CN105551040A - Method and system for automatically extracting tongue contour in NMR image sequence - Google Patents

Abstract

The present invention provides a method and system for automatically extracting a tongue contour in an NMR image sequence. The method comprises a step of obtaining a tongue contour initial edge point by using a multi-directional Sobel operator in a tongue movement area for an NMR image sequence, a step of establishing a tongue edge point mapping matrix, combining with a former frame of tongue contour position, and adjusting the mapping matrix, a step of searching an optimal edge point sequence in the adjusted mapping matrix, and obtaining the tongue contour with the aid of the quadratic spline curve fitting technology of the control point. According to the method and the system, the tongue contour can be automatically and accurately extracted from the NMR image sequence, the method and the system have the advantages that when a tongue and other organs are in contact, the method has good robustness, and the whole process is automatically completed without human interaction.

Description

Automatically the method and system of tongue position profile is extracted in nuclear-magnetism image sequence

Technical field

The present invention relates to IT trade technical field of image processing, relate to the method and system automatically extracting tongue position profile in nuclear-magnetism image sequence particularly.

Background technology

Nmr imaging technique, as a kind of medical science observation method of advanced safety, was able to widespread use in fields such as medical treatment, scientific research, criminal investigations in recent years.Median sagittal plane because of nuclear-magnetism image can provide comparatively complete speaker's vocal tract shape, and its cavity interior imaging effect is more clear compared with X-ray image, therefore this technology is usually used in vocal tract shape segmentation, the field such as tongue position motion analysis when the research of vowel articulation sound channel profile and pronunciation.In research work in the past, researchist needs the modeling having marked pronunciation contour extraction method by a large amount of craft usually.The Proctor of University of Southern California etc. propose a kind of automanual vocal organs contour extraction method, can have been marked the extraction work of vocal organs profile by a small amount of craft.But sometimes visually there is error clearly in the tongue position profile that the method is extracted, when especially touching other organ contours in the motion process of tongue position.For this situation, some researchers have done corresponding research work, but still need by a large amount of craft mark or revise.

Inventor finds: owing to there is a large amount of noise in nuclear-magnetism image, imaging resolution is lower, and when tongue position and other vocal organs (as maxilla, soft palate, throat etc.) come in contact, its contour edge becomes very fuzzy and even disappears, therefore, in nuclear-magnetism image sequence, very large challenge is also faced with to the contours extract work of tongue position.

Summary of the invention

The embodiment of the present invention provides a kind of method automatically extracting tongue position profile in nuclear-magnetism image sequence, with solve at least in part how without the need to man-machine interactively automatically to extract the technical matters of tongue position profile in nuclear-magnetism image sequence.In addition, a kind of system automatically extracting tongue position profile in nuclear-magnetism image sequence is also provided.

To achieve these goals, according to an aspect, following technical scheme is provided:

In nuclear-magnetism image sequence, automatically extract a method for tongue position profile, described method at least comprises:

In described nuclear-magnetism image, in moving region, tongue position, utilize multi-direction Sobel operator, extract tongue position profile initial edge points;

Based on described tongue position profile initial edge points, set up tongue position marginal point mapping matrix;

According to described tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence;

Using described tongue position optimal edge point sequence as reference mark, curve fitting algorithm is utilized to obtain tongue position profile.

According to another aspect, additionally provide a kind of system automatically extracting tongue position profile in nuclear-magnetism image sequence, described system at least comprises:

Extraction module, is configured in described nuclear-magnetism image, in moving region, tongue position, utilizes multi-direction Sobel operator, extracts tongue position profile initial edge points;

Matrix sets up module, is configured to, based on described tongue position profile initial edge points, set up tongue position marginal point mapping matrix;

Search module, is configured to according to described tongue position marginal point mapping matrix, and utilizes the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence;

Curve fitting module, to be configured to described tongue position optimal edge point sequence, as reference mark, utilize curve fitting algorithm to obtain tongue position profile.

Compared with prior art, technique scheme at least has following beneficial effect:

The embodiment of the present invention, by adopting multi-direction Sobel operator in nuclear-magnetism image, in moving region, tongue position, extracts tongue position profile initial edge points; Then based on tongue position profile initial edge points, tongue position marginal point mapping matrix is set up; According to tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence; Using tongue position optimal edge point sequence as reference mark, curve fitting algorithm is utilized to obtain tongue position profile.Thus, the embodiment of the present invention can extract tongue position profile more exactly automatically from nuclear-magnetism image sequence, and whole process is without the need to man-machine interactively.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet automatically extracting the method for tongue position profile in nuclear-magnetism image sequence according to an exemplary embodiment;

Fig. 2 a is moving region, the nuclear-magnetism image median sagittal plane tongue position schematic diagram according to an exemplary embodiment;

Fig. 2 b is the tongue position contour edge gradient direction schematic diagram according to an exemplary embodiment;

Fig. 3 a is the tongue position profile initial edge points schematic diagram according to an exemplary embodiment;

Fig. 3 b is that schematic diagram is distributed in the non-homogeneous sector according to an exemplary embodiment;

Fig. 4 a is the initial tongue position marginal point mapping matrix schematic diagram according to an exemplary embodiment;

Fig. 4 b is the former frame tongue position contour edge point position view according to an exemplary embodiment;

Fig. 4 c is the tongue position marginal point mapping matrix schematic diagram after the adjustment according to an exemplary embodiment;

Fig. 5 a is the tongue position optimal edge point sequence schematic diagram according to an exemplary embodiment;

Fig. 5 b is the schematic diagram of the tongue position optimal edge point sequence correspondence position in protokaryon magnetic image according to an exemplary embodiment;

Fig. 6 a is existing methodical experimental result schematic diagram;

Fig. 6 b is the experimental result schematic diagram of the embodiment of the present invention method according to an exemplary embodiment;

Fig. 7 is the structural representation automatically extracting the system of tongue position profile in nuclear-magnetism image sequence according to an exemplary embodiment.

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 more detail.

It should be noted that, in accompanying drawing or instructions describe, similar or identical part all uses identical figure number.And in the accompanying drawings, to simplify or convenient sign.Moreover the implementation not illustrating in accompanying drawing or describe is form known to a person of ordinary skill in the art in art.In addition, although herein can providing package containing the demonstration of the parameter of particular value, should be appreciated that, parameter without the need to definitely equaling corresponding value, but is similar to corresponding value in acceptable error margin or design constraint.

The embodiment of the present invention, in conjunction with traditional image processing method, for nuclear-magnetism image sequence, in the process of vocal organs contours extract and tracking, utilizes multi-direction Sobel to combine operator and obtains tongue position profile initial edge points; Based on tongue position profile initial edge points, set up tongue position marginal point mapping matrix, and in conjunction with former frame tongue position outline position, this mapping matrix is adjusted; Find optimal edge point sequence in mapping matrix after the adjustment, obtain tongue position profile by the Quadric spline curve fitting technique crossing reference mark.

In one exemplary embodiment of the present invention, provide a kind of method automatically extracting tongue position profile in nuclear-magnetism image sequence.As shown in Figure 1, the method comprising the steps of S100 is to step S106.

Step S100: in nuclear-magnetism image, in moving region, tongue position, utilizes multi-direction Sobel operator, extracts tongue position profile initial edge points.

In this step, multi-direction Sobel operator, it is discreteness difference operator, is mainly used as rim detection.Particularly, the approximate value of gray scale of arithmograph image brightness function is used for.Use this operator in any point of image, corresponding gray scale vector or its law vector will be produced.

Wherein, sequence is continued when nuclear-magnetism image sequence is preferably nuclear-magnetism image median sagittal plane.

In actual applications, according to the difference of tongue position contour edge gradient direction, different multi-direction Sobel operators can be selected, obtains tongue position profile initial edge points.

Fig. 2 a is moving region, the nuclear-magnetism image median sagittal plane tongue position schematic diagram according to an exemplary embodiment.Fig. 2 b is the tongue position contour edge gradient direction schematic diagram according to an exemplary embodiment.In figure 2b, be divided into two parts by white line, left-half is first half, and right half part is latter half.For tongue position first half, select, front upper, front, front lower place to Sobel operator carry out rim detection, using obtained marginal point as tongue position profile initial edge points.

Wherein, the process that obtains of multi-direction Sobel operator is specially:

If G _ffor Sobel gradient operator, original definition is as follows:

$G_F=\left(\begin{array}{ccc}{g}_{11}& g_{12}& g_{13}\\ g_{21}& g_{22}& g_{23}\\ g_{31}& g_{32}& g_{33}\end{array}\right)$

For the gradient of this four direction upper, front upper, front, front lower, to even things up, make the Grad of this four direction equal (be equal to n), then the gradient of this four direction is defined as follows:

$\left\{\begin{array}{c}(g_{31}+g_{32}+g_{33})-(g_{11}+g_{12}+g_{13})=n\\ (g_{23}+g_{32}+g_{33})-(g_{11}+g_{12}+g_{21})=n\\ (g_{13}+g_{23}+g_{33})-(g_{11}+g_{21}+g_{31})=n\\ (g_{12}+g_{13}+g_{23})-(g_{21}+g_{31}+g_{32})=n\end{array}\right.$

Solution is one group of equation with many unknowns above, obtains one of them feasible solution to be:

$G_{F1}=n\×\left(\begin{array}{ccc}-1& 1& -1\\ -1& 0& 1\\ 1& -1& 1\end{array}\right)$

For the ease of calculating, in above-mentioned formula, n gets 1, and the multi-direction Sobel operator obtained is as follows:

-1	1	-1
			-1	0	1 3 -->
1	-1	1

。

In like manner, for tongue position latter half, select, rear upper, rear, the back lower place to Sobel operator carry out the extraction of tongue position profile initial edge points.

Sobel gradient operator G _fa feasible solution be:

$G_{F2}=n\×\left(\begin{array}{ccc}-1& 1& -1\\ 1& 0& -1\\ 1& -1& 1\end{array}\right)$

For the ease of calculating, in above-mentioned formula, n gets 1, and corresponding multi-direction Sobel operator is as follows:

-1	1	-1
			1	0	-1
1	-1	1

。

The process that Sobel operator carries out edge extracting is prior art, does not repeat them here.

For the forward and backward two halves in tongue position, utilize tongue position profile initial edge points that corresponding multi-direction Sobel operator extraction arrives as shown in Figure 3 a.

Step S102: based on tongue position profile initial edge points, sets up tongue position marginal point mapping matrix.

Particularly, this step can be divided into:

Step S1022: with the central point of moving region, tongue position for the center of circle, is divided into N number of sector by the image of tongue position profile initial edge points; Wherein, described N gets positive integer.

In this step, with moving region, tongue position central point for the center of circle, tongue position profile initial edge dot image is divided into several sectors and to its number consecutively, uneven distribution can be carried out according to features such as tongue position different parts kinematic dexterities in these sectors, and Fig. 3 b provides a kind of non-homogeneous sector according to an exemplary embodiment and distributes schematic diagram.

Step S1024: in sector, from the center of circle, from the close-by examples to those far off chooses tongue position profile initial edge points the strongest on each camber line, equally spacedly as the edge intensity value computing of camber line in this sector.

In this step, in each sector, from the center of circle, from the close-by examples to those far off choose tongue position contour edge point the strongest on each section of camber line equally spacedly, as the edge intensity value computing of camber line corresponding in this sector.Wherein, the strongest tongue position contour edge point is on the camber line that forms with the point of center of circle same distance in sector, the point that brightness (gray-scale value) is the strongest.

Step S1026: according to edge intensity value computing, sets up the tongue position marginal point mapping matrix of N × D; Wherein, D is the ultimate range of tongue position profile initial edge points apart from the center of circle.

In this step, set up the tongue position marginal point mapping matrix K of a N × D, each of matrix K arranges corresponding sector in corresponding protokaryon magnetic image.Wherein, N is the total number of sector, and D is the ultimate range of tongue position profile initial edge points apart from the center of circle, the i-th row jth column element K in this matrix _ijrepresent that in i-th sector, the distance center of circle is the segmental arc edge intensity value computing of j.According to image pixel size and experiment demand, preferably get N=10, D=17.Fig. 4 a shows the initial tongue position marginal point mapping matrix schematic diagram according to an exemplary embodiment.

In the process setting up tongue position marginal point mapping matrix, in order to obtain better effect, can also adjust this matrix.

In an optional embodiment, nuclear-magnetism image sequence comprises the first frame nuclear-magnetism image and the second frame nuclear-magnetism image, and the second frame nuclear-magnetism image is after the first frame nuclear-magnetism image.

The method also comprises:

In the tongue position marginal point mapping matrix of the second frame nuclear-magnetism image, determine the correspondence position of the tongue position profile optimal edge point sequence of the first frame nuclear-magnetism image, and utilize the tongue position marginal point mapping matrix of following formula to the second frame nuclear-magnetism image to adjust:

$M_{ij}=K_{ij}\×\exp (-{\left(\frac{i-i_j}{{\mathrm{\σ}}_M}\right)}^2);$

Wherein, M _ijfor the i-th row jth column element in the tongue position marginal point mapping matrix after adjustment; K _ijfor the i-th row jth column element in the marginal point mapping matrix of tongue position; i _jfor the line position in the first frame tongue position profile optimal edge point sequence residing for jth column border point; σ _mfor adjustment variance parameter.

Particularly, due to adjacent two frame nuclear-magnetism images time closely, the tongue position contours extract of tongue position outline position to a rear frame of former frame has certain booster action.So, from the second frame nuclear-magnetism image, concerning each frame nuclear-magnetism image, according to the tongue position outline position in former frame nuclear-magnetism image, tongue position marginal point mapping matrix can be adjusted.Therefore, in the tongue position marginal point mapping matrix K of a rear frame nuclear-magnetism image, find the correspondence position (as shown in black point in Fig. 4 b) of the tongue position profile optimal edge point sequence of former frame nuclear-magnetism image, then utilize following formula to adjust tongue position marginal point mapping matrix K:

$M_{ij}=K_{ij}\×\exp (-{\left(\frac{i-i_j}{{\mathrm{\σ}}_M}\right)}^2)$

Wherein, M _ijfor the tongue position marginal point mapping matrix i-th row jth column element after adjustment; K _ijfor the i-th row jth column element in the marginal point mapping matrix of tongue position, represent that in i-th sector, the distance center of circle is the camber line edge intensity value computing of j; i _jfor the line position in the profile optimal edge point sequence of former frame tongue position residing for jth column border point; σ _mfor adjustment variance parameter.In the present embodiment, experimentally demand, preferably gets σ _m=5.Fig. 4 c is the tongue position marginal point mapping matrix schematic diagram after the adjustment according to an exemplary embodiment.

The embodiment of the present invention considers the relativeness of tongue position outline position between upper and lower frame, improves the accuracy of tongue position contours extract.

Step S104: according to tongue position marginal point mapping matrix, utilizes the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence.

Particularly, this step can also comprise:

Step S1042: according to tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, build the transition probability between the marginal point of tongue position.

Particularly, because the location comparison of adjacent sectors is close, namely there is certain restricting relation between adjacent key point position, therefore, when carrying out the search of optimal edge point sequence in tongue position marginal point mapping matrix M after the adjustment, introduce transition function:

$T_{hi}=\exp (-{\left(\frac{h-i}{{\mathrm{\σ}}_T}\right)}^2)$

Wherein, T _hifor in adjacent two row of mapping matrix M, residing line position is respectively the transition probability between the tongue position marginal point of h and i; σ _tfor transfer variance parameter.Experimentally demand, preferably gets σ _t=6.

Step S1044: according to transition probability, by solving the optimum solution of following formula, obtains tongue position optimal edge point sequence:

$\begin{array}{c}{r}^{*}=\underset{r}{\arg \max }P\left(r\right|M,{\mathrm{\σ}}_T)\\ =\underset{r}{\arg \max }{\Π}_{j=1}^N{T}_{r(j-1)r\left(j\right)}{M}_{r\left(j\right)j}\end{array}$

Wherein, r ^*for optimal edge point sequence; P (.) is probability function; R is possible marginal point sequence; R (j) is the element of jth in r, preferably, in order to symbol is unified, by T _{r (0) r (1)}be set to 1.Preferably, the embodiment of the present invention adopts Viterbi (Viterbi) searching algorithm to obtain tongue position optimal edge point sequence, as shown in Figure 5 a.This optimal edge point sequence is reliable tongue position contour edge point.

Step S106: using tongue position optimal edge point sequence as reference mark, utilizes curve fitting algorithm to obtain tongue position profile.

In this step, quafric curve or other SPL can be adopted to carry out matching.Preferably, Quadric spline curve fitting algorithm can be adopted.Can obtain so comparatively smoothly without the tongue position contour curve of corner angle.Wherein concrete fit procedure is prior art, does not repeat them here.

Fig. 5 b gives the correspondence position of tongue position optimal edge point sequence in protokaryon magnetic image, and using tongue position optimal edge point sequence as reference mark, utilizes Quadric spline curve fitting algorithm, obtains the matched curve by this reference mark, as final tongue position profile.

Fig. 6 a is existing methodical experimental result schematic diagram.Fig. 6 b is the experimental result schematic diagram of the embodiment of the present invention method according to an exemplary embodiment.Comparison diagram 6a and Fig. 6 b, find when tongue position and other vocal organs come in contact, method provided by the present invention also has good robustness.

In embodiments of the present invention the mode of each step according to above-mentioned precedence is described; it will be appreciated by those skilled in the art that; in order to realize the effect of the present embodiment; need not perform according to such order between different steps; it can perform simultaneously or execution order is put upside down, and these simply change all within protection scope of the present invention.

Based on the technical conceive identical with embodiment of the method, the embodiment of the present invention also provides a kind of system automatically extracting tongue position profile in nuclear-magnetism image sequence.

Following used term " module " can realize the software of predetermined function and/or the combination of hardware.Although the system described by following examples preferably realizes in the mode of software, hardware also or the realization of the combination of software and hardware also may be conceived.

As shown in Figure 7, this system 70 at least comprises: extraction module 72, matrix set up module 74, search module 76 and curve fitting module 78.Wherein, extraction module 72 is configured in described nuclear-magnetism image, in moving region, tongue position, utilizes multi-direction Sobel operator, extracts tongue position profile initial edge points.Matrix is set up module 74 and is configured to, based on described tongue position profile initial edge points, set up tongue position marginal point mapping matrix.Search module 76 is configured to according to described tongue position marginal point mapping matrix, and utilizes the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence.Curve fitting module 78 to be configured to described tongue position optimal edge point sequence, as reference mark, utilize curve fitting algorithm to obtain tongue position profile.

In an optional embodiment, extraction module is specifically configured to:

At the first half of tongue position, use, front upper, front, front lower place to Sobel operator extraction tongue position profile initial edge points;

At the latter half of tongue position, use, rear upper, rear, the back lower place to Sobel operator extraction tongue position profile initial edge points.

In an optional embodiment, matrix is set up module and is specifically configured to:

With the central point of moving region, tongue position for the center of circle, the image of tongue position profile initial edge points is divided into N number of sector; Wherein, N gets positive integer;

In sector, from the center of circle, from the close-by examples to those far off choose tongue position profile initial edge points the strongest on each camber line equally spacedly, as the edge intensity value computing of camber line in this sector;

According to edge intensity value computing, set up the tongue position marginal point mapping matrix of N × D; Wherein, D is the ultimate range of tongue position profile initial edge points apart from the center of circle.

In an optional embodiment, search module is specifically configured to:

According to tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, build the transition probability between the marginal point of tongue position:

$T_{hi}=\exp (-{\left(\frac{h-i}{{\mathrm{\σ}}_T}\right)}^2),$

Wherein, T _hifor in adjacent two row of tongue position marginal point mapping matrix, residing line position is respectively the transition probability between the tongue position marginal point of h and i; σ _tfor transfer variance parameter;

According to transition probability, by solving the optimum solution of following formula, obtain tongue position optimal edge point sequence:

$\begin{array}{c}{r}^{*}=\underset{r}{\arg \max }P\left(r\right|M,{\mathrm{\σ}}_T)\\ =\underset{r}{\arg \max }{\Π}_{j=1}^N{T}_{r(j-1)r\left(j\right)}{M}_{r\left(j\right)j}\end{array},$

Wherein, r ^*for optimal edge point sequence; P (.) is probability function; R is possible marginal point sequence; R (j) is the element of jth in r.

Said system embodiment may be used for performing said method embodiment, the technique effect of its know-why, the technical matters solved and generation is similar, person of ordinary skill in the field can be well understood to, for convenience and simplicity of description, the specific works process of the system of foregoing description, with reference to the corresponding process in preceding method embodiment, can not repeat them here.

It should be noted that: the system of tongue position profile of automatically extracting in nuclear-magnetism image sequence that above-described embodiment provides is when carrying out the extraction of tongue position profile, only be illustrated with the division of above-mentioned each functional module, in actual applications, can distribute to have been come by different functional modules as required and by above-mentioned functions, inner structure by system is divided into different functional modules, to complete all or part of function described above.

Respectively system and method embodiment of the present invention is described respectively above being to be noted that, but another embodiment also be can be applicable to the details that an embodiment describes.

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. in nuclear-magnetism image sequence, automatically extract a method for tongue position profile, it is characterized in that, described method at least comprises:

In described nuclear-magnetism image, in moving region, tongue position, utilize multi-direction Sobel operator, extract tongue position profile initial edge points;

Based on described tongue position profile initial edge points, set up tongue position marginal point mapping matrix;

According to described tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence;

Using described tongue position optimal edge point sequence as reference mark, curve fitting algorithm is utilized to obtain tongue position profile.

2. method according to claim 1, is characterized in that, described in described nuclear-magnetism image, in moving region, tongue position, utilizes multi-direction Sobel operator, extracts tongue position profile initial edge points, specifically comprises:

At the first half of described tongue position, use, front upper, front, front lower place to Sobel operator extraction described in tongue position profile initial edge points;

At the latter half of described tongue position, use, rear upper, rear, the back lower place to Sobel operator extraction described in tongue position profile initial edge points.

3. method according to claim 1, is characterized in that, described based on described tongue position profile initial edge points, sets up tongue position marginal point mapping matrix, specifically comprises:

With the central point of moving region, described tongue position for the center of circle, the image of described tongue position profile initial edge points is divided into N number of sector; Wherein, described N gets positive integer;

In described sector, from the described center of circle, from the close-by examples to those far off choose tongue position profile initial edge points the strongest on each camber line equally spacedly, as the edge intensity value computing of camber line described in this sector;

According to described edge intensity value computing, set up the tongue position marginal point mapping matrix of N × D; Wherein, described D is the ultimate range of tongue position profile initial edge points apart from the center of circle.

4. method according to claim 1, is characterized in that, described according to described tongue position marginal point mapping matrix, and utilizes the restricting relation of neighboring edge point position, and search tongue position optimal edge point sequence, specifically comprises:

According to described tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, build the transition probability between the marginal point of described tongue position:

$T_{hi}=\exp (-{\left(\frac{h-i}{{\mathrm{\σ}}_T}\right)}^2),$

Wherein, described T _hifor in adjacent two row of described tongue position marginal point mapping matrix, residing line position is respectively the transition probability between the tongue position marginal point of h and i; σ _tfor transfer variance parameter;

According to described transition probability, by solving the optimum solution of following formula, obtain tongue position optimal edge point sequence:

$\begin{array}{c}{r}^{*}=\underset{r}{\mathrm{argmax}}P\left(r\right|M,{\mathrm{\σ}}_T)\\ =\underset{r}{\mathrm{argmax}}{\mathrm{\Π}}_{j=1}^N{T}_{r(j-1)r\left(j\right)}{M}_{r\left(j\right)j}\end{array},$

Wherein, described r ^*for optimal edge point sequence; Described P (.) is probability function; Described r is possible marginal point sequence; Described r (j) is the element of jth in r.

5. method according to claim 1, is characterized in that, described curve fitting algorithm is Quadric spline curve fitting algorithm.

6. method according to claim 1, is characterized in that, described nuclear-magnetism image sequence comprises the first frame nuclear-magnetism image and the second frame nuclear-magnetism image, and described second frame nuclear-magnetism image is after described first frame nuclear-magnetism image;

Described method also comprises:

In the tongue position marginal point mapping matrix of described second frame nuclear-magnetism image, determine the correspondence position of the tongue position profile optimal edge point sequence of described first frame nuclear-magnetism image;

The described tongue position marginal point mapping matrix of following formula to described second frame nuclear-magnetism image is utilized to adjust:

$M_{ij}=K_{ij}\×\exp (-{\left(\frac{i-i_j}{{\mathrm{\σ}}_M}\right)}^2)$

Wherein, M _ijfor the i-th row jth column element in the tongue position marginal point mapping matrix after adjustment; K _ijfor the i-th row jth column element in the marginal point mapping matrix of tongue position; i _jfor the line position in the first frame tongue position profile optimal edge point sequence residing for jth column border point; σ _mfor adjustment variance parameter.

7. in nuclear-magnetism image sequence, automatically extract a system for tongue position profile, it is characterized in that, described system at least comprises:

Extraction module, is configured in described nuclear-magnetism image, in moving region, tongue position, utilizes multi-direction Sobel operator, extracts tongue position profile initial edge points;

Matrix sets up module, is configured to, based on described tongue position profile initial edge points, set up tongue position marginal point mapping matrix;

Search module, is configured to according to described tongue position marginal point mapping matrix, and utilizes the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence;

Curve fitting module, to be configured to described tongue position optimal edge point sequence, as reference mark, utilize curve fitting algorithm to obtain tongue position profile.

8. system according to claim 7, is characterized in that, described extraction module is specifically configured to:

At the first half of described tongue position, use, front upper, front, front lower place to Sobel operator extraction described in tongue position profile initial edge points;

At the latter half of described tongue position, use, rear upper, rear, the back lower place to Sobel operator extraction described in tongue position profile initial edge points.

9. method according to claim 7, is characterized in that, described matrix is set up module and is specifically configured to:

With the central point of moving region, described tongue position for the center of circle, the image of described tongue position profile initial edge points is divided into N number of sector; Wherein, described N gets positive integer;

In described sector, from the described center of circle, from the close-by examples to those far off choose tongue position profile initial edge points the strongest on each camber line equally spacedly, as the edge intensity value computing of camber line described in this sector;

According to described edge intensity value computing, set up the tongue position marginal point mapping matrix of N × D; Wherein, described D is the ultimate range of tongue position profile initial edge points apart from the center of circle.

10. system according to claim 7, is characterized in that, described search module is specifically configured to:

According to described tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, build the transition probability between the marginal point of described tongue position:

$T_{hi}=\exp (-{\left(\frac{h-i}{{\mathrm{\σ}}_T}\right)}^2),$

Wherein, described T _hifor in adjacent two row of described tongue position marginal point mapping matrix, residing line position is respectively the transition probability between the tongue position marginal point of h and i; σ _tfor transfer variance parameter;

According to described transition probability, by solving the optimum solution of following formula, obtain tongue position optimal edge point sequence:

$\begin{array}{c}{r}^{*}=\underset{r}{\mathrm{argmax}}P\left(r\right|M,{\mathrm{\σ}}_T)\\ =\underset{r}{\mathrm{argmax}}{\mathrm{\Π}}_{j=1}^N{T}_{r(j-1)r\left(j\right)}{M}_{r\left(j\right)j}\end{array},$

Wherein, described r ^*for optimal edge point sequence; Described P (.) is probability function; Described r is possible marginal point sequence; Described r (j) is the element of jth in r.