture that the present invention adopts below in conjunction with Fig. 3 to Fig. 9.

In one embodiment of the invention, the employing template matching method calculates the degree of correlation between the picture.The principle of so-called template matches is to judge the degree of correlation or the similarity of two pictures by the difference of two squares of the pixel value of two width of cloth images, and these two pictures are divided into and are called " by than figure " (perhaps " searched figure ") and " template figure " (perhaps " template ").Fig. 3 is the synoptic diagram of template matching method.As shown in Figure 3, the size of establishing template figure T is M * M, and establishing by the size than figure S is N * N.Template figure T is by on than figure S during translation, and the quilt under template figure T covers is called subgraph S than subgraph ^{I, j}, i, j are subgraph S ^{I, j}Than the coordinate among the figure S, this pixel is defined as reference point to top left corner pixel point at quilt.The span of i and j is among the figure: 1≤i, j≤N-M+1; Weigh template T and subgraph Si, the related function of the similarity degree of j can draw by following similarity measurement:

$D(i,j)=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{[S^{i,j}(m,n)-T(m,n)]}^2$ Equation (1)

Wherein, (m is that coordinate is (m, the pixel value of some n) (for example gray-scale value), S among the template figure n) to T ^{I, j}(m, n) be by than among the figure with template figure (m, the n) pixel value of the point of Chong Heing (for example gray-scale value), and D (i, j) representation template figure T with the quilt than subgraph Si, the difference of j (difference of two squares of pixel value).D (i, j) more little, illustrate template figure T with by than subgraph Si, the degree of correlation of j or similarity degree are high more.For example, when template figure T with by than subgraph Si, when j mated fully, (i j) obtained minimum value 0 to D.

Equation (1) above launching obtains:

$D(i,j)=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[S^{i,j}(m,n)\right]}^2-2\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{S}^{i,j}(m,n)\&times;T(m,n)+\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[T(m,n)\right]}^2$ Equation (2)

Can see that from equation (2) the 3rd on equation (2) the right is only relevant with template figure T, be one with (i, j) irrelevant constant.Order:

$R(i,j)=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{S}^{i,j}(m,n)\&times;T(m,n)}{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[S^{i,j}(m,n)\right]}^2}$ Equation (3)

Equation (3) is carried out normalization, obtains:

$R=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{S}^{i,j}(m,n)\&times;T(m,n)}{\sqrt{\left(\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[S^{i,j}(m,n)\right]}^2\right)}\sqrt{\left(\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\underset{n=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[T(m,n)\right]}^2\right)}}$ Equation (4)

Can see from equation (3), when template figure T with by than subgraph Si, when j is consistent, have R (i, j)=1, otherwise R (i, j)＜1.In like manner, when template figure T is more consistent than figure S with quilt, R=1 is arranged, otherwise R＜1.Obviously, R is big more, template T and just high more by degree of correlation or similarity degree than figure S.Therefore, R is defined as template figure T and the related coefficient of quilt than figure S.

Can see from equation (1), computing machine to template figure T with by than subgraph S ^{I, j}Carry out a template matches and will carry out M * M subtraction, M * M time square, M * M-1 sub-addition, it is inferior that a matching complete quilt will be repeated aforesaid operations (N-M+1) * (N-M+1) than figure S.In order to reduce the operation times of computing machine effectively, improve Computer Processing speed, can be according to actual conditions with template figure (for example x width of cloth picture in the first phase sequence chart) picture in its entirety as template figure T.In view of the picture size of a lot of sequence chart equates, therefore, as template figure, be feasible with the x width of cloth picture integral body in certain phase sequence chart, i.e. M=N.

As the case may be, can earlier sequence chart be converted to after the gray-scale map, carry out the calculating of picture degree of correlation again.For example, can earlier the CT sequence chart be converted to the gray-scale map of .BMP form, and then carry out the calculating of degree of correlation.

As mentioned above, template matching method can judge by than whether comprising template figure among the figure, and this template figure described by than the coordinate position among the figure.For example, as template figure, as by than figure, template matching method can be judged by than having comprised template figure among the figure with picture shown in Figure 4 with picture shown in Figure 6, and can determine this template figure at this quilt than the coordinate position among the figure.Again for example, as template figure, as by than figure, template matching method can be judged by than not comprised template figure among the figure with picture shown in Figure 5 with picture shown in Figure 6.Therefore, the present invention is applicable to other field, for example, respectively the different information (waters, hills, greening territory) of same geographic area are carried out can adopting preprocess method provided by the invention that image of many phases is carried out pre-service after the map collection according to identical or opposite order.

As a kind of improved plan, in the applying template matching method, can at first be rotated at default angular range inner formword figure, and in default zoom ranges, template figure is carried out convergent-divergent, and then to template figure with by being carried out the template matches computing than figure, and write down resulting maximum correlation coefficient R, rotation angle value and the zoom factor corresponding with this maximum correlation coefficient R.This improvement project can judge angle difference, difference in size between two width of cloth pictures, the angle difference value equals the corresponding rotation angle value with this maximum correlation coefficient R, the difference in size value equals the corresponding zoom factor with this maximum correlation coefficient R.Can make picture angle unanimity, the equal and opposite in direction of two phase sequence chart according to the rotation of this angle difference value, difference in size value, the convergent-divergent picture of first phase sequence chart wherein.Therefore, this improved plan can also be applicable to the pre-service of CT sequence chart medically, the splicing of map photo that satellite is taken etc.

Those skilled in the art should recognize, can also adopt difference shadow method to judge two degrees of correlation between the picture.Difference shadow method is to judge the degree of correlation of two width of cloth pictures or similarity degree according to the poor shadow value between two width of cloth pictures.For example, picture shown in Figure 7 as first by than figure, as second by than figure, first is differed from than figure, second picture shown in Figure 8 after the shadow value this than figure, just can remove first by than figure and second the quilt than the identical content among the figure, obtain poor shadow picture as shown in Figure 9.Obviously, two width of cloth difference shadow values are more little, and the degree of correlation or the similarity degree of this two width of cloth picture are high more.

Similarly, in the difference shadow method, also can in default zoom ranges, rotating range, be carried out convergent-divergent and rotation than figure earlier, be compared again one of them.

In conjunction with the accompanying drawings preprocess method of the present invention, system are set forth above.Though top elaboration is an example with two phase of pre-service sequence chart, the invention is not restricted to this situation, the present invention is applicable to handling sequence chart of many phases.

Figure 10 and Figure 11 related coefficient curve map for obtaining in an alternative embodiment of the invention.This embodiment is this 3 phase CT sequence chart of pre-service arteria hepatica phase, vena hepatica phase and vena portae hepatica phase, and each issue CT sequence chart is 396 width of cloth pictures.This 3 phase CT sequence chart is that the doctor carries out 3 CT scan according to the sequence of flow of contrast preparation to the liver position of same human body and obtains.The main processing procedure of present embodiment is as follows:

At first, read certain width of cloth picture in any first phase CT sequence chart, in certain experiment, what read is the 100th width of cloth CT picture of arteria hepatica phase.

Then, with the 100th width of cloth CT picture integral body as template figure, call the related coefficient that template matching algorithm calculates the picture of the 100th width of cloth CT picture of arteria hepatica phase and vena hepatica phase CT sequence chart, vena portae hepatica phase CT sequence chart, obtain two groups of related coefficients, respectively as Fig. 9 and shown in Figure 10.

Then, find out the maximal value of these two groups of relative coefficients respectively, and find out in view of the above in the vena hepatica phase CT sequence chart and the most similar residing serial number of picture of template figure (the 100th width of cloth CT picture of arteria hepatica phase), and in the vena portae hepatica phase CT sequence chart to the serial number of the most similar picture of template figure (the 100th width of cloth CT picture of arteria hepatica phase).In this test, the most similar to template figure in the vena hepatica phase CT sequence chart is the 296th width of cloth picture, and the most similar to template figure in vena portae hepatica phase CT sequence chart be the 100th width of cloth picture.In view of people's the physiological structure and the characteristic of CT scan, can tentatively obtain conclusion: the order of vena portae hepatica phase CT sequence chart and arteria hepatica phase CT sequence are sought for identical, and the order of vena hepatica phase CT sequence chart is sought for opposite with arteria hepatica phase CT sequence.

In order further to confirm above-mentioned conclusion, above-mentioned test can be carried out repeatedly.Following table 1 is to get this carries out template matches as template with the 096th width of cloth-0105 width of cloth CT figure of patient's arterial phase CT sequence chart result:

Table 1: template matches experimental result

96?	300?	96?
			97?	299?	97?
98?	298?	98?
			99?	297?	99?
100?	296?	100?
			101?	295?	101?

[0070]?

Template figure (arteria hepatica phase CT sequence chart) serial number	The serial number of the most similar picture in the vena hepatica phase CT sequence chart	The serial number of the most similar picture in the vena portae hepatica phase CT sequence chart
			102?	294?	102?
103?	293?	103?
			104?	292?	104?
105?	291?	105?

As above shown in the table, the conclusion that these 10 test for data obtain all is consistent, therefore, conclusion above further having confirmed, that is: the order of vena portae hepatica phase CT sequence chart and arteria hepatica phase CT sequence are sought for identically, and the order of vena hepatica phase CT sequence chart is sought for opposite with arteria hepatica phase CT sequence.

Above-described embodiment of the present invention does not constitute the qualification to protection domain of the present invention.Any modification of being done within the spirit and principles in the present invention, be equal to and replace and improvement etc., all should be included within the claim protection domain of the present invention.

Claims (8)

1. the method for a preprocessing sequence chart is characterized in that, may further comprise the steps:

Read the picture number X of first phase sequence chart, read the picture number Y of second phase sequence chart, according to | the magnitude relationship of the X-Y| and the first preset value α is judged the validity of the described first phase and second phase sequence chart;

If the described first phase and second phase sequence chart are effective, just read a secondary picture of non-zone line in the first phase sequence chart, the serial number of the picture that is read in first phase sequence chart is x;

Calculate the degree of correlation of the included Y width of cloth picture of picture that described serial number is x and second phase sequence chart according to the difference of pixel value, determine in the second phase sequence chart with described serial number to be the maximally related picture of picture of x according to described degree of correlation, and write down the serial number y of this maximally related picture in described second phase sequence chart;

Judge | the magnitude relationship of x-y| and the 3rd preset value δ, if | x-y|≤δ then the described first phase and second phase sequence chart are considered as in proper order identical; Otherwise, the described first phase and second phase sequence chart are considered as reversed in order and the picture of described first phase sequence chart or second phase sequence chart are carried out sorting by reversals.

2. the method for preprocessing sequence chart according to claim 1 is characterized in that, described difference according to pixel value is calculated in the step of described degree of correlation, comprising:

Calculate

Wherein,

T (m, n) for described serial number be in the picture of x coordinate for (m, the pixel value of some n),

S ^{I, j}(m n) is subgraph S in certain width of cloth picture of second phase sequence chart ^{I, j}With described serial number be (m, n) pixel value of the point of Chong Heing of the picture T of x; I, j are S ^{I, j}The coordinate of top left corner pixel point in certain width of cloth picture of described second phase sequence chart, described serial number are that length and the width value of the picture T of x is M;

R is that described serial number is the related coefficient of described certain width of cloth picture of the picture of x and second phase sequence chart, and the degree of correlation of picture is directly proportional with the value of coefficient R.

3. the method for preprocessing sequence chart according to claim 1, it is characterized in that, described difference according to pixel value is calculated in the step of described degree of correlation, with described serial number be the picture integral body of x as template figure, adopt template matching method to calculate described degree of correlation.

4. the method for preprocessing sequence chart according to claim 3 is characterized in that:

In the step of the degree of correlation of calculating picture: in default rotation angle range, be that the picture of x is rotated earlier to described serial number, and be that the picture of x carries out convergent-divergent to described serial number in default zoom ranges, again picture behind rotation and the convergent-divergent and the Y width of cloth picture in the second phase sequence chart are carried out described template matches computing, and write down resulting maximum correlation coefficient R, rotation angle value and the zoom factor corresponding with this maximum correlation coefficient R; And

The method of described preprocessing sequence chart also comprises the X width of cloth picture in the described first phase sequence chart is rotated and convergent-divergent, the anglec of rotation equals the corresponding rotation angle value with described maximum correlation coefficient R, and the convergent-divergent multiple equals the corresponding zoom factor with described maximum correlation coefficient R.

5. according to the method for any described preprocessing sequence chart in the claim 1 to 4, it is characterized in that, described picture in described first phase sequence chart or the second phase sequence chart is carried out after the sorting by reversals, also comprise: with the picture-storage behind the described sorting by reversals is third phase sequence chart.

6. the system of a preprocessing sequence chart is characterized in that, comprising:

Initialization module (1), be used to read first phase sequence chart picture number X, read the picture number Y of second phase sequence chart, and according to | the magnitude relationship of the X-Y| and the first preset value α is judged the validity of the described first phase and second phase sequence chart;

Decimation blocks (3) is used for reading a secondary picture of the non-zone line of first phase sequence chart, and the serial number of the picture that is read in first phase sequence chart is x;

Picture comparison module (5), be used for calculating the degree of correlation of the included Y width of cloth picture of picture that described serial number is x and second phase sequence chart according to the difference of pixel value, determine in the second phase sequence chart with described serial number to be the maximally related picture of picture of x according to described degree of correlation, and write down the serial number y of this maximally related picture in described second phase sequence chart;

Judge and processing module (7), be used for judging | the magnitude relationship of x-y| and the 3rd preset value δ, if | x-y|≤δ then the described first phase and second phase sequence chart are considered as in proper order identical; Otherwise, the described first phase and second phase sequence chart are considered as reversed in order and the picture of described first phase sequence chart or second phase sequence chart are carried out sorting by reversals.

7. the system of preprocessing sequence chart according to claim 6 is characterized in that, described picture comparison module (5) is used for:

Calculate

Wherein,

T (m, n) for described serial number be in the picture of x coordinate for (m, the pixel value of some n),

S ^{I, j}(m n) is S in certain width of cloth picture of second phase sequence chart ^{I, j}With described serial number be (m, n) pixel value of the point of Chong Heing of the picture T of x; I, j are S ^{I, j}Top left corner pixel point is at S ^{I, j}In coordinate, the M value for described serial number be length or the width value of the picture T of x;

R is that described serial number is the related coefficient of described certain width of cloth picture of the picture of x and second phase sequence chart, and the degree of correlation of picture is directly proportional with the value of coefficient R.

8. according to the system of claim 6 or 7 described preprocessing sequence charts, it is characterized in that it is third phase sequence chart that described judgement and processing module (7) also are used for the picture-storage behind the described sorting by reversals.

Images