CN109636864A - A kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks - Google Patents
A kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks Download PDFInfo
- Publication number
- CN109636864A CN109636864A CN201811558699.8A CN201811558699A CN109636864A CN 109636864 A CN109636864 A CN 109636864A CN 201811558699 A CN201811558699 A CN 201811558699A CN 109636864 A CN109636864 A CN 109636864A
- Authority
- CN
- China
- Prior art keywords
- tongue
- image
- pixel
- region
- original image
- 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.)
- Pending
Links
- 238000012937 correction Methods 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 28
- 230000011218 segmentation Effects 0.000 claims abstract description 44
- 230000008439 repair process Effects 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims description 30
- 238000013135 deep learning Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 2
- 238000003745 diagnosis Methods 0.000 abstract description 16
- 230000000694 effects Effects 0.000 abstract description 6
- 238000005457 optimization Methods 0.000 abstract description 3
- 239000007787 solid Substances 0.000 abstract description 2
- 210000002105 tongue Anatomy 0.000 description 126
- 239000003814 drug Substances 0.000 description 6
- 201000010099 disease Diseases 0.000 description 5
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 5
- 210000001835 viscera Anatomy 0.000 description 5
- 230000007613 environmental effect Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 210000005181 root of the tongue Anatomy 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000037237 body shape Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000002425 crystallisation Methods 0.000 description 1
- 230000008025 crystallization Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 238000012432 intermediate storage Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 210000000952 spleen Anatomy 0.000 description 1
- 210000002784 stomach Anatomy 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- 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
- G06T2207/10024—Color image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks, which comprises color correction is carried out to original image according to preset rules；It is split according to tongue parted pattern trained in advance to by the tongue body in the original image of color correction, obtains the binary image of tongue body segmentation；The connected region in the binary image of the tongue body segmentation is detected, and confirms largest connected region according to preset rules, obtains largest connected area image；Repair operation is carried out to the largest connected area image according to preset rules, obtains final tongue cut zone；The region of the corresponding original image after color correction of final tongue cut zone, as tongue segmentation result.The method and system are operated by the correction to original image rgb value and based on depth convolutional neural networks and a variety of optimizations, obtain more accurate tongue segmented image；Solid ground foundation has been established to improve subsequent lingual diagnosis diagnosis effect.
Description
Technical field
The present invention relates to field of medical image processing, more particularly, to one kind based on color correction and depth convolution mind
Tongue dividing method and system through network.
Background technique
Chinese medicine is the rarity of our Chinese nation, is the wisdom crystallization much constantly improve for people by thousands of years.With
The development in epoch and social progress and Chinese medicine are preventiveed treatment of disease being rooted in the hearts of the people of theory, and Chinese medicine and modern science and technology combine, and generate
A series of modernization achievement.In addition to Chinese medicine modernization extract and production, the diagnostic method of Chinese medicine also towards automation,
Digitized direction is developed.As described in ancient and Modern Medicine: " four word of the four methods of diagnosis is really the guiding principle of doctor." i.e. four methods of diagnosis group
At the four methods of diagnosis of tcm diagnosis." the dirty piece of Ling Shu Miraculous Pivot or Divine Axis sheet ": " depending on answering outside it, to know its internal organ, then know institute's disease." know observation tool
There is very important effect.Observation can be divided into facial diagnosis and lingual diagnosis." Tongue Differentiation Guide ": " actual situation of the distinguishable internal organs of tongue examination matter, depending on tongue
Tongue fur can examine the shallow deep of the six external factors which cause diseases." tongue be the heart seedling, wait except spleen, tongue fur is given birth to by stomach Qi.Internal organs are connected by passages through which vital energy circulates and tongue, internal organs
Lesion can reflect in tongue nature and tongue fur.Lingual diagnosis Main Diagnosis tongue nature and the form of tongue fur, color etc. judge disease with this
Property, the shallow depth of patient's condition, the prosperity and decline of qi and blood and the actual situation of internal organs etc. of disease.
In recent years, with the gradually development of image processing techniques, the artificial intelligence technologys such as machine learning and deep learning
Constantly mature, depth convolutional neural networks start to be applied among TCM tongue diagnosis, and produce a variety of methods.Wherein, tongue body point
Cutting is processing step very crucial in lingual diagnosis system.Since tongue body shape is different, tongue body color very close to lip and
Face color causes existing many tongue body segmentation effects unsatisfactory, and traditional tongue dividing method all has certain limitation
Property, may generate tongue edge has missing, the root of the tongue to have situations such as missing, tongue body centre hole, adhesion lip and chin；Very
To because of shooting environmental colour cast, cause not to be partitioned into tongue at all；These all directly affect the feasibility of tongue reconstruction
With accuracy.
Summary of the invention
In order to solve existing tongue segmentation existing for background technique because of shooting environmental colour cast and image processing effect difference etc.
Reason influences the problem of tongue segmentation quality, and the present invention provides a kind of tongues based on color correction and depth convolutional neural networks point
Segmentation method and system；The method and system by the correction to original image rgb value and are based on depth convolutional neural networks
It is operated with a variety of optimizations, obtains more accurate tongue segmented image；Described one kind is based on color correction and depth convolutional neural networks
Tongue dividing method include:
Color correction is carried out to original image according to preset rules；
It is split, is obtained to by the tongue body in the original image of color correction according to tongue parted pattern trained in advance
The binary image of tongue body segmentation；The tongue parted pattern is obtained according to the training of depth convolutional neural networks；
The connected region in the binary image of the tongue body segmentation is detected, and confirms largest connected area according to preset rules
Domain obtains largest connected area image；
Repair operation is carried out to the largest connected area image according to preset rules, obtains final tongue cut zone；
The region of the corresponding original image after color correction of final tongue cut zone, as tongue segmentation result.
It is further, described that color correction is carried out to original image according to preset rules, comprising:
The RGB color value information of original image top face skin area is obtained, and confirms the RGB mean value in the region；
According to preset constant and the RGB mean value in the region, the RGB in entire image after each pixel correction is calculated
Value；According to the color of the rgb value adjustment whole picture original image after each pixel correction.
Further, the calculation formula of the rgb value after each pixel correction are as follows:
Wherein, R`, G` and B` are the rgb value after the correction of original image top face skin area, and R, G and B are original
The rgb value of beginning image current pixel, it is describedAndFor the RGB mean value of original image top face skin area；a,b
And c is preset constant.
Further, the tongue parted pattern trained in advance is under Caffe deep learning frame, according to training sample
The training of this progress FCN depth convolutional neural networks obtains；
The training sample is obtained by carrying out tongue body mark to sample image in advance, includes tongue body in the training sample
Pixel and non-tongue body pixel.
Further, before the connected region in the binary image for detecting the tongue body segmentation, the method also includes:
To the tongue body segmentation binary image carry out out operation, eliminate the binary image profile burr and
Less than the narrow interruption of pre-set dimension.
Further, before carrying out repair operation to the largest connected area image according to preset rules, the method
Further include:
Closed operation is carried out to the largest connected area image, eliminates and is less than default ruler in the largest connected area image
Very little narrow interruption and the cavity less than pre-set dimension.
Further, the connected region in the binary image of the detection tongue body segmentation, it is true according to preset rules
Recognizing largest connected region includes:
Confirm that each pixel is in eight unconnected pixels and the pixel centered on the pixel in the binary image
Whether gray value having the same；It will be with same grayscale value and connected pixel is denoted as a connected region；
According to the pixel quantity judgement in each region, maximum connected region is obtained.
A kind of tongue segmenting system based on color correction and depth convolutional neural networks includes:
Color correction unit, the color correction unit are used to carry out color correction to original image according to preset rules；
Tongue parted pattern unit, the tongue parted pattern unit are used for according to tongue parted pattern trained in advance to by face
Tongue body in the original image of color correction unit correction is split, the binary image of output tongue body segmentation；The tongue segmentation
Model is obtained according to the training of depth convolutional neural networks；
Connected region computing unit, the connected region computing unit are used to detect the tongue parted pattern unit output
Connected region in the binary image of tongue body segmentation, and largest connected region is confirmed according to preset rules, it exports largest connected
Area image；
Computing unit is repaired, what the repairing computing unit exported the connected region computing unit according to preset rules
Largest connected area image carries out repair operation, obtains final tongue cut zone；
Tongue divides output unit, and the tongue segmentation output unit is used for the final tongue cut zone is corresponding in color school
The region of original image after just exports tongue segmentation result.
Further, the color correction unit is used to obtain the RGB color value of original image top face skin area
Information, and confirm the RGB mean value in the region；
The color correction unit is used for the RGB mean value according to preset constant and the region, calculates entire image
In rgb value after each pixel correction；According to the color of the rgb value adjustment whole picture original image after each pixel correction.
Further, the tongue parted pattern that the tongue parted pattern unit is trained in advance is in Caffe deep learning frame
Under, the training of FCN depth convolutional neural networks is carried out according to training sample and is obtained；
The training sample is obtained by carrying out tongue body mark to sample image in advance, includes tongue body in the training sample
Pixel and non-tongue body pixel.
Further, the system also includes open operation computing unit；
It is described open operation computing unit be used for the tongue parted pattern unit output tongue body segmentation binary image into
Row opens operation, eliminates the profile burr of the binary image and the narrow interruption less than pre-set dimension；And it will open
Binary image after operation is exported to the connected region computing unit.
Further, the system also includes closed operation computing units；
The largest connected area image that the closed operation computing unit is used to export the connected region computing unit into
Row closed operation is eliminated and is less than the narrow interruption of pre-set dimension in the largest connected area image and less than pre-set dimension
Cavity；And the largest connected area image after progress closed operation is exported to the repairing computing unit.
Further, the connected region computing unit is for confirming that each pixel is with this in the binary image
Eight unconnected pixels and pixel gray value whether having the same centered on pixel；There to be same grayscale value and be connected
Pixel be denoted as a connected region；
The connected region computing unit judges according to the pixel quantity in each region, obtains maximum connected region.
The invention has the benefit that technical solution of the present invention, gives one kind based on color correction and depth convolution
The tongue dividing method and system of neural network, the method and system are solved tongue body and are shone by the correction to original image rgb value
Piece color difference problem caused by because of shooting environmental colour cast difference；The binary picture of tongue segmentation is obtained by depth convolutional neural networks
Picture, and operation, connected region calculating, closed operation and repair operation etc. are opened according to preset, binary image is carried out into one
Step optimization plays denoising, the smooth of the edge and the effect for repairing tongue body broken partial section；The accuracy that tongue is divided is promoted, is
It improves subsequent lingual diagnosis diagnosis effect and has established solid ground foundation.
Detailed description of the invention
By reference to the following drawings, exemplary embodiments of the present invention can be more fully understood by:
Fig. 1 is a kind of tongue segmentation side based on color correction Yu depth convolutional neural networks of the specific embodiment of the invention
The flow chart of method；
Fig. 2 is that a kind of tongue based on color correction and depth convolutional neural networks of the specific embodiment of the invention divides system
The structure chart of system.
Specific embodiment
Exemplary embodiments of the present invention are introduced referring now to the drawings, however, the present invention can use many different shapes
Formula is implemented, and is not limited to the embodiment described herein, and to provide these embodiments be at large and fully disclose
The present invention, and the scope of the present invention is sufficiently conveyed to person of ordinary skill in the field.Show for what is be illustrated in the accompanying drawings
Term in example property embodiment is not limitation of the invention.In the accompanying drawings, identical cells/elements use identical attached
Icon note.
Unless otherwise indicated, term (including scientific and technical terminology) used herein has person of ordinary skill in the field
It is common to understand meaning.Further it will be understood that with the term that usually used dictionary limits, should be understood as and its
The context of related fields has consistent meaning, and is not construed as Utopian or too formal meaning.
Fig. 1 is a kind of tongue segmentation side based on color correction Yu depth convolutional neural networks of the specific embodiment of the invention
The flow chart of method, as shown in Figure 1, which comprises
Step 110, color correction is carried out to original image according to preset rules；
Tongue dividing method involved in the application is the antecedent basis for realizing lingual diagnosis, the patient of lingual diagnosis is needed by inputting
Tongue body photo, the tongue body in photo image recognition and individually split, more it is subsequent objectively to tongue body into
Row analysis；
It is the tongue body original image shading value of input, bright because of shooting environmental disunity and in the tongue body photo actually entered
There are larger differences for degree, contrast etc., and the prior art is mainly by being converted to tongue body according to preset mode for original image
The binary image of segmentation, and if similarly there are larger difference, outputs for shading value, brightness, contrast of original image etc.
Binary image can also have differences, the original image under the poor environment in part can not effectively be partitioned into tongue body；
For the original image of tongue body, tongue shooting is stretched out from mouth for people, there are face skins on the top of original image
People is arranged for the crowd of similar skin color difference because being not much different for specific region crowd's face skin color difference in the region of skin
Face skin region preset constant (white people, yellow, the corresponding preset constant of black race's ethnic group is such as respectively set, it is described
The setting of preset constant can do further subdivision according to demand), it is described that color correction is carried out to original image according to preset rules
Can further comprise:
The RGB color value information of original image top face skin area is obtained, and confirms the RGB mean value in the region；
According to preset constant and the RGB mean value in the region, the RGB in entire image after each pixel correction is calculated
Value；
According to the color of the rgb value adjustment whole picture original image after each pixel correction.
Further, the calculation formula of the rgb value after each pixel correction are as follows:
Wherein, R`, G` and B` are the rgb value after the correction of original image top face skin area, and R, G and B are original
The rgb value of beginning image current pixel, it is describedAndFor the RGB mean value of original image top face skin area；a,b
And c is preset constant.
In the present embodiment, corresponding yellow, preset constant value is a=1.55, b=1.15, c=0.95.
Step 120, it is carried out according to tongue parted pattern trained in advance to by the tongue body in the original image of color correction
Segmentation obtains the binary image of tongue body segmentation；The tongue parted pattern is obtained according to the training of depth convolutional neural networks；
Further, the tongue parted pattern trained in advance is under Caffe deep learning frame, according to training sample
The training of this progress FCN depth convolutional neural networks obtains；
The training sample by advance to sample image carry out tongue body mark (such as by multidigit expert to tongue body carry out
More wheel marks obtain) it obtains, it include tongue body pixel and non-tongue body pixel in the training sample；
Further, for the binary image of output, operation is carried out out, the profile burr of the binary image is eliminated
And the narrow interruption less than pre-set dimension.The first binary image of original image is obtained, tongue body contour edge
There are the thin protrusions such as burr, or there are narrow interruptions (such as the binary conversion treatment for tongue body part similar in tooth
When), these defects are eliminated by opening operation.
In the present embodiment, operation is opened to even numbers binary image by using the function imopen realization in Matlab.
Step 130, the connected region in the binary image of the tongue body segmentation is detected, and most according to preset rules confirmation
Big connected region obtains largest connected area image；
In the present embodiment, connected region is detected by the way of 8 connections；
Confirm that each pixel is in eight unconnected pixels and the pixel centered on the pixel in the binary image
Whether gray value having the same；It will be with same grayscale value and connected pixel is denoted as a connected region；
In the present embodiment, connected region is obtained using the function bwlabel in Matlab.
According to the pixel quantity judgement in each region, maximum connected region is obtained；
Equally, in the present embodiment, the face of each region in image can be measured by the regionprops function in Matlab
Product.
The maximum connected region corresponds to tongue and cuts corresponding image-region；
Further, closed operation is carried out to the largest connected area image, eliminated in the largest connected area image
Less than the narrow interruption of pre-set dimension and less than the cavity of pre-set dimension.In the shooting of original image, because tongue body stretches
Degree is different, it is understood that there may be depression points；Or because there is exposure point, bad point in shooting reason；Defect is eliminated by closed operation.
In the present embodiment, closed operation is realized by using the imclose function in Matlab.
Step 140, repair operation is carried out to the largest connected area image according to preset rules, obtains final tongue segmentation
Region；
The repair operation includes hole or the root of the tongue intercalary delection part repaired among largest connected region；
The largest connected region corresponds to the tongue body part in original image, the i.e. final area of tongue segmentation；If therefore should
The largest connected closed hole of region intermediate storage, the corresponding part of the hole also must be a part of tongue body, Ke Nengyin
It is depression points caused by stretching, extension sufficiently for tongue body or exposure point, bad point occurs because shooting reason, by repair operation, supplements this portion
Divide hole, to guarantee the integrality of tongue cut zone；
And due to the habit that lolls of patient, Great Wall root of the tongue middle section can be recessed so that this part on the original image compared with
Secretly, can not effectively be divided；By the setting to tongue body preset shape, the recess of root of the tongue excalation is supplemented, automatically with reality
The now integrality of final tongue cut zone.
Step 150, the region of the corresponding original image after color correction of the final tongue cut zone, as tongue are divided
As a result.
Fig. 2 is that a kind of tongue based on color correction and depth convolutional neural networks of the specific embodiment of the invention divides system
The structure chart of system.As shown in Fig. 2, the system comprises:
Color correction unit 210, the color correction unit 210 are used to carry out color to original image according to preset rules
Correction；
Further, the color correction unit 210 is used to obtain the RGB color of original image top face skin area
Value information, and confirm the RGB mean value in the region；
The color correction unit 210 is used for the RGB mean value according to preset constant and the region, calculates whole picture figure
Rgb value as in after each pixel correction；According to the face of the rgb value adjustment whole picture original image after each pixel correction
Color.
Tongue parted pattern unit 220, the tongue parted pattern unit 220 are used for according to tongue parted pattern pair trained in advance
The tongue body in original image corrected by color correction unit 210 is split, the binary image of output tongue body segmentation；Institute
Tongue parted pattern is stated to be obtained according to the training of depth convolutional neural networks；
Further, the tongue parted pattern that the tongue parted pattern unit 220 is trained in advance, is in Caffe deep learning
Under frame, the training of FCN depth convolutional neural networks is carried out according to training sample and is obtained；
The training sample is obtained by carrying out tongue body mark to sample image in advance, includes tongue body in the training sample
Pixel and non-tongue body pixel.
Connected region computing unit 230, the connected region computing unit 230 is for detecting the tongue parted pattern unit
Connected region in the binary image of the tongue body segmentation of 220 outputs, and largest connected region, output are confirmed according to preset rules
Largest connected area image；
Further, the connected region computing unit 230 is for confirming that each pixel exists in the binary image
Eight unconnected pixels and pixel gray value whether having the same centered on the pixel；To have same grayscale value and
Connected pixel is denoted as a connected region；
The connected region computing unit 230 judges according to the pixel quantity in each region, obtains maximum connected region
Domain.
Computing unit 240 is repaired, the repairing computing unit 240 is according to preset rules to the connected region computing unit
The largest connected area image of 230 outputs carries out repair operation, obtains final tongue cut zone；
Tongue divides output unit 250, and the tongue segmentation output unit 250 is used to correspond to the final tongue cut zone
The region of original image after color correction exports tongue segmentation result.
Further, the system also includes open operation computing unit；
The binary image opened operation computing unit and be used to divide the tongue parted pattern unit 220 output tongue body
Operation is carried out out, the profile burr of the binary image and the narrow interruption less than pre-set dimension are eliminated；And it will carry out
Binary image after opening operation is exported to the connected region computing unit 230.
Further, the system also includes closed operation computing units；
The closed operation computing unit is used for the largest connected area image exported to the connected region computing unit 230
Closed operation is carried out, the narrow interruption for being less than pre-set dimension in the largest connected area image is eliminated and is less than pre-set dimension
Cavity；And the largest connected area image after progress closed operation is exported to the repairing computing unit 240.
In the instructions provided here, numerous specific details are set forth.It is to be appreciated, however, that the implementation of the disclosure
Example can be practiced without these specific details.In some instances, well known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this specification.
Those skilled in the art will understand that can be carried out adaptively to the module in the equipment in embodiment
Change and they are arranged in one or more devices different from this embodiment.It can be the module or list in embodiment
Member or component are combined into a module or unit or component, and furthermore they can be divided into multiple submodule or subelement or
Sub-component.Other than such feature and/or at least some of process or unit exclude each other, it can use any
Combination is to all features disclosed in this specification (including adjoint claim, abstract and attached drawing) and so disclosed
All process or units of what method or apparatus are combined.Unless expressly stated otherwise, this specification is (including adjoint power
Benefit require, abstract and attached drawing) disclosed in each feature can carry out generation with an alternative feature that provides the same, equivalent, or similar purpose
It replaces.The step of being related in this specification, which numbers, is only used for distinguishing each step, and the time being not limited between each step
Or the relationship of logic, restriction unless the context clearly, otherwise the relationship between each step includes the case where various possible.
In addition, it will be appreciated by those of skill in the art that although some embodiments described herein include other embodiments
In included certain features rather than other feature, but the combination of the feature of different embodiments means to be in the disclosure
Within the scope of and form different embodiments.For example, embodiment claimed in detail in the claims is one of any
Can in any combination mode come using.
The various component embodiments of the disclosure can be implemented in hardware, or to run on one or more processors
Software module realize, or be implemented in a combination thereof.The disclosure is also implemented as executing side as described herein
The some or all equipment or system program (for example, computer program and computer program product) of method.It is such
It realizes that the program of the disclosure can store on a computer-readable medium, or can have the shape of one or more signal
Formula.Such signal can be downloaded from an internet website to obtain, and perhaps be provided on the carrier signal or with any other shape
Formula provides.
The disclosure is limited it should be noted that above-described embodiment illustrates rather than the disclosure, and ability
Field technique personnel can be designed alternative embodiment without departing from the scope of the appended claims.Word "comprising" is not arranged
Except there are element or steps not listed in the claims.Word "a" or "an" located in front of the element does not exclude the presence of more
A such element.The disclosure can be by means of including the hardware of several different elements and by means of properly programmed calculating
Machine is realized.If several in these systems can be by same in the unit claim for listing dry systems
Hardware branch embodies.
The above is only the specific embodiment of the disclosure, it is noted that for the ordinary skill people of this field
Member for, do not depart from the disclosure spirit under the premise of, can make several improvements, modify and deform, these improve, modification,
It is regarded as falling within the scope of protection of this application with deformation.
Claims (13)
1. a kind of tongue dividing method based on color correction Yu depth convolutional neural networks, which comprises
Color correction is carried out to original image according to preset rules；
It is split according to tongue parted pattern trained in advance to by the tongue body in the original image of color correction, obtains tongue body
The binary image of segmentation；The tongue parted pattern is obtained according to the training of depth convolutional neural networks；
The connected region in the binary image of the tongue body segmentation is detected, and confirms largest connected region according to preset rules,
Obtain largest connected area image；
Repair operation is carried out to the largest connected area image according to preset rules, obtains final tongue cut zone；
The region of the corresponding original image after color correction of final tongue cut zone, as tongue segmentation result.
2. according to the method described in claim 1, it is characterized by: described carry out color school to original image according to preset rules
Just, comprising:
The RGB color value information of original image top face skin area is obtained, and confirms the RGB mean value in the region；
According to preset constant and the RGB mean value in the region, the rgb value in entire image after each pixel correction is calculated；
According to the color of the rgb value adjustment whole picture original image after each pixel correction.
3. according to the method described in claim 2, it is characterized in that, the calculation formula of the rgb value after each pixel correction
Are as follows:
Wherein, R`, G` and B` are the rgb value after the correction of original image top face skin area, and R, G and B are original graph
It is described as the rgb value of current pixelAndFor the RGB mean value of original image top face skin area；A, b and
C is preset constant.
4. according to the method described in claim 1, it is characterized by: the tongue parted pattern trained in advance, is in Caffe depth
It spends under learning framework, the training of FCN depth convolutional neural networks is carried out according to training sample and is obtained；
The training sample is obtained by carrying out tongue body mark to sample image in advance, includes tongue body pixel in the training sample
And non-tongue body pixel.
5. the method according to claim 1, wherein the company in the binary image for detecting the tongue body segmentation
Before logical region, the method also includes:
Operation is carried out out to the binary image of tongue body segmentation, the profile burr of the binary image is eliminated and is less than
The narrow interruption of pre-set dimension.
6. according to the method described in claim 1, it is characterized by: according to preset rules to the largest connected area image
Before carrying out repair operation, the method also includes:
Closed operation is carried out to the largest connected area image, eliminates and is less than pre-set dimension in the largest connected area image
Narrow interruption and the cavity less than pre-set dimension.
7. according to the method described in claim 1, it is characterized by: in the binary image of the detection tongue body segmentation
Connected region confirms that largest connected region includes: according to preset rules
Confirm each pixel in the binary image centered on the pixel eight unconnected pixels and the pixel whether
Gray value having the same；It will be with same grayscale value and connected pixel is denoted as a connected region；
According to the pixel quantity judgement in each region, maximum connected region is obtained.
8. a kind of tongue segmenting system based on color correction Yu depth convolutional neural networks, the system comprises:
Color correction unit, the color correction unit are used to carry out color correction to original image according to preset rules；
Tongue parted pattern unit, the tongue parted pattern unit are used for according to tongue parted pattern trained in advance to by color school
Tongue body in the original image of positive unit correction is split, the binary image of output tongue body segmentation；The tongue parted pattern
It is obtained according to the training of depth convolutional neural networks；
Connected region computing unit, the connected region computing unit are used to detect the tongue body of the tongue parted pattern unit output
Connected region in the binary image of segmentation, and largest connected region is confirmed according to preset rules, export largest connected region
Image；
Repair computing unit, the maximum that the repairing computing unit exports the connected region computing unit according to preset rules
Connected region image carries out repair operation, obtains final tongue cut zone；
Tongue divides output unit, and the tongue segmentation output unit is used for the final tongue cut zone is corresponding after color correction
Original image region, export tongue segmentation result.
9. system according to claim 8, it is characterised in that:
The color correction unit is used to obtain the RGB color value information of original image top face skin area, and confirms institute
State the RGB mean value in region；
The color correction unit is used for the RGB mean value according to preset constant and the region, calculates every in entire image
Rgb value after a pixel correction；According to the color of the rgb value adjustment whole picture original image after each pixel correction.
10. system according to claim 8, it is characterised in that: the tongue segmentation that the tongue parted pattern unit is trained in advance
Model is to carry out the training of FCN depth convolutional neural networks under Caffe deep learning frame according to training sample and obtain；
The training sample is obtained by carrying out tongue body mark to sample image in advance, includes tongue body pixel in the training sample
And non-tongue body pixel.
11. system according to claim 8, it is characterised in that: the system also includes open operation computing unit；
The operation computing unit of opening is used to open the binary image of tongue parted pattern unit output tongue body segmentation
Operation, eliminates the profile burr of the binary image and the narrow interruption less than pre-set dimension；And operation will be carried out open
Binary image afterwards is exported to the connected region computing unit.
12. system according to claim 8, it is characterised in that: the system also includes closed operation computing units；
The largest connected area image that the closed operation computing unit is used to export the connected region computing unit closes
The narrow interruption for being less than pre-set dimension in the largest connected area image and the sky less than pre-set dimension are eliminated in operation
Hole；And the largest connected area image after progress closed operation is exported to the repairing computing unit.
13. system according to claim 8, it is characterised in that:
The connected region computing unit is for confirming that each pixel is centered on the pixel in the binary image
Eight unconnected pixels and pixel gray value whether having the same；It will be with same grayscale value and connected pixel is denoted as one
A connected region；
The connected region computing unit judges according to the pixel quantity in each region, obtains maximum connected region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811558699.8A CN109636864A (en) | 2018-12-19 | 2018-12-19 | A kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811558699.8A CN109636864A (en) | 2018-12-19 | 2018-12-19 | A kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109636864A true CN109636864A (en) | 2019-04-16 |
Family
ID=66075594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811558699.8A Pending CN109636864A (en) | 2018-12-19 | 2018-12-19 | A kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109636864A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110298829A (en) * | 2019-06-19 | 2019-10-01 | 上海国民集团健康科技有限公司 | A kind of lingual diagnosis method, apparatus, system, computer equipment and storage medium |
| CN112489053A (en) * | 2020-11-26 | 2021-03-12 | 深圳市艾合芯科智慧医疗有限公司 | Tongue image segmentation method and device and storage medium |
| CN114511567A (en) * | 2022-04-20 | 2022-05-17 | 天中依脉(天津)智能科技有限公司 | Tongue body and tongue coating image identification and separation method |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW200623017A (en) * | 2004-12-31 | 2006-07-01 | Committee On Chinese Medicine And Pharmacy Dept Of Health | Tongue diagnosis systems and methods |
| CN103745217A (en) * | 2013-12-31 | 2014-04-23 | 北京工业大学 | Automatic analysis method of tongue color and coating color in traditional Chinese medicine based on image retrieval |
| CN104537379A (en) * | 2014-12-26 | 2015-04-22 | 上海大学 | High-precision automatic tongue partition method |
| CN104572538A (en) * | 2014-12-31 | 2015-04-29 | 北京工业大学 | K-PLS regression model based traditional Chinese medicine tongue image color correction method |
| CN104658003A (en) * | 2015-03-16 | 2015-05-27 | 北京理工大学 | Tongue image segmentation method and device |
| CN106023151A (en) * | 2016-05-09 | 2016-10-12 | 厦门大学 | Traditional Chinese medicine tongue manifestation object detection method in open environment |
| CN106295139A (en) * | 2016-07-29 | 2017-01-04 | 姹ゅ钩 | A kind of tongue body autodiagnosis health cloud service system based on degree of depth convolutional neural networks |
| CN107194937A (en) * | 2017-05-27 | 2017-09-22 | 厦门大学 | Tongue image partition method under a kind of open environment |
| CN107316307A (en) * | 2017-06-27 | 2017-11-03 | 北京工业大学 | A kind of Chinese medicine tongue image automatic segmentation method based on depth convolutional neural networks |
| CN107507250A (en) * | 2017-06-02 | 2017-12-22 | 北京工业大学 | A kind of complexion tongue color image color correction method based on convolutional neural networks |
| CN107977671A (en) * | 2017-10-27 | 2018-05-01 | 浙江工业大学 | A kind of tongue picture sorting technique based on multitask convolutional neural networks |
| CN108810509A (en) * | 2018-07-06 | 2018-11-13 | 北京中安未来科技有限公司 | A kind of image color correction method and device |
-
2018
- 2018-12-19 CN CN201811558699.8A patent/CN109636864A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW200623017A (en) * | 2004-12-31 | 2006-07-01 | Committee On Chinese Medicine And Pharmacy Dept Of Health | Tongue diagnosis systems and methods |
| CN103745217A (en) * | 2013-12-31 | 2014-04-23 | 北京工业大学 | Automatic analysis method of tongue color and coating color in traditional Chinese medicine based on image retrieval |
| CN104537379A (en) * | 2014-12-26 | 2015-04-22 | 上海大学 | High-precision automatic tongue partition method |
| CN104572538A (en) * | 2014-12-31 | 2015-04-29 | 北京工业大学 | K-PLS regression model based traditional Chinese medicine tongue image color correction method |
| CN104658003A (en) * | 2015-03-16 | 2015-05-27 | 北京理工大学 | Tongue image segmentation method and device |
| CN106023151A (en) * | 2016-05-09 | 2016-10-12 | 厦门大学 | Traditional Chinese medicine tongue manifestation object detection method in open environment |
| CN106295139A (en) * | 2016-07-29 | 2017-01-04 | 姹ゅ钩 | A kind of tongue body autodiagnosis health cloud service system based on degree of depth convolutional neural networks |
| CN107194937A (en) * | 2017-05-27 | 2017-09-22 | 厦门大学 | Tongue image partition method under a kind of open environment |
| CN107507250A (en) * | 2017-06-02 | 2017-12-22 | 北京工业大学 | A kind of complexion tongue color image color correction method based on convolutional neural networks |
| CN107316307A (en) * | 2017-06-27 | 2017-11-03 | 北京工业大学 | A kind of Chinese medicine tongue image automatic segmentation method based on depth convolutional neural networks |
| CN107977671A (en) * | 2017-10-27 | 2018-05-01 | 浙江工业大学 | A kind of tongue picture sorting technique based on multitask convolutional neural networks |
| CN108810509A (en) * | 2018-07-06 | 2018-11-13 | 北京中安未来科技有限公司 | A kind of image color correction method and device |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110298829A (en) * | 2019-06-19 | 2019-10-01 | 上海国民集团健康科技有限公司 | A kind of lingual diagnosis method, apparatus, system, computer equipment and storage medium |
| CN112489053A (en) * | 2020-11-26 | 2021-03-12 | 深圳市艾合芯科智慧医疗有限公司 | Tongue image segmentation method and device and storage medium |
| CN112489053B (en) * | 2020-11-26 | 2021-07-13 | 深圳市艾合芯科智慧医疗有限公司 | Tongue image segmentation method and device and storage medium |
| CN114511567A (en) * | 2022-04-20 | 2022-05-17 | 天中依脉(天津)智能科技有限公司 | Tongue body and tongue coating image identification and separation method |
| CN114511567B (en) * | 2022-04-20 | 2022-08-05 | 天中依脉(天津)智能科技有限公司 | Tongue body and tongue coating image identification and separation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109859203B (en) | Defect tooth image identification method based on deep learning | |
| CN110600122B (en) | Digestive tract image processing method and device and medical system | |
| CN109670510A (en) | A kind of gastroscopic biopsy pathological data screening system and method based on deep learning | |
| CN108765392B (en) | Digestive tract endoscope lesion detection and identification method based on sliding window | |
| CN109637660B (en) | Tongue diagnosis analysis method and system based on deep convolutional neural network | |
| CN109636864A (en) | A kind of tongue dividing method and system based on color correction Yu depth convolutional neural networks | |
| CN112967285B (en) | Chloasma image recognition method, system and device based on deep learning | |
| Qureshi et al. | Detection of glaucoma based on cup-to-disc ratio using fundus images | |
| CN109146897A (en) | Oral cavity radiation image method of quality control and device | |
| CN102567734B (en) | Specific value based retina thin blood vessel segmentation method | |
| CN107516312A (en) | A kind of Chinese medicine complexion automatic classification method using shallow-layer neutral net | |
| CN109087310A (en) | Dividing method, system, storage medium and the intelligent terminal of Meibomian gland texture region | |
| CN113139971A (en) | Tongue picture identification method and system based on artificial intelligence | |
| CN106570530A (en) | Extraction method for extracting hard exudates in ophthalmoscopic image | |
| CN109241963A (en) | Blutpunkte intelligent identification Method in capsule gastroscope image based on Adaboost machine learning | |
| CN114119950A (en) | Artificial intelligence-based oral cavity curved surface fault layer dental image segmentation method | |
| Maqsood et al. | Detection of macula and recognition of aged-related macular degeneration in retinal fundus images | |
| Chen et al. | Automated Detection System Based on Convolution Neural Networks for Retained Root, Endodontic Treated Teeth, and Implant Recognition on Dental Panoramic Images | |
| CN109949299A (en) | A kind of cardiologic medical image automatic segmentation method | |
| CN108765431A (en) | A kind of dividing method of image and its application in medical domain | |
| CN113539476A (en) | Stomach endoscopic biopsy Raman image auxiliary diagnosis method and system based on artificial intelligence | |
| CN117275677A (en) | Method for effectively identifying benign and malignant breast ultrasound image tumor | |
| CN112419246A (en) | Depth detection network for quantifying esophageal mucosa IPCLs blood vessel morphological distribution | |
| CN109711306A (en) | A kind of method and apparatus obtaining facial characteristics based on depth convolutional neural networks | |
| CN110610502A (en) | Automatic aortic arch region positioning and segmentation method based on CT image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |