CN105551040B - The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence - Google Patents

The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence Download PDF

Info

Publication number
CN105551040B
CN105551040B CN201510929281.3A CN201510929281A CN105551040B CN 105551040 B CN105551040 B CN 105551040B CN 201510929281 A CN201510929281 A CN 201510929281A CN 105551040 B CN105551040 B CN 105551040B
Authority
CN
China
Prior art keywords
mrow
tongue position
msub
tongue
mapping matrix
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201510929281.3A
Other languages
Chinese (zh)
Other versions
CN105551040A (en
Inventor
陶建华
张大伟
杨明浩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Automation of Chinese Academy of Science
Original Assignee
Institute of Automation of Chinese Academy of Science
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Institute of Automation of Chinese Academy of Science filed Critical Institute of Automation of Chinese Academy of Science
Priority to CN201510929281.3A priority Critical patent/CN105551040B/en
Publication of CN105551040A publication Critical patent/CN105551040A/en
Application granted granted Critical
Publication of CN105551040B publication Critical patent/CN105551040B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Quality & Reliability (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a kind of method and system that tongue position profile is automatically extracted in nuclear-magnetism image sequence.Wherein this method includes：For nuclear-magnetism image sequence, in the moving region of tongue position, tongue position profile initial edge points are obtained using multi-direction Sobel operators；Tongue position marginal point mapping matrix is established, and combines former frame tongue position outline position, the mapping matrix is adjusted；Optimal edge point sequence is found in mapping matrix after the adjustment, tongue position profile is obtained by the Quadric spline curve fitting technique for crossing control point.The present invention can accurately extract tongue position profile from nuclear-magnetism image sequence automatically, it is advantageous that when tongue position and other vocal organs are in contact, this method also has preferable robustness, and whole process is automatically performed, without man-machine interactively.

Description

The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence
Technical field
The present invention relates to IT trade technical field of image processing, more particularly to automatic in nuclear-magnetism image sequence Extract the method and system of tongue position profile.
Background technology
Nmr imaging technique is as a kind of advanced safe medical science observation method, in recent years in medical treatment, scientific research, criminal investigation Extensive use is able to Deng field.Because the median sagittal plane of nuclear-magnetism image can provide more complete speaker's vocal tract shape, and Its cavity interior imaging effect becomes apparent from compared with X-ray image, therefore the technology is usually used in vocal tract shape segmentation, vowel articulation sound channel The field such as tongue position motion analysis when profile is studied and pronounced.In conventional research work, researcher is usually required by big The manual mark of amount completes the modeling of pronunciation contour extraction method.Proctor of University of Southern California etc. proposes a kind of semi-automatic Vocal organs contour extraction method, can lead to it is too small amount of by hand mark complete vocal organs profile extraction work.However, Visually there is apparent error sometimes in the tongue position profile that this method is extracted, is especially touched in the motion process of tongue position In the case of other organ contours.For this case, some researchers have done corresponding research work, but still need by big The manual mark of amount or amendment.
Inventor has found：Because, there is a large amount of noises, imaging resolution is relatively low in nuclear-magnetism image, and when tongue position and other When vocal organs (such as maxilla, soft palate, throat) are in contact, its contour edge becomes very fuzzy or even disappeared, therefore, Very big challenge is also faced to the work of tongue position contours extract in nuclear-magnetism image sequence.
The content of the invention
The embodiment of the present invention provides a kind of method that tongue position profile is automatically extracted in nuclear-magnetism image sequence, with least partly Ground solve how without man-machine interactively to automatically extract the technical problem of tongue position profile in nuclear-magnetism image sequence.In addition, also A kind of system that tongue position profile is automatically extracted in nuclear-magnetism image sequence is provided.
To achieve these goals, according on one side, there is provided following technical scheme：
A kind of method that tongue position profile is automatically extracted in nuclear-magnetism image sequence, methods described comprise at least：
In the nuclear-magnetism image, in the moving region of tongue position, using multi-direction Sobel operators, at the beginning of the profile of extraction tongue position Beginning marginal point；
Based on the tongue position profile initial edge points, tongue position marginal point mapping matrix is established；
According to the tongue position marginal point mapping matrix, and using the restricting relation of neighboring edge point position, search tongue position is most Excellent edge point sequence；
Using the tongue position optimal edge point sequence as control point, tongue position profile is obtained using curve fitting algorithm.
According on the other hand, a kind of system that tongue position profile is automatically extracted in nuclear-magnetism image sequence, institute are additionally provided The system of stating comprises at least：
Extraction module, it is configured as in the nuclear-magnetism image, in the moving region of tongue position, is calculated using multi-direction Sobel Son, extraction tongue position profile initial edge points；
Matrix establishes module, is configured as being based on the tongue position profile initial edge points, establishes tongue position marginal point mapping square Battle array；
Search module, it is configured as according to the tongue position marginal point mapping matrix, and utilizes the system of neighboring edge point position About relation, search tongue position optimal edge point sequence；
Curve fitting module, it is configured as, using the tongue position optimal edge point sequence as control point, utilizing curve matching Algorithm obtains tongue position profile.
Compared with prior art, above-mentioned technical proposal at least has the advantages that：
The embodiment of the present invention in nuclear-magnetism image, in the moving region of tongue position, is extracted by using multi-direction Sobel operators Tongue position profile initial edge points；Tongue position profile initial edge points are then based on, establish tongue position marginal point mapping matrix；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；By tongue position most Excellent edge point sequence obtains tongue position profile as control point using curve fitting algorithm.Thus, the embodiment of the present invention can be automatic Tongue position profile is accurately extracted from nuclear-magnetism image sequence, and whole process is without man-machine interactively.
Brief description of the drawings
Fig. 1 is the method that tongue position profile is automatically extracted in nuclear-magnetism image sequence according to an exemplary embodiment Schematic flow sheet；
Fig. 2 a are the nuclear-magnetism image median sagittal plane tongue position moving region schematic diagram according to an exemplary embodiment；
Fig. 2 b are the tongue position contour edge gradient direction schematic diagram according to an exemplary embodiment；
Fig. 3 a are the tongue position profile initial edge points schematic diagram according to an exemplary embodiment；
Fig. 3 b are that schematic diagram is distributed in the non-homogeneous sector according to an exemplary embodiment；
Fig. 4 a are the initial tongue position marginal point mapping matrix schematic diagram according to an exemplary embodiment；
Fig. 4 b are the former frame tongue position contour edge point position view according to an exemplary embodiment；
Fig. 4 c are the tongue position marginal point mapping matrix schematic diagram after the adjustment according to an exemplary embodiment；
Fig. 5 a are the tongue position optimal edge point sequence schematic diagram according to an exemplary embodiment；
Fig. 5 b are that the tongue position optimal edge point sequence according to an exemplary embodiment corresponds to position in protokaryon magnetic image The schematic diagram put；
Fig. 6 a are the experimental result schematic diagram of existing method；
Fig. 6 b are the experimental result schematic diagram of the present invention method according to an exemplary embodiment；
Fig. 7 is the system that tongue position profile is automatically extracted in nuclear-magnetism image sequence according to an exemplary embodiment Structural representation.
Embodiment
For the object, technical solutions and advantages of the present invention are more clearly understood, below in conjunction with specific embodiment, and reference Accompanying drawing, the present invention is described in more detail.
It should be noted that in accompanying drawing or specification description, similar or identical part all uses identical figure number.And In the accompanying drawings, with simplified or convenient sign.Furthermore the implementation for not illustrating or describing in accompanying drawing, it is art In form known to a person of ordinary skill in the art.In addition, though the demonstration of the parameter comprising particular value can be provided herein, it is to be understood that Parameter is similar to be worth accordingly without being definitely equal to corresponding value in acceptable error margin or design constraint.
The embodiment of the present invention combines traditional image processing method, for nuclear-magnetism image sequence, is carried in vocal organs profile During taking with tracking, tongue position profile initial edge points are obtained using multi-direction Sobel joint operators；Based at the beginning of the profile of tongue position Beginning marginal point, tongue position marginal point mapping matrix is established, and combine former frame tongue position outline position, the mapping matrix is adjusted It is whole；Optimal edge point sequence is found in mapping matrix after the adjustment, by the Quadric spline curve fitting technique for crossing control point Obtain tongue position profile.
In one exemplary embodiment of the present invention, there is provided one kind automatically extracts tongue position wheel in nuclear-magnetism image sequence Wide method.As shown in figure 1, the method comprising the steps of S100 to step S106.
Step S100：In nuclear-magnetism image, in the moving region of tongue position, multi-direction Sobel operators, extraction tongue position wheel are utilized Wide initial edge points.
In this step, multi-direction Sobel operators, it is discreteness difference operator, is mainly used as rim detection.Specifically Ground, for the approximation of the gray scale of computing brightness of image function.This operator is used in any point of image, it will is produced corresponding Gray scale vector or its law vector.
Wherein, continuous sequence when nuclear-magnetism image sequence is preferably nuclear-magnetism image median sagittal plane.
In actual applications, can be according to the difference of tongue position contour edge gradient direction, from different multi-direction Sobel Operator, obtain tongue position profile initial edge points.
Fig. 2 a are the nuclear-magnetism image median sagittal plane tongue position moving region schematic diagram according to an exemplary embodiment.Figure 2b is the tongue position contour edge gradient direction schematic diagram according to an exemplary embodiment.In figure 2b, it is divided into two by white line Individual part, left-half are first half, and right half part is latter half.For tongue position first half, in selection, preceding The Sobel operators in upper, preceding, front lower direction carry out rim detection, using resulting marginal point as tongue position profile initial edge points.
Wherein, the process that obtains of multi-direction Sobel operators is specially：
If GFIt is as follows for Sobel gradient operators, original definition：
For the gradient of this upper, preceding upper, preceding, front lower four direction, to even things up, the Grad phase of this four direction is made Deng (being equal to n), then the gradient of this four direction is defined as follows：
Solution one group of equation with many unknowns above, obtaining one of feasible solution is：
For the ease of calculating, in above-mentioned formula, n takes 1, and obtained multi-direction Sobel operators are as follows：
-1 1 -1
-1 0 1
1 -1 1
Similarly, for tongue position latter half, in selection, rear upper, the rear, back lower place to Sobel operators carry out tongue position profile The extraction of initial edge points.
Sobel gradient operators GFA feasible solution be：
For the ease of calculating, in above-mentioned formula, n takes 1, and corresponding multi-direction Sobel operators are as follows：
-1 1 -1
1 0 -1
1 -1 1
The process that Sobel operators carry out edge extracting is prior art, be will not be repeated here.
For forward and backward two half part in tongue position, the tongue position profile initial edge arrived using corresponding multi-direction Sobel operator extractions Point is as shown in Figure 3 a.
Step S102：Based on tongue position profile initial edge points, tongue position marginal point mapping matrix is established.
Specifically, the step can be divided into：
Step S1022：Using the central point of tongue position moving region as the center of circle, the image of tongue position profile initial edge points is divided into N number of sector；Wherein, the N takes positive integer.
In this step, using tongue position moving region central point as the center of circle, if tongue position profile initial edge dot image is divided into Dry sector simultaneously can carry out uneven to its number consecutively, these sectors according to the features such as tongue position different parts kinematic dexterity Distribution, Fig. 3 b provide a kind of non-homogeneous sector distribution schematic diagram according to an exemplary embodiment.
Step S1024：In sector, from the center of circle, tongue most strong on each camber line is from the close-by examples to those far off equally spacedly chosen Position profile initial edge points, the edge intensity value computing as camber line in the sector.
In this step, in each sector, from the center of circle, from the close-by examples to those far off equally spacedly choose on each section of camber line most Strong tongue position contour edge point, the edge intensity value computing as corresponding camber line in the sector.Wherein, most strong tongue position contour edge point It is a most strong point of brightness (gray value) on the camber line formed in sector with the point of center of circle same distance.
Step S1026：According to edge intensity value computing, N × D tongue position marginal point mapping matrix is established；Wherein, D takes turns for tongue position Ultimate range of the wide initial edge points away from the center of circle.
In this step, N × D tongue position marginal point mapping matrix K is established, each row of matrix K correspond to protokaryon magnetic Corresponding sector in image.Wherein, N is the total number of sector, and D is ultimate range of the tongue position profile initial edge points away from the center of circle, I-th row jth column element K in the matrixijRepresent in i-th of sector away from the segmental arc edge intensity value computing that the center of circle is j.According to image pixel Size and experiment demand, preferably take N=10, D=17.Fig. 4 a show the initial tongue position according to an exemplary embodiment Marginal point mapping matrix schematic diagram.
During tongue position marginal point mapping matrix is established, in order to obtain more preferable effect, the matrix can also be entered Row adjustment.
In an optional embodiment, nuclear-magnetism image sequence includes the first frame nuclear-magnetism image and the second frame nuclear-magnetism image, Second frame nuclear-magnetism image is after the first frame nuclear-magnetism image.
This method also includes：
In the tongue position marginal point mapping matrix of the second frame nuclear-magnetism image, the tongue position profile of the first frame nuclear-magnetism image is determined most The correspondence position of excellent edge point sequence, and the tongue position marginal point mapping matrix of the second frame nuclear-magnetism image is carried out using below equation Adjustment：
Wherein, MijFor the i-th row jth column element in the tongue position marginal point mapping matrix after adjustment；KijFor tongue position marginal point I-th row jth column element in mapping matrix；ijFor residing for jth column border point in the first frame tongue position profile optimal edge point sequence Line position；σMTo adjust variance parameter.
Specifically, because the time of adjacent two frames nuclear-magnetism image is very close, the tongue position outline position of former frame is to latter The tongue position contours extract of frame has certain booster action.So since the second frame nuclear-magnetism image, each frame nuclear-magnetism image is come Say, tongue position marginal point mapping matrix can be adjusted according to the tongue position outline position in former frame nuclear-magnetism image.Therefore, In the tongue position marginal point mapping matrix K of a later frame nuclear-magnetism image, the tongue position profile optimal edge of former frame nuclear-magnetism image is found The correspondence position (as shown in black dot in Fig. 4 b) of point sequence, then using below equation to tongue position marginal point mapping matrix K It is adjusted：
Wherein, MijFor tongue position marginal point mapping matrix the i-th row jth column element after adjustment；KijMapped for tongue position marginal point I-th row jth column element in matrix, represent in i-th of sector away from the camber line edge intensity value computing that the center of circle is j；ijFor former frame tongue position Line position in profile optimal edge point sequence residing for jth column border point；σMTo adjust variance parameter.In the present embodiment, according to reality Demand is tested, preferably takes σM=5.Fig. 4 c are the tongue position marginal point mapping matrix after the adjustment according to an exemplary embodiment Schematic diagram.
The embodiment of the present invention considers the relativeness of tongue position outline position between upper and lower frame, improves tongue position profile and carries The accuracy taken.
Step S104：According to tongue position marginal point mapping matrix, the restricting relation of neighboring edge point position, search tongue position are utilized Optimal edge point sequence.
Specifically, the step can also include：
Step S1042：According to tongue position marginal point mapping matrix, and utilize the restricting relation of neighboring edge point position, structure Transition probability between the marginal point of tongue position.
Specifically, it is located closer to due to adjacent sectors, i.e., certain restriction is present between adjacent key point position Relation, when therefore, in tongue position marginal point mapping matrix M after the adjustment carrying out the search of optimal edge point sequence, introduce transfer Probability function：
Wherein, ThiIt is respectively the transfer between h and i tongue position marginal point for residing line position in adjacent two row of mapping matrix M Probability；σTTo shift variance parameter.According to experiment demand, σ is preferably takenT=6.
Step S1044：According to transition probability, by solving the optimal solution of below equation, tongue position optimal edge point sequence is obtained Row：
Wherein, r*For optimal edge point sequence；P () is probability function；R is possible edge point sequence；R (j) is in r J-th of element, it is preferable that in order to which symbol is unified, by Tr(0)r(1)It is set to 1.Preferably, the embodiment of the present invention uses Viterbi (Viterbi) searching algorithm obtains tongue position optimal edge point sequence, as shown in Figure 5 a.The optimal edge point sequence is as reliable Tongue position contour edge point.
Step S106：Using tongue position optimal edge point sequence as control point, tongue position profile is obtained using curve fitting algorithm.
In this step, conic section or other SPLs can be used to be fitted.Preferably, can use secondary Spline curve fitting algorithm.The more smooth tongue position contour curve without corner angle can so be obtained.Wherein specific fit procedure For prior art, will not be repeated here.
Fig. 5 b give correspondence position of the tongue position optimal edge point sequence in protokaryon magnetic image, and by tongue position optimal edge Point sequence, using Quadric spline curve fitting algorithm, obtains the matched curve by the control point, as final as control point Tongue position profile.
Fig. 6 a are the experimental result schematic diagram of existing method.Fig. 6 b are that the present invention according to an exemplary embodiment is real Apply the experimental result schematic diagram of a method.Comparison diagram 6a and Fig. 6 b, find when tongue position and other vocal organs are in contact, this The there is provided method of invention also has preferable robustness.
Each step is described in the way of above-mentioned precedence in embodiments of the present invention, art technology Personnel are appreciated that to realize the effect of the present embodiment, are performed between different steps not necessarily in such order, it can Overturned with execution or execution order simultaneously, these simply change all within protection scope of the present invention.
Based on embodiment of the method identical technical concept, the embodiment of the present invention also provide one kind in nuclear-magnetism image sequence The system for automatically extracting tongue position profile.
Term " module " used below can realize the combination of the software and/or hardware of predetermined function.It is although following System described by embodiment preferably realized in a manner of software, still, hardware also or software and hardware combination Realize to be also what may be contemplated.
As shown in fig. 7, the system 70 comprises at least：Extraction module 72, matrix establish module 74, search module 76 and curve Fitting module 78.Wherein, extraction module 72 is configured as in the nuclear-magnetism image, in the moving region of tongue position, using multi-party To Sobel operators, extraction tongue position profile initial edge points.Matrix establish module 74 be configured as based on tongue position profile it is initial Marginal point, establish tongue position marginal point mapping matrix.Search module 76 is configured as according to the tongue position marginal point mapping matrix, and Utilize the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence.Curve fitting module 78 be configured as by The tongue position optimal edge point sequence obtains tongue position profile as control point using curve fitting algorithm.
In an optional embodiment, extraction module is specifically configured to：
First half in tongue position, use, the Sobel operator extraction tongues position profile initial edge in preceding upper, preceding, front lower direction Point；
Latter half in tongue position, use, afterwards upper, the rear, back lower place to Sobel operator extraction tongues position profile initial edge Point.
In an optional embodiment, matrix is established module and is specifically configured to：
Using the central point of tongue position moving region as the center of circle, the image of tongue position profile initial edge points is divided into N number of sector；Its In, N takes positive integer；
In sector, from the center of circle, it is initial from the close-by examples to those far off equally spacedly to choose tongue position profile most strong on each camber line Marginal point, the edge intensity value computing as camber line in the sector；
According to edge intensity value computing, N × D tongue position marginal point mapping matrix is established；Wherein, D is tongue position profile initial edge Ultimate range of the point away 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, structure tongue position marginal point Between transition probability：
Wherein, ThiFor tongue position marginal point mapping matrix it is adjacent two row in residing line position be respectively h and i tongue position marginal point it Between transition probability；σTTo shift variance parameter；
According to transition probability, by solving the optimal solution of below equation, tongue position optimal edge point sequence is obtained：
Wherein, r*For optimal edge point sequence；P () is probability function；R is possible edge point sequence；R (j) is in r J-th of element.
Said system embodiment can be used for performing above method embodiment, its technical principle, the technical problem solved And caused technique effect is similar, person of ordinary skill in the field can be understood that, convenience and letter for description It is clean, the specific work process of the system of foregoing description, the corresponding process in preceding method embodiment is may be referred to, it is no longer superfluous herein State.
It should be noted that：The system that tongue position profile is automatically extracted in nuclear-magnetism image sequence of above-described embodiment offer exists When carrying out the extraction of tongue position profile, only carried out with the division of above-mentioned each functional module for example, in actual applications, Ke Yigen Above-mentioned function distribution is completed by different functional modules according to needs, i.e., the internal structure of system is divided into different work( Energy module, to complete all or part of function described above.
It should be noted that the system and method embodiment of the present invention is described respectively respectively above, but to one The details of individual embodiment description can also be applied to another embodiment.
Particular embodiments described above, the purpose of the present invention, technical scheme and beneficial effect are carried out further in detail Describe in detail it is bright, should be understood that the foregoing is only the present invention specific embodiment, be not intended to limit the invention, it is all Within the spirit and principles in the present invention, any modification, equivalent substitution and improvements done etc., it should be included in the guarantor of the present invention Within the scope of shield.

Claims (10)

  1. A kind of 1. method that tongue position profile is automatically extracted in nuclear-magnetism image sequence, it is characterised in that methods described comprises at least：
    In the nuclear-magnetism image, in the moving region of tongue position, multi-direction Sobel operators, extraction tongue position profile initial edge are utilized Edge point；
    Based on the tongue position profile initial edge points, tongue position marginal point mapping matrix is established；
    According to the tongue position marginal point mapping matrix, and using the restricting relation of neighboring edge point position, search for the optimal side in tongue position Edge point sequence；
    Using the tongue position optimal edge point sequence as control point, tongue position profile is obtained using curve fitting algorithm.
  2. 2. according to the method for claim 1, it is characterised in that it is described in the nuclear-magnetism image, in tongue position moving region It is interior, using multi-direction Sobel operators, extraction tongue position profile initial edge points, specifically include：
    First half in the tongue position, use, tongue position profile described in the Sobel operator extractions in preceding upper, preceding, front lower direction it is initial Marginal point；
    Latter half in the tongue position, use, afterwards upper, the rear, back lower place to Sobel operator extractions described in tongue position profile it is initial Marginal point.
  3. 3. according to the method for claim 1, it is characterised in that it is described to be based on the tongue position profile initial edge points, establish Tongue position marginal point mapping matrix, is specifically included：
    Using the central point of tongue position moving region as the center of circle, the image of the tongue position profile initial edge points is divided into N number of fan Area；Wherein, the N takes positive integer；
    In the sector, from the center of circle, tongue position profile most strong on each camber line is from the close-by examples to those far off equally spacedly chosen Initial edge points, the edge intensity value computing as the camber line in the sector；
    According to the edge intensity value computing, N × D tongue position marginal point mapping matrix is established；Wherein, the D is that tongue position profile is initial Ultimate range of the marginal point away from the center of circle.
  4. 4. according to the method for claim 1, it is characterised in that it is described according to the tongue position marginal point mapping matrix, and profit With the restricting relation of neighboring edge point position, search tongue position optimal edge point sequence, specifically include：
    According to the tongue position marginal point mapping matrix, and using the restricting relation of neighboring edge point position, build tongue position side Transition probability between edge point：
    <mrow> <msub> <mi>T</mi> <mrow> <mi>h</mi> <mi>i</mi> </mrow> </msub> <mo>=</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <msup> <mrow> <mo>(</mo> <mfrac> <mrow> <mi>h</mi> <mo>-</mo> <mi>i</mi> </mrow> <msub> <mi>&amp;sigma;</mi> <mi>T</mi> </msub> </mfrac> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>)</mo> </mrow> <mo>,</mo> </mrow>
    Wherein, the ThiIt is respectively h and i tongue position edge for residing line position in adjacent two row of the tongue position marginal point mapping matrix Transition probability between point；σTTo shift variance parameter；
    According to the transition probability, by solving the optimal solution of below equation, tongue position optimal edge point sequence is obtained：
    <mrow> <mtable> <mtr> <mtd> <mrow> <msup> <mi>r</mi> <mo>*</mo> </msup> <mo>=</mo> <munder> <mi>argmax</mi> <mi>r</mi> </munder> <mi>P</mi> <mrow> <mo>(</mo> <mi>r</mi> <mo>|</mo> <mi>M</mi> <mo>,</mo> <msub> <mi>&amp;sigma;</mi> <mi>T</mi> </msub> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mo>=</mo> <munder> <mi>argmax</mi> <mi>r</mi> </munder> <msubsup> <mi>&amp;Pi;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </msubsup> <msub> <mi>T</mi> <mrow> <mi>r</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> <mi>r</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </msub> <msub> <mi>M</mi> <mrow> <mi>r</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>)</mo> </mrow> <mi>j</mi> </mrow> </msub> </mrow> </mtd> </mtr> </mtable> <mo>,</mo> </mrow>
    Wherein, the r*For optimal edge point sequence；The P () is probability function；The r is possible edge point sequence；Institute R (j) is stated as j-th of element in r；The M represents the tongue position marginal point mapping matrix after adjustment；The Tr(j-1)r(j)Described in expression Residing line position is respectively the transfer between r (j-1) and r (j) tongue position marginal point in adjacent two row of tongue position marginal point mapping matrix Probability；The Mr(j)jRepresent r (j) row jth column elements in the tongue position marginal point mapping matrix after adjustment.
  5. 5. according to the method for claim 1, it is characterised in that the curve fitting algorithm is that Quadric spline curve fitting is calculated Method.
  6. 6. according to the method for claim 1, it is characterised in that the nuclear-magnetism image sequence include the first frame nuclear-magnetism image and Second frame nuclear-magnetism image, the second frame nuclear-magnetism image is after the first frame nuclear-magnetism image；
    Methods described also includes：
    In the tongue position marginal point mapping matrix of the second frame nuclear-magnetism image, the tongue position wheel of the first frame nuclear-magnetism image is determined The correspondence position of wide optimal edge point sequence；
    The tongue position marginal point mapping matrix of the second frame nuclear-magnetism image is adjusted using below equation：
    <mrow> <msub> <mi>M</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>=</mo> <msub> <mi>K</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>&amp;times;</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <msup> <mrow> <mo>(</mo> <mfrac> <mrow> <mi>i</mi> <mo>-</mo> <msub> <mi>i</mi> <mi>j</mi> </msub> </mrow> <msub> <mi>&amp;sigma;</mi> <mi>M</mi> </msub> </mfrac> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>)</mo> </mrow> </mrow>
    Wherein, MijFor the i-th row jth column element in the tongue position marginal point mapping matrix after adjustment；KijMapped for tongue position marginal point I-th row jth column element in matrix；ijFor the line position residing for jth column border point in the first frame tongue position profile optimal edge point sequence； σMTo adjust variance parameter.
  7. 7. a kind of system that tongue position profile is automatically extracted in nuclear-magnetism image sequence, it is characterised in that the system comprises at least：
    Extraction module, it is configured as in the nuclear-magnetism image, in the moving region of tongue position, using multi-direction Sobel operators, carries Take tongue position profile initial edge points；
    Matrix establishes module, is configured as being based on the tongue position profile initial edge points, establishes tongue position marginal point mapping matrix；
    Search module, it is configured as according to the tongue position marginal point mapping matrix, and is closed using the restriction of neighboring edge point position System, search tongue position optimal edge point sequence；
    Curve fitting module, it is configured as, using the tongue position optimal edge point sequence as control point, utilizing curve fitting algorithm Obtain tongue position profile.
  8. 8. system according to claim 7, it is characterised in that the extraction module is specifically configured to：
    First half in the tongue position, use, tongue position profile described in the Sobel operator extractions in preceding upper, preceding, front lower direction it is initial Marginal point；
    Latter half in the tongue position, use, afterwards upper, the rear, back lower place to Sobel operator extractions described in tongue position profile it is initial Marginal point.
  9. 9. system according to claim 7, it is characterised in that the matrix is established module and is specifically configured to：
    Using the central point of tongue position moving region as the center of circle, the image of the tongue position profile initial edge points is divided into N number of fan Area；Wherein, the N takes positive integer；
    In the sector, from the center of circle, tongue position profile most strong on each camber line is from the close-by examples to those far off equally spacedly chosen Initial edge points, the edge intensity value computing as the camber line in the sector；
    According to the edge intensity value computing, N × D tongue position marginal point mapping matrix is established；Wherein, the D is that tongue position profile is initial Ultimate range of the marginal point away from the center of circle.
  10. 10. system according to claim 7, it is characterised in that the search module is specifically configured to：
    According to the tongue position marginal point mapping matrix, and using the restricting relation of neighboring edge point position, build tongue position side Transition probability between edge point：
    <mrow> <msub> <mi>T</mi> <mrow> <mi>h</mi> <mi>i</mi> </mrow> </msub> <mo>=</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <msup> <mrow> <mo>(</mo> <mfrac> <mrow> <mi>h</mi> <mo>-</mo> <mi>i</mi> </mrow> <msub> <mi>&amp;sigma;</mi> <mi>T</mi> </msub> </mfrac> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>)</mo> </mrow> <mo>,</mo> </mrow>
    Wherein, the ThiIt is respectively h and i tongue position edge for residing line position in adjacent two row of the tongue position marginal point mapping matrix Transition probability between point；σTTo shift variance parameter；
    According to the transition probability, by solving the optimal solution of below equation, tongue position optimal edge point sequence is obtained：
    <mrow> <mtable> <mtr> <mtd> <mrow> <msup> <mi>r</mi> <mo>*</mo> </msup> <mo>=</mo> <munder> <mi>argmax</mi> <mi>r</mi> </munder> <mi>P</mi> <mrow> <mo>(</mo> <mi>r</mi> <mo>|</mo> <mi>M</mi> <mo>,</mo> <msub> <mi>&amp;sigma;</mi> <mi>T</mi> </msub> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mo>=</mo> <munder> <mi>argmax</mi> <mi>r</mi> </munder> <msubsup> <mi>&amp;Pi;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </msubsup> <msub> <mi>T</mi> <mrow> <mi>r</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> <mi>r</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </msub> <msub> <mi>M</mi> <mrow> <mi>r</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>)</mo> </mrow> <mi>j</mi> </mrow> </msub> </mrow> </mtd> </mtr> </mtable> <mo>,</mo> </mrow>
    Wherein, the r*For optimal edge point sequence；The P () is probability function；The r is possible edge point sequence；Institute R (j) is stated as j-th of element in r；The M represents the tongue position marginal point mapping matrix after adjustment；The Tr(j-1)r(j)Described in expression Residing line position is respectively the transfer between r (j-1) and r (j) tongue position marginal point in adjacent two row of tongue position marginal point mapping matrix Probability；The Mr(j)jRepresent r (j) row jth column elements in the tongue position marginal point mapping matrix after adjustment.
CN201510929281.3A 2015-12-15 2015-12-15 The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence Active CN105551040B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510929281.3A CN105551040B (en) 2015-12-15 2015-12-15 The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510929281.3A CN105551040B (en) 2015-12-15 2015-12-15 The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence

Publications (2)

Publication Number Publication Date
CN105551040A CN105551040A (en) 2016-05-04
CN105551040B true CN105551040B (en) 2018-03-09

Family

ID=55830216

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510929281.3A Active CN105551040B (en) 2015-12-15 2015-12-15 The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence

Country Status (1)

Country Link
CN (1) CN105551040B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110638523B (en) * 2019-08-12 2021-01-05 杭州英库医疗科技有限公司 Method for automatically cutting edges based on circular track curve profile
CN113052866B (en) * 2021-03-29 2022-11-18 天津大学 Ultrasonic image tongue contour extraction method based on local binary fitting model

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102750549A (en) * 2012-06-12 2012-10-24 天津大学 Automatic tongue contour extraction method based on nuclear magnetic resonance images
CN102831606A (en) * 2012-08-01 2012-12-19 中国科学院自动化研究所 Method for acquiring vocal organ profile in medical image

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102750549A (en) * 2012-06-12 2012-10-24 天津大学 Automatic tongue contour extraction method based on nuclear magnetic resonance images
CN102831606A (en) * 2012-08-01 2012-12-19 中国科学院自动化研究所 Method for acquiring vocal organ profile in medical image

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
《A SHAPE-BASED FRAMEWORK TO SEGMENTATION》;Ting Peng et al.;《2010 IEEE INTERNATIONAL CONFERENCE ON ICASSP》;20100314;662-665 *
《Rapid Semi-automatic Segmentation of Real-time Magnetic Resonance Images for Parametric Vocal Tract Analysis》;Michael I. Proctor et al.;《INTERSPEECH 2010》;20100926;1575-1579 *

Also Published As

Publication number Publication date
CN105551040A (en) 2016-05-04

Similar Documents

Publication Publication Date Title
CN106485695B (en) Medical image Graph Cut dividing method based on statistical shape model
Cao et al. Deformable image registration using a cue-aware deep regression network
CN103310449B (en) Based on the lung dividing method improving shape
Li et al. Lvit: language meets vision transformer in medical image segmentation
Yassa et al. A quantitative evaluation of cross-participant registration techniques for MRI studies of the medial temporal lobe
CN107977952A (en) Medical image cutting method and device
CN107862694A (en) A kind of hand-foot-and-mouth disease detecting system based on deep learning
Cao et al. Region-adaptive deformable registration of CT/MRI pelvic images via learning-based image synthesis
CN106204718A (en) A kind of simple and efficient 3 D human body method for reconstructing based on single Kinect
CN104732546B (en) The non-rigid SAR image registration method of region similitude and local space constraint
CN103903013A (en) Optimization algorithm of unmarked flat object recognition
CN105389811A (en) Multi-modality medical image processing method based on multilevel threshold segmentation
CN109509193B (en) Liver CT atlas segmentation method and system based on high-precision registration
CN105719278A (en) Organ auxiliary positioning segmentation method based on statistical deformation model
Wang et al. Liver segmentation from CT images using a sparse priori statistical shape model (SP-SSM)
Zhu et al. Multi-level colonoscopy malignant tissue detection with adversarial CAC-UNet
CN105551040B (en) The method and system of tongue position profile is automatically extracted in nuclear-magnetism image sequence
CN102831614A (en) Sequential medical image quick segmentation method based on interactive dictionary migration
Wu et al. Auto-contouring via automatic anatomy recognition of organs at risk in head and neck cancer on CT images
CN107451594A (en) A kind of various visual angles Approach for Gait Classification based on multiple regression
Gong et al. Innovative Deep Learning Methods for Precancerous Lesion Detection
Jin et al. Object recognition in medical images via anatomy-guided deep learning
CN104933672B (en) Method based on quick convex optimized algorithm registration three dimensional CT with ultrasonic liver image
CN104933719B (en) One kind integration segment spacing is from detection image edge method
Jiao et al. TransDose: Transformer-based radiotherapy dose prediction from CT images guided by super-pixel-level GCN classification

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant