CN1162798C - Chinese medicine tongue colour, fur colour and tongue fur thickness analysis method based on multiclass support vector machine - Google Patents
Chinese medicine tongue colour, fur colour and tongue fur thickness analysis method based on multiclass support vector machine Download PDFInfo
- Publication number
- CN1162798C CN1162798C CNB021037957A CN02103795A CN1162798C CN 1162798 C CN1162798 C CN 1162798C CN B021037957 A CNB021037957 A CN B021037957A CN 02103795 A CN02103795 A CN 02103795A CN 1162798 C CN1162798 C CN 1162798C
- Authority
- CN
- China
- Prior art keywords
- tongue
- pixel
- fur
- coating
- type
- 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.)
- Expired - Fee Related
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 44
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 38
- 239000003814 drug Substances 0.000 title claims abstract description 14
- 238000000034 method Methods 0.000 claims abstract description 65
- 238000003745 diagnosis Methods 0.000 claims abstract description 14
- 238000007417 hierarchical cluster analysis Methods 0.000 claims abstract description 7
- 238000000576 coating method Methods 0.000 claims description 110
- 239000011248 coating agent Substances 0.000 claims description 108
- 210000005182 tip of the tongue Anatomy 0.000 claims description 55
- 210000005181 root of the tongue Anatomy 0.000 claims description 37
- 238000002474 experimental method Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 15
- 229940079593 drug Drugs 0.000 claims description 11
- 238000011002 quantification Methods 0.000 claims description 9
- 238000012113 quantitative test Methods 0.000 claims description 7
- 238000013459 approach Methods 0.000 claims description 6
- MCSXGCZMEPXKIW-UHFFFAOYSA-N 3-hydroxy-4-[(4-methyl-2-nitrophenyl)diazenyl]-N-(3-nitrophenyl)naphthalene-2-carboxamide Chemical compound Cc1ccc(N=Nc2c(O)c(cc3ccccc23)C(=O)Nc2cccc(c2)[N+]([O-])=O)c(c1)[N+]([O-])=O MCSXGCZMEPXKIW-UHFFFAOYSA-N 0.000 claims description 5
- 101100129500 Caenorhabditis elegans max-2 gene Proteins 0.000 claims description 4
- 241001092500 Photinia x fraseri Species 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 10
- 230000008901 benefit Effects 0.000 abstract description 2
- 235000012571 Ficus glomerata Nutrition 0.000 abstract 3
- 244000153665 Ficus glomerata Species 0.000 abstract 3
- 210000002105 tongue Anatomy 0.000 description 358
- 230000006870 function Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 238000007689 inspection Methods 0.000 description 4
- 239000003086 colorant Substances 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000000047 product Substances 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 239000000872 buffer Substances 0.000 description 2
- 238000009672 coating analysis Methods 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 206010043946 Tongue conditions Diseases 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Landscapes
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Image Analysis (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The present invention relates to a method for analyzing tongue color, fur color and the thickness of tongue fur in traditional Chinese medicine on the basis of various vector support machines, which is used in the field of computerized medical image processing. The method comprises the following steps that images are acquired by a digital camera, the images are input into a computer processor by means of a USB interface, and then, tongue body areas are separated from tongue images in the processor. The present invention is characterized in that the method also orderly comprises the following steps that on the basis of making hierarchical cluster analysis so as to form a cluster tree and adjusting the cluster tree, pixels of the tongue body areas are identified by the support vector machine (CTSVM) on the basis of the cluster tree, and the pixels of the tongue body areas are divided into 15 varieties; then, a tongue surface is divided into 5 areas, and pixel varieties with the largest number in each area and the number of the pixels of each variety are counted; next, tongue color is described, tongue fur color is described, and the thickness of tongue fur is quantitatively analyzed and described; finally, analysis results are displayed by means of characters and pseudo colored pictures. The method has the advantages of rapidness, accuracy, practicality and conformation to diagnosis principles of traditional Chinese medicine experts.
Description
Technical field
The present invention relates to the computing machine field of medical image processing, designed a kind of traditional Chinese medical science tongue image analytic system, the tongue body image is carried out the quantitative test and the qualitative description of tongue color, coating colour, thickness of the tongue coating, auxiliary tcm diagnosis based on multi-class support vector machine.
Background technology
Lingual diagnosis is the important content of observation in the traditional Chinese medical science four methods of diagnosis.Lingual diagnosis is by the doctor tongue body to be observed, thereby judges.This method has very strong subjectivity, ambiguity, and examination result and doctor's experience is closely related.People strive for the tongue picture feature is carried out objective quantitative analysis at present, overcome the subjectivity of traditional visualization, improve the reliability of lingual diagnosis, have very big advantage.
Adopt computer image processing to carry out the tongue picture signature analysis, can reflect the situation of tongue picture more all sidedly, meet the custom of Evolution of Tongue Inspection of TCM, be convenient to the collection and the preservation of lingual diagnosis data, have good practical values.It is under certain illumination condition, gathers experimenter's colored tongue image, imports computing machine after the digitizing, utilizes image analysis technology, the tongue picture feature is analyzed automatically, and the result is stored in the computing machine.
When existing image processing techniques is used for the tongue picture analysis, just adopt the analysis of tongue image color value simple statistics mostly.We find by analysis, and the distribution of tongue color, coating colour and thickness has locality, and the zones of different on lingual surface often shows different features, observe the part during Evolution of Tongue Inspection of TCM earlier, carry out describe, in general terms again.Mainly emphasize to have the feature of dialectical meaning during description.
At image processing or analysis field, because the condition difference of picture-taking, the parameter in processing or the analytical approach may be according to the actual conditions adjustment.The method of adjusting is, from the image that the same terms is taken, takes out a width of cloth or a few width of cloth, is determined by experiment the occurrence of parameter.Each parameter in this method is substantially all determined in this way.
Support vector machine in the existing mode identification technology (SVM) method is to solve relatively effective method of small sample identification problem at present, and basic thought can be used the bidimensional situation explanation of Fig. 3.Among Fig. 3, solid dot and hollow dots are represented two class samples, and H is a sorting track, H
1, H
2Be respectively all kinds of in from the nearest sample of sorting track and be parallel to the straight line of sorting track, the distance between them is called class interval (margin).So-called optimal classification line requires sorting track not only two classes correctly can be separated (the training error rate is 0) exactly, and makes the class interval maximum.
As can be seen from Figure 3, (b) class interval in is big, and the possibility of its classification error (a) is medium and small, and the sample that is positioned on the optimal classification line is called support vector, more than is the example of linear separability situation.For nonlinear situation, be transformed into the feature space (Feature Space) of higher-dimension by nonlinear transformation, the structure linear discriminant function is realized the Nonlinear Discriminant Function in the former space in higher dimensional space.
The tongue picture sample comprises the digitizing tongue image of collection and the clinical lingual diagnosis conclusion that the lingual diagnosis expert provides.Sample size is many more, and clinical judgment is reliable more, and then the model of setting up on this basis is realistic more, and the accuracy of new tongue picture being carried out signature analysis is also just high more.
Yet the tongue picture of some type is very rare (as grey and black coat, pale tongue, blue-purple tongue etc.).So the skewness of all kinds of samples, limited amount, as the mode identification method adopting certainly will influence accuracy and reliability that the tongue picture feature is analyzed automatically.Therefore adopt support vector machine (SVM) mode identification method to carry out pixel identification,, reduce the classification error rate to improve the accuracy of classification.
Classical SVM is generally at two class identification problems, and tongue body zone pixel identification is a multiclass problem.When being used for the multiclass pattern-recognition, be a plurality of two class identification problems usually with former PROBLEM DECOMPOSITION.Existing method comprises 1-a-a (one-against-all) and 1-a-1 (one-against-one) etc.Wherein 1-a-a is by k two class SVM (f
1..., f
k) form f
jBe category label with j class sample be decided to be+1, and the category label of other all samples is decided to be-1,1-a-1 then constitutes two class identification problems respectively with each class sample and other are all kinds of, constitutes altogether
Individual two class SVM.During application, sample obtains through two all class SVM
Individual recognition result, wherein the most dominant category attribute is final recognition result.The shortcoming of these two kinds of methods is that the support vector number that obtains after the training is more, the recognition speed when influence is used.
Summary of the invention
In order to solve foregoing problems, can be automatically, accurately, features such as the tongue color of express-analysis tongue body image, coating colour, thickness of the tongue coating, the present invention has designed a kind of new multiclass method---based on support vector machine (CTSVM) method of clustering tree.This method adopts the pixel identification (partial analysis) of carrying out the tongue body zone earlier, carries out " two-step approach " analytical procedure of describe, in general terms again.In pixel when identification, combine support vector machine and can effectively finish the small sample identification problem, and clustering tree adopts stage division that the multiclass problem is converted into the characteristics of two class problems, thereby fast, accurately reach practical.Describe, in general terms carries out according to main pixel class, meets the examination principle of Traditional Chinese Medicine experts.Introduce the thick index of tongue simultaneously, made the thickness of the tongue coating quantification.Adopt this method can obtain the qualitative description and the quantitative test of tongue color in the tongue image, coating colour, tongue condition thickness with computing machine.Technical thought of the present invention is characterised in that:
When 1, image processing method being used for the tongue picture signature analysis, be different from general method, but adopted the pixel identification (partial analysis) of carrying out the tongue body zone earlier, carry out " two-step approach " analytical procedure of describe, in general terms again.Owing to observe the part during Evolution of Tongue Inspection of TCM earlier, carry out describe, in general terms again.Thereby the diagnosis of this method and traditional Chinese medical science custom matches, and can effectively discern the tongue picture feature.
2, we have adopted the pixel identification (partial analysis) of carrying out the tongue body zone earlier, carry out the analytical procedure of describe, in general terms again.Also promptly the feature of tongue nature, tongue fur is carried out partial analysis.According to color attribute, adopt multi-class support vector machine to be divided into different classifications the pixel on the tongue body.We are made as 15 kinds to the classification number of tongue body district pixel identification, comprise that tongue nature is light, tongue nature is light red, tongue nature is red, dimly red tongue, tongue nature deep red red, tongue nature is dark violet, thin and whitish fur, white tongue, white thick coating, thin and yellowish fur, yellow tongue, yellow thick coating, grayish fur, brown tongue, black tongue etc.Wherein, the first six kind is the tongue nature type, and back nine kinds is the tongue fur type.During pixel identification, generally need provide the pixel of known class earlier, be called sample.If sample relatively seldom, such identification problem is called small sample problem.
Discern the classification problem of this multiclass, small sample at pixel, adopted a kind of new multi-class support vector machine based on clustering tree (CTSVM) method.The main thought of CTSVM is earlier the hierarchical clustering analysis to be carried out at all kinds of centers of training sample set, generates the multistage classifier that is called clustering tree, constitutes (k-1) individual two class SVM more in view of the above.Utilizing it to be converted into several simple classification problems to the multi-class classification problem of a complexity solves.The non-leaf node of each of clustering tree is represented a support vector sorter.The structure of clustering tree can be determined according to artificial experience, also can adopt the hierarchical clustering method to obtain.So-called cluster is exactly the principle according to " Things of a kind come together ", and it is a type that the big sample of similarity is assembled.This method can effectively be converted into the multiclass identification problem the less thereby recognition speed of number two class classification problems faster, and can overcome the less influence of sample in the tongue picture analysis to discrimination, thereby can carry out pixel identification comparatively accurately, and obtain clustering tree by hierarchical clustering and in conjunction with the method for Traditional Chinese Medicine experts experience, referring to Fig. 4.
The form of clustering tree is determined according to the distribution situation and the minimum distance criterion self-adaptation of training sample set.This clustering tree is combined with two quasi-modes identification SVM, just constituted CTSVM.In CTSVM, each non-leaf node (root node and intermediate node) all is one two class support vector machines.
We are according to said method, and the clustering tree of determining in conjunction with the examination experience of Traditional Chinese Medicine experts as shown in Figure 4.Among the figure, H~M tongue nature type, N~V is the tongue fur type.Among the figure, X is called root node, and A~G is called non-leaf node.The upper layer node that links to each other with a node is called father node, and corresponding, this node is called the child node of its father node.Among Fig. 3, A has two child node C and D, and C has three child node H, I, J, at first divides two maximum big classes of distance during pixel identification, carries out thinner division more step by step.Promptly set out, use with this node respective classified device and classify,,, promptly obtain the final recognition result of this pixel until arriving certain leaf node with its ownership and a certain child node by root node.In the present invention, all adopt support vector machine with the corresponding sorter of each non-leaf node.If a non-leaf node has two child nodes, then directly adopt two class support vector machines to get final product; If a non-leaf node has three child nodes, just there are three two class support vector machines to finish common pixel identification mission at this node, each two class support vector machines separates the class of one of them child node representative with the class of other two child node representatives.
3, lingual surface is divided into five zones, the maximum pixel class of quantity in each district of statistics, is carried out in conjunction with the experience of Traditional Chinese Medicine experts mainly according to this classification this description of distinguishing tongue color, coating colour.This meets Evolution of Tongue Inspection of TCM mainly emphasizes to have the principle of the feature of dialectical meaning when describing.
4, thickness of the tongue coating divides time-like, and each pixel in tongue body zone is divided into " no tongue ", " thin tongue ", " middle thick coating ", " thick coating " 4 types.By the classification of pixel thickness of the tongue coating, its foundation is " thickness of tongue fur is a standard to see the bottom, not see the bottom " earlier.Carry out the thickness of the tongue coating quantification again,, introduce the thick index of tongue, the thickness of the tongue coating of view picture tongue image is carried out quantification describe by on the basis of pixel identification.
5, in order to make analysis result easy to understand and clinical practice, adopt literal and pseudocolour picture display analysis result.The pixel identification result is expressed as tongue color, coating colour distribution pseudocolour picture.To be expressed as thickness of the tongue coating pseudo-colours distribution plan by the thickness classification results of pixel.Adopt literal that tongue color, coating colour thickness of the tongue coating are carried out qualitative description, and provide the thickness of the tongue coating exponential quantity.
Technical scheme of the present invention is seen Fig. 1, Fig. 2, Fig. 5, Fig. 6, Fig. 8, Fig. 9, Figure 10, method step is for to carry out image acquisition by digital camera, and image is input in the computer processor by USB interface, in processor, adopt current techique, the tongue body zone is split from tongue image, be characterised in that this method also comprises the steps: successively
1) on the basis of carrying out hierarchical clustering analysis generation clustering tree and adjusting, adopts based on support vector machine (CTSVM) method of clustering tree and carry out the identification of tongue body area pixel;
The hierarchical clustering analysis generates the method step of clustering tree and sees Fig. 5:
(1), calculates k class centralization: (x for k class sample
1, x
2..., x
k),
(2) the 1st grades of divisions are that each class center respectively becomes a class, i.e. k leaf node,
(3) ask distance D between per two class centers
Ij, i=1 ... k, j=1 ... k, i ≠ j,
(4) nearest two centers are merged, form new class center, i.e. intermediate node,
(5) repeat (3)-(4) step, up to the k-1 level, 1 class is merged at all centers, i.e. root node,
(6) as required, adjust the structure of decision tree, make that the degree of depth of tree is the least possible; The method of tongue body area pixel identification and step see Fig. 6:
(1) calculating three chroma color value R in each pixel 3 * 3 neighborhood, the average of G, B in the input imagery, is other input feature vector as pixel,
(2) read R, G, the B value of a pixel in the tongue body,
(3) from root node, the svm classifier device according to this node is divided into a certain branch with this pixel,
(4) if branch node is not a leaf node, then repeat the process in (3): according to the svm classifier device of this node, this pixel is divided into a certain branch, if branch node is a leaf node, then the class of this leaf node representative is the classification of the pixel of being discerned, the sorter of each node is two class support vector machines sorters, and assorting process realizes according to the method for routine
(5) process of repetition (2)-(4), the whole pixels classification up to the tongue body zone finish.
As can be seen, during pixel identification, at first divide two maximum big classes of distance, carry out thinner division more step by step,, just obtain pixel and get category attribute until leaf node.Owing to reduced the number of times that two classes are differentiated, take into account the distribution situation of sample simultaneously, this method can guarantee the accuracy rate of identification when accelerating recognition speed.
2) add up various types of number of picture elements in each subregion;
3) carry out tongue color and describe, see Fig. 8;
4) carry out coating colour and describe, see Fig. 9;
5) carry out thick libngual fur quantitative test and description, see Figure 10;
6) adopt literal and pseudocolour picture display analysis result.In order to make analysis result easy to understand and clinical practice, adopt literal and pseudocolour picture display analysis result.The pixel identification result is expressed as tongue color, coating colour distribution pseudocolour picture.To be expressed as thickness of the tongue coating pseudo-colours distribution plan by the thickness classification results of pixel.Adopt literal that tongue color, coating colour thickness of the tongue coating are carried out qualitative description, and provide the thickness of the tongue coating exponential quantity.Figure 11 is display analysis result's a example.
The description feature of tongue color, coating colour, thickness of the tongue coating also is in addition, after pixel identification is finished, adds up in each subregion behind various types of number of picture elements, makes N
i jRepresent that the j class resembles wherein i=1 in the i district, 5 represent in the root of the tongue, the tongue respectively, 5 zones such as the tip of the tongue, tongue left side, tongue right side, j=1,, 15 expression tongue natures are light, tongue nature is light red, tongue nature is red, dimly red tongue, tongue nature deep red red, tongue nature is dark violet, 15 kinds of tongue natures such as thin and whitish fur, white tongue, white thick coating, thin and yellowish fur, yellow tongue, yellow thick coating, grayish fur, brown tongue, black tongue and tongue fur type, and on this basis, according to the lingual diagnosis custom of Traditional Chinese Medicine experts, carry out the description of tongue color, coating colour, thickness of the tongue coating.
Wherein the describing method of tongue color is the color according to the tongue nature type specification tongue nature of tongue lateral areas and the tip of the tongue district pixel.Because tongue nature mainly is distributed in the tongue side and the tip of the tongue, thereby 3 zone consideration tongue nature types on the tip of the tongue, tongue left side, tongue right side etc. only.Calculate the sum of all pixels that belongs to each tongue nature type (belong to promptly that tongue nature is light, tongue nature is light red, tongue nature is red, deep red red, the type such as tongue nature is dark violet of dimly red tongue, tongue nature) in 3 zones such as the tip of the tongue, tongue left side, tongue right side, the tongue nature type that pixel count is maximum is the tongue nature feature of this tongue image.According to the experience of the traditional Chinese medical science, tongue nature whether part has dark violet and whether the tip of the tongue is that red tongue is very important, therefore to the both of these case individual processing.Method is: calculate 3 zones such as the tip of the tongue, tongue left side, tongue right side and belong to the number of pixels of " tongue nature is dark violet " and the ratio of whole lingual surface sum of all pixels, if this ratio, is then assert " local dark violet " greater than certain default value; Calculate the tip of the tongue district and belong to the number of pixels of " tongue nature is red " and the ratio of whole lingual surface sum of all pixels, if this ratio greater than certain default value, is then assert " tongue nature is red ".Detailed process is:
1) total area (sum of the pixel) A of calculating lingual surface;
2) calculate the sum of 6 kinds of tongue nature types of tongue side and the tip of the tongue district respectively
3) order
J then
MaxThe tongue nature type of representative is the tongue nature feature of this tongue image;
4) the area ratio of the dark violet tongue nature of calculating
If R
6>θ
pAnd j
Max≠ 6, then increase and describe " local dark violet ", wherein θ
pFor according to the experiment preset threshold;
5) the area ratio in calculating the tip of the tongue district " tongue nature is red "
If R
3>θ
rAnd j
Max≠ 3, then increase and describe " red tip of tongue ", wherein θ
rFor according to the experiment preset threshold.
The wherein description of coating colour because the tongue side seldom has tongue fur to distribute, thereby to the description of tongue fur be divided in the root of the tongue, the tongue, 3 zones of the tip of the tongue.Calculate in the root of the tongue, the tongue, the tip of the tongue etc. belongs to each tongue fur type sum of all pixels of (promptly belonging to types such as thin and whitish fur, white tongue, white thick coating, thin and yellowish fur, yellow tongue, yellow thick coating, grayish fur, brown tongue, black tongue) in 3 zones, in three zones, whether the pixel count of judging the tongue fur type that pixel count is maximum respectively and the ratio of whole lingual surface sum of all pixels be less than a certain threshold value, if less than, then assert this Qu Shaotai, otherwise this tongue fur type is the tongue fur feature in this district.These three threshold values are obtained by test among the present invention.According to the experience of the traditional Chinese medical science, the district often presents multiple tongue fur in the root of the tongue, the tongue, therefore also the tongue fur type that district's pixel is many for several times in the root of the tongue, the tongue is judged.Method is: calculate the ratio of the number of pixels of the tongue fur type that number of pixels and the pixel count of many tongue fur types is maximum for several times of district's pixel in the root of the tongue, the tongue, if this ratio is greater than certain default value, then should the district increases and describe the second main type tongue fur.Concrete grammar is as follows:
1) calculates the total area (sum of the pixel) A of lingual surface, in the root of the tongue, the tongue, the area A in the tip of the tongue district
i, i=1 ..., 3;
2) in the root of the tongue, the tongue, the tip of the tongue district, calculate respectively
I=1 wherein, 2,3,
J then
Max iThe tongue fur type of representative is the tongue fur feature in this district;
3) calculate i=1, j in 2,3 districts
Max iThe area ratio that the tongue fur pixel is shared
4) respectively in the root of the tongue, the tongue, the tip of the tongue district is described.For root of the tongue district i=1, if
Then
Be described as " the few tongue of the root of the tongue "; Otherwise the description root of the tongue is j
Max 1The tongue fur of type.For distinguishing i=2 in the tongue, if
Then be described as " few tongue in the tongue "; Otherwise in the description tongue j
Max 2The tongue fur of type, for the tip of the tongue district i=3, if
Then be described as " the few tongue of the tip of the tongue "; Otherwise the description the tip of the tongue is j
Max 3The tongue fur of type.θ wherein
1, θ
2, θ
3For according to the experiment preset threshold;
5), calculate the second main type tongue fur respectively for distinguishing in the root of the tongue, the tongue
And calculating j
Sec iType pixels area and j
Max iThe area ratio of type
If
Then the i district increases the description second main type tongue fur.θ wherein
SecFor according to the experiment preset threshold.
Wherein the analysis of thickness of the tongue coating and description are divided into two parts: the classification of (1) thickness of the tongue coating.Each pixel in tongue body zone is divided into " no tongue ", " thin tongue ", " middle thick coating ", " thick coating " 4 types.By pixel thickness of the tongue coating The classification basis is " thickness of tongue fur is a standard to see the bottom, not see the bottom ".(2) thickness of the tongue coating quantification.By on the basis of pixel identification, the thickness of the tongue coating of view picture tongue image is carried out quantification describe.What wherein, " see the bottom " with judgements of tongue nature pixel in the neighborhood of pixels.If the tongue nature pixel is a lot of in the neighborhood of pixels, the degree that then sees the bottom is big.The tongue fur type of pixel has also reflected the thickness of tongue fur.Belong to thin and whitish fur, thin and yellowish fur and reflect that this pixel place is thin tongue; Belong to white tongue, yellow tongue, grayish fur, brown tongue and show that this pixel place is middle thick coating; Belong to white thick coating, yellow thick coating, black tongue shows that this pixel place is a thick coating.Concrete steps are:
1) reads the classification number of tongue body pixel and 5 * 5 neighborhoods thereof;
2) establish c
1, c
2For according to experiment preset threshold c
1>c
2, and.Calculate the number of picture elements S that belongs to the tongue nature type in these pixel 5 * 5 neighborhoods
Body, if S
Body>c
1, then the thickness of the tongue coating type of this point is decided to be " no tongue ";
3) if c
1〉=S
Body>c
2, then the thickness of the tongue coating type of this point is decided to be " thin tongue ",
4) if S
Body≤ c
2, then the tongue body and tongue coating type j according to this point determines the thickness type, if tongue nature (j<7), then being " thin tongue ", as if j=7 (thin and whitish fur), 10 (thin and yellowish furs), then is " thin tongue ", if j=8 (white tongue), 11 (yellow tongues), 13 (grayish furs), 14 (brown tongue) then are " middle thick coating "; If j=9 (white thick coating), 12 (yellow thick coatings), 15 (black tongue) then are " thick coating ";
5) repeat (1)-(3) step, finish up to all processes pixel;
6) according to the result of pixel identification, calculate the thick index T of tongue of view picture image, as the quantification result that thickness of the tongue coating is analyzed, computing method are:
A represents the pixel sum in tongue body district in the formula,
Expression is to all pixel summations in the tongue body zone, j
kThe thickness type of representing k pixel gets 0,1,2,3 respectively, w corresponding to " no tongue ", " thin tongue ", " middle thick coating ", " thick coating "
JkBe weights, determine that according to the experience of experiment and Traditional Chinese Medicine experts the thick index of tongue can be described the general thickness of tongue fur.
As seen from Figure 11, quantitative test and qualitative description result that this method can obtain by computing machine, and the result is consistent with the visual examination result of the traditional Chinese medical science help the objectifying of lingual diagnosis, standardization.
Description of drawings
Fig. 1 is traditional Chinese medical science tongue picture tongue color, coating colour, the thick analytic system block diagram of tongue
1, digital camera, 2, USB interface, 3, computer processor, 4, output buffers, 5, the tongue picture analysis, 6, display, 7, analysis result;
Fig. 2 is a traditional Chinese medical science tongue picture analysis method main program flow chart;
Fig. 3 support vector machine principle key diagram
(a) the less classifying face in class interval (b) has the optimal classification face of maximum class interval;
Fig. 4 is used for the clustering tree synoptic diagram that the hierarchical clustering analysis generates
H, dimly red tongue, I, tongue nature are light red, and J, tongue nature are light, and K, tongue nature are dark reddish purple, and L, tongue nature are dark violet, and M, tongue nature are red, N, white tongue, O, white thick coating, P, thin and whitish fur, Q, yellow tongue, R, thin and yellowish fur, S, yellow thick coating, T, brown tongue, U, grayish fur, V, black tongue;
Fig. 5 is the generation method subroutine flow chart of Fig. 4 clustering tree;
Fig. 6 is a tongue body area pixel recognition methods subroutine flow chart;
Fig. 7 is a lingual surface subregion synoptic diagram;
Fig. 8 is a tongue color describing method subroutine flow chart;
Fig. 9 is a coating colour describing method subroutine flow chart;
Figure 10 is thickness of the tongue coating analysis and describing method subroutine flow chart;
Figure 11 is the analysis result of tongue image
(a) tongue body image, (b) tongue color, coating colour pseudocolour picture, (c) thickness of the tongue coating pseudocolour picture, (d) the automatic analysis result of text description;
Figure 12 is the program main flow chart that the tongue picture that moves is on computers analyzed;
Figure 13 is the tongue body area pixel recognition subroutine process flow diagram that moves on computers.
Embodiment
In the traditional Chinese medical science tongue picture analytic system block diagram of Fig. 1, digital camera and USB interface all are commercially available, mainly finish the collection tongue image, the optical signalling of tongue body and colour code is converted to visual electric signal is input to computing machine, are convenient to operations such as Computer Processing, transmission; Computer Processing mainly is by USB interface software tongue image to be carried out read/write process, and the tongue image after the processing outputs to buffer, is convenient to show.Display is the output device of image, and human eye can be watched original tongue image and analysis result by display.The tongue body analysis is that the tongue image that computing machine reads in is carried out tongue color, coating colour, thick quantitative test and the qualitative description of tongue, and the output analysis result.
Original tongue image can be the image that collects in real time by digital camera, also can be to realize collecting the image that is kept in the hard disc of computer by digital camera.
Adopt support vector machine to carry out pixel identification, the key issue that relates to comprises the choosing etc. of the choosing of formation, kernel function, penalty factor of selection, the study collection in input feature vector space.The study collection refers to the set of the sample that classification is known.
In the native system, the input feature vector space is the RGB color space, and for avoiding The noise, to each pixel, the RGB average of getting its 3 * 3 neighborhood is as eigenvector.In some typical images, select the sub-piece of a series of images, determine to constitute the study collection after the classification by the Traditional Chinese Medicine experts block-by-block.Sample (visual sub-piece) is decided according to the concrete condition of image, does not have unified size.
The penalty factor C of SVM has embodied the degree of belief to the study collection among the present invention.C is big more, and degree of belief is high more, and C is big more, and degree of belief is high more.In the research of tongue body pixel identification, because the classification number of classification is more, and the separability between all kinds of is relatively poor, need choose a suitable penalty factor by experiment.Among the present invention, choose C=400 by experiment.
The corresponding learning machine of different kernel functions has dissimilar non-linear decision surfaces in the input space.We have tested kernel function forms such as polynomial expression commonly used, radial basis function (RBF), neural network.The final kernel function of using is RBF, and its form is:
σ is an important parameters in the formula, is determining the concrete form of RBF kernel function, chooses σ=50 by experiment.
In order to make analysis result easy to understand and clinical practice, carry out qualitative, the quantitative description of tongue nature and tongue fur among the present invention.With lingual surface be divided in the root of the tongue, the tongue, 5 zones such as the tip of the tongue, tongue left side, tongue right side.Division methods is to be divided into before 5 five equilibriums 1/5 with the tip of the tongue to herringbone sulcus terminalis mid point to claim the tip of the tongue, in 2/5 claim in the tongue that back 2/5 claims the root of the tongue.With the standardized line of mid point on tongue center line and tongue limit, the outer part of line claims the tongue side in addition.As shown in Figure 7.Add up various types of number of picture elements in each subregion,, carry out traditional Chinese medical science tongue color, the coating colour based on multi-class support vector machine, the computer analysis method of thickness of the tongue coating according to the lingual diagnosis of Traditional Chinese Medicine experts custom.
The tongue picture analysis mainly realizes by software.In computing machine, finish following program (master routine is seen Figure 12):
1, reads in tongue body range image data, read in support vector and the corresponding coefficient of CTSVM, initiation parameter θ
p, θ
r, θ
1, θ
2, θ
3, θ
Sec, c
1, c
2, w
0, w
1, w
2, w
3θ wherein
p, θ
rBe respectively when tongue color is analyzed the threshold value of judging " local dark violet " and " red tip of tongue ", get θ among the present invention
p=0.01, θ
r=0.06.θ
1, θ
2, θ
3In the root of the tongue, the tongue, the threshold value in the tip of the tongue three districts, get θ among the present invention
1=0.5, θ
2=0.2, θ
3=0.4.c
1, c
2Judge when analyzing for tongue is thick tongue fur have or not with the tongue fur type time threshold value that adopts.Get c among the present invention
1=20, c
2=8.w
0, w
1, w
2, w
3Weight coefficient during for the thick index of calculating tongue is defined as w among the present invention
0=0, w
1=0.2, w
2=0.7, w
3=1.3.
2, enter the pixel recognition subroutine, adopt based on support vector machine (CTSVM) method of clustering tree and carry out the identification of tongue body area pixel.Establish at this and to obtain clustering tree shown in Figure 4 by training.Each node in the tree all is a support vector machine classifier, and support vector and corresponding coefficient obtain by training.To a pixel, the assorting process of support vector machine is as follows:
1) calculate RGB average x=in these pixel 3 * 3 neighborhoods (R, G, B);
2) from root node, calculate the kernel function inner product of this pixel each support vector corresponding, and multiply by corresponding coefficient and label with this node, to all these product summations, ask the symbol that adds constant b result afterwards then.Promptly ask:
In the formula, x
iBe the support vector of this sorter, α
iBe the coefficient corresponding with support vector, y
iBe the label of support vector sample, b is a constant, and b gets 40 among the present invention, K (x
i, x) be the kernel function inner product,
‖ in the formula. ‖ is the norm of vector, and σ gets 50.
Sign () is-symbol function, if the value of independent variable greater than 0, then functional value is 1, less than 0, then is-1.
If f (x) greater than 0, then is referred to it the branch on the left side, otherwise the branch on the right of being referred to;
3) process repetition 2) is until arriving certain branch node.For example the identifying of certain pixel is: X → A → D → K, and then this pixel is differentiated and is " tongue nature is deep red red ";
4), carry out 2 to each pixel), 3), all pixels are all differentiated and are finished in the tongue body zone, each pixel has all been distributed a classification number.
3, lingual surface subregion.Scan the tongue body zone from left to right, from top to bottom, obtain the boundary rectangle in tongue body zone.If two coordinates of limit in tongue image are l, r about rectangle, two coordinates of limit in tongue image are t, b up and down, then according to the principle of aforementioned lingual surface subregion (with lingual surface be divided in the root of the tongue, the tongue, 5 zones such as the tip of the tongue, tongue left side, tongue right side, division methods is to be divided into 5 five equilibriums with the tip of the tongue to herringbone sulcus terminalis mid point, preceding 1/5 claims the tip of the tongue, in 2/5 claim in the tongue that back 2/5 claims the root of the tongue.With the standardized line of mid point on tongue center line and tongue limit, the outer part of line claims the tongue side in addition.), obtain following subregion result: establish (x, y) the longitudinal and transverse coordinate of pixel, then tongue left side in the expression tongue image: if
And (x, y) in the tongue body zone, the tongue right side: if
And (x, y) in the tongue body zone, root of the tongue district: if
And (x, y), distinguish in the tongue in the tongue body zone: if
The tip of the tongue district: if
And (x is y) in the tongue body zone.
4, add up various types of number of picture elements N in each subregion
i j
5, tongue color analysis and description.According to the tongue nature type specification tongue nature of the tongue side and the tip of the tongue, concrete grammar is:
1) total area (sum of the pixel) A of calculating lingual surface;
2) calculate the sum of 6 kinds of tongue nature types of tongue side and the tip of the tongue district respectively:
3) order
J then
MaxThe tongue nature type of representative is the tongue nature feature of this tongue image, for example j
MaxBe described as in=2 o'clock " tongue nature is light red ";
4) the area ratio of the dark violet tongue nature of calculating
If R
6>0.01 and j
Max≠ 6, then increase and describe " local dark violet ";
5) the area ratio in calculating the tip of the tongue district " tongue nature is red "
If R
3>0.05 and j
Max≠ 3, then increase and describe " red tip of tongue ".
6, coating colour analysis and description.Description to tongue fur is divided in the root of the tongue, the tongue, 3 zones of the tip of the tongue, and concrete steps are as follows:
1) calculates the total area (sum of the pixel) A of lingual surface, in the root of the tongue, the tongue, the area A in the tip of the tongue district
i, i=1 ..., 3;
2) in the root of the tongue, the tongue, the tip of the tongue district, calculate respectively
I=1 wherein, 2,3,
J then
Max iThe tongue fur type of representative is the tongue fur feature in this district;
3) calculate i=1, j in 2,3 districts
Max iThe area ratio that the tongue fur pixel is shared
4) for root of the tongue district i=1, if
Then be described as " the few tongue of the root of the tongue "; For distinguishing i=2 in the tongue,
If
Then be described as " few tongue in the tongue "; For the tip of the tongue district i=3, if
Then be described as " the few tongue of the tip of the tongue ";
5) if
Then describing the root of the tongue is j
Max iThe tongue fur of type.For example
The time, be described as " the yellow tongue of the root of the tongue ".If
Then describe and be j in the tongue
Max 2The tongue fur of type, for example
The time, be described as " white tongue in the tongue ".If
Then describing the tip of the tongue is j
Max 3The tongue fur of type, for example
The time, be described as " the tip of the tongue thin and whitish fur ";
6), calculate the second main type tongue fur respectively for distinguishing in the root of the tongue, the tongue
And calculating j
Sec iType pixels area and j
Max iThe area ratio of type
If
Then the i district increases the description second main type tongue fur.
7, thick quantitative test and the qualitative description of tongue.Concrete steps are:
1) reads the classification number of tongue body pixel and 5 * 5 neighborhoods thereof;
2) establish c
1, c
2For according to experiment preset threshold c
1>c
2, and.Calculate the number of picture elements S that belongs to the tongue nature type in these pixel 5 * 5 neighborhoods
Body, if S
Body>c
1, then the thickness of the tongue coating type of this point is decided to be " no tongue ";
3) if c
1〉=S
Body>c
2, then the thickness of the tongue coating type of this point is decided to be " thin tongue ";
4) if S
Body≤ c
2, then the tongue body and tongue coating type j according to this point determines the thickness type.If tongue nature (j<7) then is " a thin tongue "; If j=7 (thin and whitish fur), 10 (thin and yellowish furs) then are " thin tongue "; If j=8 (white tongue), 11 (yellow tongues), 13 (grayish furs), 14 (brown tongue) then are " middle thick coating "; If j=9 (white thick coating), 12 (yellow thick coatings), 15 (black tongue) then are " thick coating ";
5) repeat 1)~4) step, finish up to all processes pixel;
6), calculate the thick index T of tongue of view picture image, as the quantification result of thickness of the tongue coating analysis according to the result of pixel identification.Computing method are:
A represents the pixel sum in tongue body district in the formula,
Expression is to all pixel summations in the tongue body zone, j
kThe thickness type of representing k pixel gets 0,1,2,3 respectively, w corresponding to " no tongue ", " thin tongue ", " middle thick coating ", " thick coating "
JkBe weights, determine according to the experience of experiment and Traditional Chinese Medicine experts.Be defined as w among the present invention
0=0, w
1=0.2, w
2=0.7, w
3=1.3.The thick index of tongue can be described the general thickness of tongue fur.
8, the demonstration of analysis result.In order to make analysis result easy to understand and clinical practice, adopt literal and pseudocolour picture display analysis result.The pixel identification result is expressed as tongue color, coating colour distribution pseudocolour picture.To be expressed as thickness of the tongue coating pseudo-colours distribution plan by the thickness classification results of pixel.Adopt literal that tongue color, coating colour thickness of the tongue coating are carried out qualitative description, and provide the thickness of the tongue coating exponential quantity.
Figure 11 is display analysis result's a example.
Claims (2)
1, a kind of traditional Chinese medical science tongue color, coating colour, thickness of the tongue coating analytical approach based on multi-class support vector machine, be to carry out image acquisition by digital camera, and image is input in the computer processor by USB interface, in processor, adopt current techique, the tongue body zone is split from tongue image, the invention is characterized in that this method also comprises the steps: successively
1) on the basis of carrying out hierarchical clustering analysis generation clustering tree and adjusting, adopt support vector machine method to carry out the identification of tongue body area pixel based on clustering tree, below will abbreviate CTSVM as based on the support vector machine of clustering tree;
The method step that the hierarchical clustering analysis generates clustering tree is:
(1), calculates k class centralization: (x for k class sample
1, x
2..., x
k),
(2) the 1st grades of divisions are that each class center respectively becomes a class, i.e. k leaf node,
(3) ask distance D between per two class centers
Ij, i=1 ... k, j=1 ... k, i ≠ j,
(4) nearest two centers are merged, form new class center, i.e. intermediate node,
(5) repeat (3)-(4) step, up to the k-1 level, 1 class is merged at all centers, i.e. root node,
(6) as required, adjust the structure of clustering tree, make that the degree of depth of tree is the least possible, the clustering tree of generation combines with two quasi-modes identification support vector machine, has just constituted CTSVM;
The method of tongue body area pixel identification and step be:
(1) calculate three chroma color value R in each pixel 3 * 3 neighborhood, the average of G, B in the input imagery, as the input feature vector of pixel identification,
(2) read R, G, the B value of a pixel in the tongue body,
(3) from the root node of described clustering tree, the CTSVM sorter according to this node is divided into a certain branch with this pixel, and concrete grammar is:
Calculate the kernel function inner product of this pixel each support vector corresponding, and multiply by corresponding coefficient and label,, ask the symbol that adds constant b result afterwards then all these product summations with this node.Promptly ask:
In the formula, x
iBe the support vector of this sorter, α
iBe the coefficient corresponding with support vector, y
iBe the label of support vector sample, b is a constant, and b gets 40 among the present invention, K (x
i, x) be the kernel function inner product,
‖ x-y ‖ is the norm of vector x-y in the formula, and σ gets 50,
Sign () is-symbol function, if the value of independent variable greater than 0, then functional value is 1, less than 0, then is-1.
According to the value of f (x) and the classifying rules of this CTSVM, this pixel is divided into a certain branch then; (4) if branch node is not a leaf node, then repeat the process in (3): according to the CTSVM sorter of this node, this pixel is divided into a certain branch, if branch node is a leaf node, then the class of this leaf node representative is the classification of the pixel of being discerned, and the sorter of each node is two class support vector machines sorters, and assorting process realizes according to the method for routine, (5) process of repetition (2)-(4), the whole pixels classification up to the tongue body zone finish;
2) add up various types of number of picture elements in each subregion;
3) carrying out tongue color describes, color according to the tongue nature type specification tongue nature of tongue lateral areas and the tip of the tongue district pixel, calculate the sum of all pixels that belongs to each tongue nature type in 3 zones such as the tip of the tongue, tongue left side, tongue right side, the tongue nature type that pixel count is maximum is the tongue nature feature of this tongue image;
4) carrying out the tongue fur look describes, calculate in the root of the tongue, the tongue, belong to the sum of all pixels of each tongue fur type in 3 zones such as the tip of the tongue, in three zones, whether the pixel count of judging the tongue fur type that pixel count is maximum respectively and the ratio of whole lingual surface sum of all pixels determine the tongue fur feature in this district less than a certain threshold value;
5) carry out thickness of the tongue coating quantitative test and description, tongue color, tongue fur look type and quantity according to pixel and neighborhood interior pixel thereof, be divided into " no tongue ", " thin tongue ", " middle thick coating ", " thick coating " 4 types, thickness type and corresponding quantity according to whole tongue body zone interior pixel, calculate the thickness of the tongue coating index, describe thickness of the tongue coating according to the thickness index;
6) adopt literal and pseudocolour picture display analysis result.
2, the traditional Chinese medical science tongue color based on multi-class support vector machine according to claim 1, coating colour, thickness of the tongue coating analytical approach, the description of its tongue color, coating colour, thickness of the tongue coating is characterised in that, after pixel identification is finished, add up in each subregion behind various types of number of picture elements, make N
i jRepresent j class pixel number in the i district, i=1 wherein, 5 represent in the root of the tongue, the tongue respectively, 5 zones such as the tip of the tongue, tongue left side, tongue right side, j=1,15 expression tongue natures are light, tongue nature is light red, tongue nature is red, dimly red tongue, tongue nature deep red red, tongue nature is dark violet, 15 kinds of tongue natures such as thin and whitish fur, white tongue, white thick coating, thin and yellowish fur, yellow tongue, yellow thick coating, grayish fur, brown tongue, black tongue and tongue fur type, on this basis, according to the lingual diagnosis custom of Traditional Chinese Medicine experts, carry out the description of tongue color, coating colour, thickness of the tongue coating;
The describing method of tongue color is the color according to the tongue nature type specification tongue nature of tongue lateral areas and the tip of the tongue district pixel, calculate the sum of all pixels that belongs to each tongue nature type in 3 zones such as the tip of the tongue, tongue left side, tongue right side, the tongue nature type that pixel count is maximum is the tongue nature feature of this tongue image, and detailed process is:
1) calculate the total area (sum of pixel) A,
2) calculate the sum of 6 kinds of tongue nature types of tongue side and the tip of the tongue district respectively
3) order
J then
MaxThe tongue nature type of representative is the tongue nature feature of this tongue image,
4) the area ratio of the dark violet tongue nature of calculating
If R
6>θ
pAnd j
Max≠ 6, then increase and describe " local dark violet ", wherein θ
pFor according to the experiment preset threshold,
5) the area ratio in calculating the tip of the tongue district " tongue nature is red "
If R
3>θ
rAnd j
Max≠ 3, then increase and describe " red tip of tongue ", wherein θ
rFor according to the experiment preset threshold;
The describing method of coating colour is for calculating in the root of the tongue, the tongue, belonging to the sum of all pixels of each tongue fur type in 3 zones such as the tip of the tongue, in three zones, whether the pixel count of judging the tongue fur type that pixel count is maximum respectively and the ratio of whole lingual surface sum of all pixels be less than a certain threshold value, determine the tongue fur feature in this district, concrete grammar is as follows:
1) calculates the total area (sum of the pixel) A of lingual surface, in the root of the tongue, the tongue, the area A in the tip of the tongue district
1, i=1 ..., 3,
2) in the root of the tongue, the tongue, the tip of the tongue district, calculate respectively
I=1 wherein, 2,3,
J then
Max iThe tongue fur type of representative is the tongue fur feature in this district,
3) calculate i=1, j in 2,3 districts
Max iThe area ratio that the tongue fur pixel is shared
4) respectively in the root of the tongue, the tongue, the tip of the tongue district is described, for root of the tongue district i=1, if
Then being described as " the few tongue of the root of the tongue ", is j otherwise describe the root of the tongue
Max 1The tongue fur of type, for distinguishing i=2 in the tongue, if
Then being described as " few tongue in the tongue ", is j otherwise describe in the tongue
Max 2The tongue fur of type, for the tip of the tongue district i=3, if
Then being described as " the few tongue of the tip of the tongue ", is j otherwise describe the tip of the tongue
Max 3The tongue fur of type, wherein θ
1, θ
2, θ
3For according to the experiment preset threshold, 5) for distinguishing in the root of the tongue, the tongue, calculate the second main type tongue fur j respectively
Sec i, i=1,2, and calculate j
Sec iType pixels area and j
Max iThe area ratio of type
If
Then the i district increases description second main type tongue fur, the wherein θ
SecFor according to the experiment preset threshold;
The analysis and the description of thickness of the tongue coating are divided into two parts: thickness of the tongue coating classification and thickness of the tongue coating quantification, concrete steps are: 1) read the classification number of tongue body pixel and 5 * 5 neighborhoods thereof, 2) establish c
1, c
2For according to experiment preset threshold, and c
1>c
2, calculate the number of picture elements S that belongs to the tongue nature type in these pixel 5 * 5 neighborhoods
Body, if S
Body>c
1, then the thickness of the tongue coating type of this point is decided to be " no tongue ", 3) if c
1〉=S
Body>c
2, then the thickness of the tongue coating type of this point is decided to be " thin tongue ", 4) if S
Body≤ c
2, then the tongue body and tongue coating type j according to this point determines the thickness type, if tongue nature is j<7, then be " thin tongue ", if j=7,10 then is " a thin tongue ", if j=8,11,13,14 then is " a middle thick coating ", if j=9,12,15, then be " thick coating " 5) repeat 1)~3) step, finish up to all processes pixel, 6) according to the result of pixel identification, calculate the thick index T of tongue of view picture image, as the quantification result that thickness of the tongue coating is analyzed, computing method are:
A represents the pixel sum in tongue body district in the formula,
Expression is to all pixel summations in the tongue body zone, j
kThe thickness type of representing k pixel gets 0,1,2,3 respectively corresponding to " no tongue ", " thin tongue ", " middle thick coating ", " thick coating ",
Be weights, determine that according to the experience of experiment and Traditional Chinese Medicine experts the thick index of tongue can be described the general thickness of tongue fur.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB021037957A CN1162798C (en) | 2002-03-25 | 2002-03-25 | Chinese medicine tongue colour, fur colour and tongue fur thickness analysis method based on multiclass support vector machine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB021037957A CN1162798C (en) | 2002-03-25 | 2002-03-25 | Chinese medicine tongue colour, fur colour and tongue fur thickness analysis method based on multiclass support vector machine |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1367454A CN1367454A (en) | 2002-09-04 |
CN1162798C true CN1162798C (en) | 2004-08-18 |
Family
ID=4739901
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB021037957A Expired - Fee Related CN1162798C (en) | 2002-03-25 | 2002-03-25 | Chinese medicine tongue colour, fur colour and tongue fur thickness analysis method based on multiclass support vector machine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1162798C (en) |
Families Citing this family (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR100506085B1 (en) * | 2002-12-28 | 2005-08-05 | 삼성전자주식회사 | Apparatus for processing tongue image and health care service apparatus using tongue image |
CN1327376C (en) * | 2004-04-08 | 2007-07-18 | 上海交通大学 | Soft measuring meter moduling method based on supporting vector machine |
CN100409804C (en) * | 2006-01-19 | 2008-08-13 | 上海交通大学 | Tongue color automatic recognition method |
CN100405402C (en) * | 2006-08-18 | 2008-07-23 | 清华大学 | Coated tongue division and extracting method for colored digital photo of tongue |
CN100418469C (en) * | 2006-10-11 | 2008-09-17 | 哈尔滨工业大学 | Computerized disease sign analysis system based on tongue picture characteristics |
CN103705216B (en) * | 2013-12-16 | 2015-05-20 | 北京工业大学 | Tongue colour sensing and quantitative classification method combining equal interval scale with classificatory scale in traditional Chinese medicine |
CN103745217B (en) * | 2013-12-31 | 2017-02-15 | 北京工业大学 | Automatic analysis method of tongue color and coating color in traditional Chinese medicine based on image retrieval |
CN104586397B (en) * | 2015-01-26 | 2017-01-18 | 北京工业大学 | Traditional Chinese medicine tongue color classification perception quantification method combined by equal sense distance method and cluster analysis |
CN105160346B (en) * | 2015-07-06 | 2019-02-01 | 上海大学 | A kind of greasy recognition methods of curdy fur on tongue based on texture and distribution characteristics |
CN106980815A (en) * | 2017-02-07 | 2017-07-25 | 王俊 | Facial paralysis objective evaluation method under being supervised based on H B rank scores |
CN107169502A (en) * | 2017-03-22 | 2017-09-15 | 广东工业大学 | A kind of option recognition methods based on decision tree |
CN111340052A (en) * | 2018-12-03 | 2020-06-26 | 深圳市前海安测信息技术有限公司 | Tongue tip red detection device and method for tongue diagnosis in traditional Chinese medicine and computer storage medium |
CN110210319A (en) * | 2019-05-07 | 2019-09-06 | 平安科技(深圳)有限公司 | Computer equipment, tongue body photo constitution identification device and storage medium |
CN110363072B (en) * | 2019-05-31 | 2023-06-09 | 正和智能网络科技(广州)有限公司 | Tongue picture identification method, tongue picture identification device, computer equipment and computer readable storage medium |
CN110189383B (en) * | 2019-06-27 | 2021-07-02 | 合肥云诊信息科技有限公司 | Traditional Chinese medicine tongue color and fur color quantitative analysis method based on machine learning |
CN110929740A (en) * | 2019-11-21 | 2020-03-27 | 中电健康云科技有限公司 | LGBM model-based tongue quality and tongue coating separation method |
CN110786834B (en) * | 2019-12-09 | 2022-04-05 | 中电健康云科技有限公司 | Fatty liver prediction model based on tongue picture characteristics and BMI index |
CN115661142B (en) * | 2022-12-14 | 2023-03-28 | 广东工业大学 | Tongue diagnosis image processing method, device and medium based on key point detection |
CN115954101A (en) * | 2023-03-14 | 2023-04-11 | 南京轶诺科技有限公司 | Health degree management system and management method based on AI tongue diagnosis image processing |
CN116777930B (en) * | 2023-05-24 | 2024-01-09 | 深圳汇医必达医疗科技有限公司 | Image segmentation method, device, equipment and medium applied to tongue image extraction |
-
2002
- 2002-03-25 CN CNB021037957A patent/CN1162798C/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN1367454A (en) | 2002-09-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1162798C (en) | Chinese medicine tongue colour, fur colour and tongue fur thickness analysis method based on multiclass support vector machine | |
CN107977671B (en) | Tongue picture classification method based on multitask convolutional neural network | |
CN107330889B (en) | A kind of Chinese medicine tongue color coating colour automatic analysis method based on convolutional neural networks | |
US20230419696A1 (en) | Image analysis method, apparatus, program, and learned deep learning algorithm | |
US10445557B2 (en) | Learning pixel visual context from object characteristics to generate rich semantic images | |
US10474874B2 (en) | Applying pixelwise descriptors to a target image that are generated by segmenting objects in other images | |
CN109447998B (en) | Automatic segmentation method based on PCANet deep learning model | |
WO2021139258A1 (en) | Image recognition based cell recognition and counting method and apparatus, and computer device | |
CN103518224B (en) | Method for analysing microbial growth | |
US20220343178A1 (en) | Method and system for performing non-invasive genetic testing using an artificial intelligence (ai) model | |
CN1759804A (en) | Intelligent analyzing and differentiating method of herbalist doctor through integrated references form four parts of diagnosis | |
CN1761205A (en) | System for detecting eroticism and unhealthy images on network based on content | |
CN112580748B (en) | Method for counting classified cells of stain image | |
Vogado et al. | Unsupervised leukemia cells segmentation based on multi-space color channels | |
CN1746901A (en) | Computer intelligent process and processor for early lung cancer cell smear color image information | |
Hossain et al. | Tissue artifact segmentation and severity assessment for automatic analysis using wsi | |
Rachna et al. | Detection of Tuberculosis bacilli using image processing techniques | |
CN1908984A (en) | Coated tongue division and extracting method for colored digital photo of tongue | |
Huang et al. | HEp-2 cell images classification based on textural and statistic features using self-organizing map | |
CN1931085A (en) | Tongue coating and tongue proper color extracting and classifying method based on pixel | |
CN112446421A (en) | Silkworm cocoon counting and identifying method based on machine vision | |
CN114037868B (en) | Image recognition model generation method and device | |
JPH11507565A (en) | Image improvement apparatus and method | |
CN115880245A (en) | Self-supervision-based breast cancer disease classification method | |
CN111401119A (en) | Classification of cell nuclei |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
C19 | Lapse of patent right due to non-payment of the annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |