CN110021406A - A kind of deep learning method based on multispectral camera - Google Patents
A kind of deep learning method based on multispectral camera Download PDFInfo
- Publication number
- CN110021406A CN110021406A CN201910248658.7A CN201910248658A CN110021406A CN 110021406 A CN110021406 A CN 110021406A CN 201910248658 A CN201910248658 A CN 201910248658A CN 110021406 A CN110021406 A CN 110021406A
- Authority
- CN
- China
- Prior art keywords
- spectrum picture
- spectrum
- band
- illness diagnostic
- illness
- 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
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013135 deep learning Methods 0.000 title claims abstract description 8
- 238000001228 spectrum Methods 0.000 claims abstract description 83
- 230000000694 effects Effects 0.000 claims abstract description 27
- 208000035473 Communicable disease Diseases 0.000 claims abstract description 24
- 208000015181 infectious disease Diseases 0.000 claims abstract description 23
- 238000003745 diagnosis Methods 0.000 claims abstract description 21
- 239000000284 extract Substances 0.000 claims abstract description 8
- 238000003384 imaging method Methods 0.000 claims description 40
- 238000012706 support-vector machine Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 12
- 230000003595 spectral effect Effects 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000011218 segmentation Effects 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 230000003287 optical effect Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 238000004321 preservation Methods 0.000 claims description 3
- 238000010183 spectrum analysis Methods 0.000 abstract description 8
- 239000000203 mixture Substances 0.000 abstract description 5
- 230000008676 import Effects 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 abstract 1
- 206010001557 Albinism Diseases 0.000 description 6
- 206010025135 lupus erythematosus Diseases 0.000 description 6
- 238000000701 chemical imaging Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 201000010099 disease Diseases 0.000 description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000003759 clinical diagnosis Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 206010019799 Hepatitis viral Diseases 0.000 description 1
- 241000700605 Viruses Species 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 238000000862 absorption spectrum Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010224 classification analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006806 disease prevention Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002458 infectious effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000002574 poison Substances 0.000 description 1
- 231100000614 poison Toxicity 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 208000017520 skin disease Diseases 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Biomedical Technology (AREA)
- Theoretical Computer Science (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Radiology & Medical Imaging (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Pathology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The deep learning method based on multispectral camera that the invention discloses a kind of, including carrying out light spectrum image-forming to different crowd, the gray value and characteristic wave bands of spectrum picture are obtained by image procossing, construct illness diagnostic characteristic vector, and import in illness diagnostic classification device and learnt, classified.In real-time diagnosis identification, obtain the spectrum picture of public arena activity scene in real time using multispectral camera, the filter center wavelength that wherein multispectral camera is selected is the characteristic wave bands of the characteristic wave bands of healthy population and the characteristic wave bands composition of patient groups, extract the gray value of the spectrum picture, again pull up illness diagnostic characteristic vector, illness diagnostic characteristic vector is imported into illness diagnostic classification device, to determine in the spectrum picture whether there are patient groups；Spectral analysis technique is applied to the real time monitoring in public arena activity scene to communicable disease for the first time by the present invention, meets the intellectually and automatically requirement that public arena identifies communicable disease.
Description
Technical field
The invention belongs to infectious disease diagnosis identification technology fields in public arena, more particularly to one kind is based on multispectral
The deep learning method of camera.
Background technique
Foreign countries have the clinical diagnosis that spectral analysis technique is applied to infectious disease, such as virus hepatitis, new city at present
The detection of the infectious diseases such as epidemic disease poison, wherein the clinical case based on absorption spectroanalysis technology, which is examined, most widely to be applied,
Its principle be using the feature of absorption spectrum pedigree possessed by infectious substance, come determine the property of the substance, structure or
Content；However, spectral analysis technique to be applied to infectious disease identification, diagnosis and the control of public arena, at present both at home and abroad not
It is studied and is applied, for the public arenas such as large-scale community, shopping center, food plaza, recreation center, Ren Yuanmi
Degree is higher, and flowing is larger, provides preferable route of transmission for infectious disease.
Under normal conditions, certain communicable diseases, such as surface characteristics of albinism, skin disease, patient are likely to be at
The also unobvious stage with the naked eye can not be identified and be handled, the control and prevention of disease being an impediment under public arena, but studies have shown that
The body surface of such communicable disease patient causes surface characteristics to change, the shape under optical spectrum imagers due to the infringement of virus
At the spectral signature wave band of spectrum picture have different from normal population, how spectral analysis technique to be applied under public arena
Identification, diagnosis and the control of special communicable disease become one of the research hotspot of field of spectral analysis technology instantly.
Summary of the invention
It is an object of the invention to: the present invention provides a kind of deep learning method based on multispectral camera, solves existing
Some spectral analysis techniques are only applied in the clinical examination of infectious disease, and are the knowledge of the communicable disease under public arena
Not, diagnosis provides a set of feasible technical solution with control.
The technical solution adopted by the invention is as follows:
A kind of deep learning method based on multispectral camera, includes the following steps:
Step 1: utilizing multi-optical spectrum imaging system, healthy population and patient groups are carried out respectively in darkroom multispectral
Imaging, obtains the spectrum picture of different crowd, filter center wave-length coverage involved in the multi-optical spectrum imaging system covers
Visible wavelength is to infrared wavelength；
Step 2: region segmentation being carried out to the spectrum picture, gray value is extracted respectively to each region；
Step 3: the band index value P of different each imaging bands of healthy population is calculated using band index methodi, according to it is each at
As channel filter central wavelength and the corresponding band index value Pi, obtain band index curve graph；
Step 4: choosing principle, the bigger imaging band of band index value, contained light according to band index characteristic wave bands
Spectrum information amount is also bigger, and the multi-optical spectrum imaging system captures out band index value office from the band index curve graph automatically
The maximum data point in portion, and the filter center wavelength of corresponding imaging band is saved, to form healthy population spectrum
The characteristic wave bands of image；
Step 5: repeating step 3-4, obtain the characteristic wave bands of patient groups' spectrum picture and preservation；
Step 6: the spectral information of the healthy population and patient groups being combined into initial characteristics collection, wherein the ash
Angle value is as test sample collection, the trained sample of the characteristic wave bands composition of the characteristic wave bands of the healthy population and the patient groups
This collection constitutes illness diagnostic characteristic vector, and constructs illness diagnostic classification device；
Step 7: when carrying out live communicable disease real-time diagnosis identification, obtaining public arena in real time using multispectral camera
The spectrum picture of activity scene, wherein the filter center wavelength that the multispectral camera is selected includes the spy of the healthy population
Wave band and the characteristic wave bands of the patient groups are levied, by the spectrum picture real-time Transmission of the public arena activity scene to prison
Control center, the monitoring center carries out image preprocessing to the spectrum picture of the public arena activity scene, and extracts institute
State the gray value of spectrum picture；
Step 8: for the field diagnostic in step 7, illness diagnostic characteristic vector is constructed, including the public arena
The illness diagnostic characteristic vector is imported the illness and diagnosed by the gray value and characteristic wave bands of the spectrum picture of activity scene
Classifier, the illness diagnostic classification device predict the illness diagnostic characteristic vector in real time, to determine the public field
Whether there are patient groups in the spectrum picture of conjunction activity scene.
Preferably, optical filter wavelength described in the step 1 be specially 425,475,509,515,558,578,620,
650, totally 14 channels 680,717,750,800,832 and 850nm, corresponding half-band width is 100 respectively, 100,20,10,5,
10,10,10,10,10,10,10,10 and 5nm.
Preferably, wherein the band index value P of different each imaging bands of healthy population is calculated in the step 3iSpecific packet
It includes:
Step 3.1: calculating the standard deviation of the gray value of different healthy population spectrum pictures；
Step 3.2: calculating the related coefficient between each imaging band of the multi-optical spectrum imaging system；
Step 3.3: the band index value P of each imaging band of healthy population is calculated using formula (1)i:
In formula: parameter σiFor the standard deviation of the i-th channel image of spectrum picture, parameter RiFor the phase between wave band i and wave band j
Relationship number.
Preferably, in the step 6, the illness diagnostic classification implement body is constructed are as follows: make 2 support vector machines series connection
For illness diagnostic classification device, i.e., the output of one support vector machines is the input of another support vector machines.
It preferably, further include the mode of classification based training in the step 6, specifically: by the illness diagnostic characteristic vector
It is input to the illness diagnostic classification device, classification based training is carried out to the illness diagnostic classification device, to determine nicety of grading, with
Verifying is reliable as identify whether to the infectious disease using the characteristic wave bands.
Preferably, the image preprocessing that the spectrum picture of the scene of public arena activity described in step 7 carries out specifically: right
The spectrum picture of the public arena activity scene carries out region segmentation, so as to the extraction of subsequent gray value.
In conclusion by adopting the above-described technical solution, the beneficial effects of the present invention are:
(1) spectral analysis technique is applied to real-time diagnosis and knowledge in public arena to communicable disease by the present invention for the first time
Not, real-time diagnosis, identification and control that existing clinical diagnosis can not be applied to communicable disease are avoided, public field is met
Close the requirement to the intellectually and automatically of infectious disease diagnosis；
(2) present invention obtains the light of public arena activity scene using multispectral camera in real-time diagnosis identification in real time
Spectrogram picture, the filter center wavelength that wherein multispectral camera is selected are the characteristic wave bands of healthy population and the spy of patient groups
The characteristic wave bands for levying wave band composition, extract the gray value of the spectrum picture, again pull up illness diagnostic characteristic vector, will suffer from
Sick diagnostic characteristic vector imports illness diagnostic classification device, to determine in the spectrum picture whether there is patient groups；This creation
Property design so that illness of the invention diagnosis is simple and efficient, be not required to excessively complicated image procossing, alleviate processing system
Computational load further improves the real-time of the diagnosis identification of public arena communicable disease.
(3) present invention for the first time identifies the diagnosis that support vector machines combination spectral analysis technique is applied to communicable disease,
Spectral information can be effectively obtained, transmits, handles, regenerates and utilize, accurately diagnosis identifies whether public arena has trouble
Patient group, is a kind of intelligent infectious disease real-time diagnosis recognition methods.
Detailed description of the invention
Fig. 1 is the flow diagram of the method for the present invention；
Fig. 2 is schematic diagram of the method for the present invention for albinism diagnosis identification；
Fig. 3 is the schematic diagram that the method for the present invention is used for lupus erythematosus diagnosis identification.
Specific embodiment
In order to which technical problems, technical solutions and advantages to be solved are more clearly understood, tie below
Embodiment is closed, the present invention will be described in detail.It should be noted that specific embodiment described herein is only to explain
The present invention is not intended to limit the present invention, and the product for being able to achieve said function belongs to equivalent replacement and improvement, is all contained in this hair
Within bright protection scope.
It elaborates below to the present invention.
Referring to attached drawing 1, a kind of deep learning method based on multispectral camera includes the following steps:
Step 1: utilizing multi-optical spectrum imaging system, healthy population and patient groups are carried out respectively in darkroom multispectral
Imaging, obtains the spectrum picture of different crowd, filter center wave-length coverage involved in the multi-optical spectrum imaging system covers
Visible wavelength is to infrared wavelength；
Step 2: region segmentation being carried out to the spectrum picture, gray value is extracted respectively to each region；
Step 3: the band index value P of different each imaging bands of healthy population is calculated using band index methodi, according to it is each at
As channel filter central wavelength and the corresponding band index value Pi, obtain band index curve graph；
Step 4: choosing principle, the bigger imaging band of band index value, contained light according to band index characteristic wave bands
Spectrum information amount is also bigger, and the multi-optical spectrum imaging system captures out band index value office from the band index curve graph automatically
The maximum data point in portion, and the filter center wavelength of corresponding imaging band is saved, to form healthy population spectrum
The characteristic wave bands of image；
Step 5: repeating step 3-4, obtain the characteristic wave bands of patient groups' spectrum picture and preservation；
Step 6: the spectral information of the healthy population and patient groups being combined into initial characteristics collection, wherein the ash
Angle value is as test sample collection, the trained sample of the characteristic wave bands composition of the characteristic wave bands of the healthy population and the patient groups
This collection constitutes illness diagnostic characteristic vector, and constructs illness diagnostic classification device；
Step 7: when carrying out live communicable disease real-time diagnosis identification, obtaining public arena in real time using multispectral camera
The spectrum picture of activity scene, wherein the filter center wavelength that the multispectral camera is selected includes the spy of the healthy population
Wave band and the characteristic wave bands of the patient groups are levied, by the spectrum picture real-time Transmission of the public arena activity scene to prison
Control center, the monitoring center carries out image preprocessing to the spectrum picture of the public arena activity scene, and extracts institute
State the gray value of spectrum picture；
Step 8: for the field diagnostic in step 7, illness diagnostic characteristic vector is constructed, including the public arena
The illness diagnostic characteristic vector is imported the illness and diagnosed by the gray value and characteristic wave bands of the spectrum picture of activity scene
Classifier, the illness diagnostic classification device predict the illness diagnostic characteristic vector in real time, to determine the public field
Whether there are patient groups in the spectrum picture of conjunction activity scene.
Optical filter wavelength described in the step 1 is specially 425,475,509,515,558,578,620,650,680,
717, totally 14 channels 750,800,832 and 850nm, corresponding half-band width is 100 respectively, 100,20,10,5,10,10,
10,10,10,10,10,10 and 5nm.
Wherein, the band index value P of different each imaging bands of healthy population is calculated in the step 3iIt specifically includes:
Step 3.1: calculating the standard deviation of the gray value of different healthy population spectrum pictures；
Step 3.2: calculating the related coefficient between each imaging band of the multi-optical spectrum imaging system；
Step 3.3: the band index value P of each imaging band of healthy population is calculated using formula (1)i:
In formula: parameter σiFor the standard deviation of the i-th channel image of spectrum picture, parameter RiFor the phase between wave band i and wave band j
Relationship number.
In the step 6, the illness diagnostic classification implement body is constructed are as follows: regard 2 support vector machines series connection as illness
Diagnostic classification device, i.e., the output of one support vector machines are the input of another support vector machines.In machine learning
In, support vector machines (SVM, also known as support vector network) is supervised learning model related to relevant learning algorithm, can
To analyze data, recognition mode, for classification and regression analysis.Give one group of training sample, it is each label be two classes, one
A SVM training algorithm establishes a model, distributes new example as a kind of or other classes, becomes non-probability binary linearity
Classification.The example of one SVM model, point such as in space, mapping, so that the example of the different classification is bright by one
Aobvious gap is the expression of division as wide as possible.New embodiment is then mapped in identical space, and predicts to fall based on them
Belong to a classification in the clearance side.It further include the mode of classification based training in the step 6, specifically: by the illness
Diagnostic characteristic vector is input to the illness diagnostic classification device, classification based training is carried out to the illness diagnostic classification device, to determine
Nicety of grading out, it is using verifying that the characteristic wave bands are reliable as identify whether to the infectious disease.
The image preprocessing that the spectrum picture of the scene of public arena activity described in step 7 carries out specifically: to the public affairs
The spectrum picture of occasion activity scene carries out region segmentation altogether, so as to the extraction of subsequent gray value.
The support vector machines that the present invention uses learns and classifies to the illness diagnostic characteristic vector, basic thought
To be developed from the optimal classification surface in the case of linear separability: assuming that there is two class samples, record sort line and it is all kinds of in from
The nearest sample of classification line and the straight line for being parallel to classification line, the distance between they are class interval, so-called optimal classification line
It is exactly to require classification line not only can correctly separate two classes, and keep class interval maximum.
Using the support vector machines of structural risk minimization, as a kind of general learning machine, support vector machines is
Statistical Learning Theory is used for the specific implementation of solving practical problems.It is to solve for convex quadratic programming problem in itself.From theory
Upper theory can obtain globally optimal solution, to effectively overcome the unavoidable local extremum problem of the methods of neural network.
The learning machine designed specifically for finite sample situation, it uses structural risk minimization, at the same to empiric risk and
The complexity of learning machine is controlled, and the generation of overfitting phenomenon is effectively prevented from, and can be obtained more excellent than conventional learning algorithms
Good generalization ability.Time-consuming higher-dimension inner product operation is preferably avoided by the introducing of kernel function.Support vector machines passes through non-
The learning sample of the low-dimensional input space is mapped to high-dimensional feature space by Linear Mapping, then dexterously by the introducing of kernel function
Time-consuming higher-dimension inner product operation is avoided, so that the complexity of algorithm is unrelated with the dimension of feature space.
Embodiment one
The present embodiment causes spectral information using the difference of the white or red Cheng Cheng of body skin for detecting albinism
Difference realize disease detection.
Based on the albinism recognition methods of multispectral camera activity scene in public, as shown in Fig. 2, detailed process is such as
Under:
Using multi-optical spectrum imaging system, the body skin of healthy population and albino is carried out respectively in darkroom
Multispectral imaging obtains the spectrum picture of different crowd；Region segmentation is carried out to the spectrum picture, each region is distinguished
Extract gray value；Using the band index value P for each imaging band of body skin that band index method calculates different healthy populationsi,
According to each imaging band filter center wavelength and the corresponding band index value Pi, obtain band index curve graph；
The multi-optical spectrum imaging system captures out band index value local maxima from the band index curve graph automatically
Data point, and the filter center wavelength of corresponding imaging band is saved, to form healthy population body skin light
The characteristic wave bands of spectrogram picture；In the same way, the characteristic wave bands of the spectrum picture of albino's body skin are obtained simultaneously
It saves.
The spectral information of the healthy population and albino's body skin is combined into initial characteristics collection, wherein described
Gray value is as test sample collection, the characteristic wave bands composition training of the characteristic wave bands of the healthy population and the patient groups
Sample set constitutes illness diagnostic characteristic vector, and constructs illness diagnostic classification device；The illness diagnostic characteristic vector is inputted
To the illness diagnostic classification device, classification based training is carried out to the illness diagnostic classification device, to determine nicety of grading, with verifying
Identify whether using the characteristic wave bands as whitening disease reliable.
When carrying out live albinism real-time diagnosis identification, public arena activity scene is obtained in real time using multispectral camera
Spectrum picture, wherein the multispectral camera select filter center wavelength include the healthy population characteristic wave bands and
The characteristic wave bands of albino, by the spectrum picture real-time Transmission to monitoring center, the monitoring center is to the spectrum
Image carries out image preprocessing, and extracts the gray value of the spectrum picture；The illness diagnostic characteristic vector is imported into institute
Illness diagnostic classification device is stated, the illness diagnostic classification device predicts the illness diagnostic characteristic vector in real time, to determine
Whether albinism patient groups are had in the spectrum picture.
Embodiment two
The present embodiment is used for the early detection to lupus erythematosus, using body skin and the spectral information of interior tissue and is good for
The difference of health people realizes disease detection.
Based on the lupus erythematosus recognition methods of multispectral camera activity scene in public, as shown in figure 3, detailed process
It is as follows:
Using multi-optical spectrum imaging system, in darkroom respectively to the body skin of healthy population and patients with SLE into
Row multispectral imaging obtains the spectrum picture of different crowd；Region segmentation is carried out to the spectrum picture, to each region point
Indescribably take gray value；Using the band index value for each imaging band of body skin that band index method calculates different healthy populations
Pi, according to each imaging band filter center wavelength and the corresponding band index value Pi, obtain band index curve graph；
The multi-optical spectrum imaging system captures out band index value local maxima from the band index curve graph automatically
Data point, and the filter center wavelength of corresponding imaging band is saved, to form healthy population body skin light
The characteristic wave bands of spectrogram picture；In the same way, the characteristic wave bands of the spectrum picture of patients with SLE body skin are obtained
And it saves.
The spectral information of the healthy population and patients with SLE body skin is combined into initial characteristics collection, wherein institute
Gray value is stated as test sample collection, the characteristic wave bands of the characteristic wave bands of the healthy population and the patient groups form instruction
Practice sample set, constitutes illness diagnostic characteristic vector, and construct illness diagnostic classification device；The illness diagnostic characteristic vector is defeated
Enter to the illness diagnostic classification device, classification based training is carried out to the illness diagnostic classification device, to determine nicety of grading, to test
Card is reliable as identify whether to lupus erythematosus using the characteristic wave bands.
When carrying out live lupus erythematosus real-time diagnosis identification, public arena activity scene is obtained in real time using multispectral camera
Spectrum picture, wherein the filter center wavelength that the multispectral camera is selected include the characteristic wave bands of the healthy population with
And the characteristic wave bands of patients with SLE, by the spectrum picture real-time Transmission to monitoring center, the monitoring center is to described
Spectrum picture carries out image preprocessing, and extracts the gray value of the spectrum picture；The illness diagnostic characteristic vector is led
Enter the illness diagnostic classification device, the illness diagnostic classification device predicts the illness diagnostic characteristic vector in real time, comes
Whether determine has lupus erythematosus patient groups in the spectrum picture.
The foregoing is merely the preferred embodiments of the invention, the claims that are not intended to limit the invention.
Simultaneously it is described above, for those skilled in the technology concerned it would be appreciated that and implement, therefore other be based on institute of the present invention
The equivalent change that disclosure is completed, should be included in the covering scope of the claims.
Claims (6)
1. a kind of deep learning method based on multispectral camera, which comprises the steps of:
Step 1: utilize multi-optical spectrum imaging system, in darkroom respectively to healthy population and patient groups carry out it is multispectral at
Picture obtains the spectrum picture of different crowd, and filter center wave-length coverage involved in the multi-optical spectrum imaging system covers can
Light-exposed wavelength is to infrared wavelength；
Step 2: region segmentation being carried out to the spectrum picture, gray value is extracted respectively to each region；
Step 3: the band index value P of different each imaging bands of healthy population is calculated using band index methodi, logical according to each imaging
Channel filter central wavelength and the corresponding band index value Pi, obtain band index curve graph；
Step 4: the multi-optical spectrum imaging system captures out band index value part most from the band index curve graph automatically
Big data point, and the filter center wavelength of corresponding imaging band is saved, to form healthy population spectrum picture
Characteristic wave bands；
Step 5: repeating step 3-4, obtain the characteristic wave bands of patient groups' spectrum picture and preservation；
Step 6: the spectral information of the healthy population and patient groups being combined into initial characteristics collection, wherein the gray value
As test sample collection, the characteristic wave bands of the characteristic wave bands of the healthy population and the patient groups form training sample
Collection constitutes illness diagnostic characteristic vector, and constructs illness diagnostic classification device；
Step 7: when live communicable disease real-time diagnosis identifies, multispectral camera obtains the light of public arena activity scene in real time
Spectrogram picture, by the spectrum picture real-time Transmission of the public arena activity scene to monitoring center, the monitoring center is to described
The spectrum picture of public arena activity scene carries out image preprocessing, and extracts the gray value of the spectrum picture；
Step 8: for the field diagnostic of step 7, constructing illness diagnostic characteristic vector, the illness diagnostic characteristic vector is imported
The illness diagnostic classification device is predicted whether there is trouble in the spectrum picture to determine the public arena activity scene in real time
Patient group.
2. the method according to claim 1, wherein optical filter wavelength described in the step 1 be specially 425,
475, totally 14 channels 509,515,558,578,620,650,680,717,750,800,832 and 850nm, corresponding half band
Wide is 100,100,20,10,5,10,10,10,10,10,10,10,10 and 5nm respectively.
3. according to the method described in claim 2, wherein, the wave of different each imaging bands of healthy population is calculated in the step 3
Section index value PiIt specifically includes:
Step 3.1: calculating the standard deviation of the gray value of different healthy population spectrum pictures；
Step 3.2: calculating the related coefficient between each imaging band of the multi-optical spectrum imaging system；
Step 3.3: the band index value P of each imaging band of healthy population is calculated using formula (1)i:
In formula: parameter σiFor the standard deviation of the i-th channel image of spectrum picture, parameter RiFor the phase relation between wave band i and wave band j
Number.
4. the method according to claim 1, wherein constructing the illness diagnostic classification utensil in the step 6
Body are as follows: regard 2 support vector machines series connection as illness diagnostic classification device, i.e., the output of one support vector machines is another
The input of the support vector machines.
5. according to the method described in claim 4, it is characterized in that, further include the mode of classification based training in the step 6, tool
Body are as follows: the illness diagnostic characteristic vector is input to the illness diagnostic classification device, the illness diagnostic classification device is carried out
Classification based training, to determine nicety of grading, using verifying by the characteristic wave bands as identify whether to the infectious disease can
It leans on.
6. according to the method described in claim 3, it is characterized in that, to the public arena activity scene in the step 7
The image preprocessing that spectrum picture carries out specifically: region segmentation is carried out to the spectrum picture of the public arena activity scene,
So as to the extraction of subsequent gray value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910248658.7A CN110021406A (en) | 2019-03-29 | 2019-03-29 | A kind of deep learning method based on multispectral camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910248658.7A CN110021406A (en) | 2019-03-29 | 2019-03-29 | A kind of deep learning method based on multispectral camera |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110021406A true CN110021406A (en) | 2019-07-16 |
Family
ID=67190217
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910248658.7A Pending CN110021406A (en) | 2019-03-29 | 2019-03-29 | A kind of deep learning method based on multispectral camera |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110021406A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202421190U (en) * | 2012-01-06 | 2012-09-05 | 孔令华 | Multispectral microscopic chip and device for diagnosing cancers and infectious diseases |
| CN104083154A (en) * | 2014-07-24 | 2014-10-08 | 成都市晶林科技有限公司 | Population health detection system and method |
| CN107064089A (en) * | 2017-04-13 | 2017-08-18 | 浙江大学 | A kind of Hot Pepper Seedling epidemic disease early monitoring apparatus and method based on Internet of Things |
| CN108272437A (en) * | 2017-12-27 | 2018-07-13 | 中国科学院西安光学精密机械研究所 | Spectral detection system and sorter model construction method for skin disease diagnosis |
-
2019
- 2019-03-29 CN CN201910248658.7A patent/CN110021406A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202421190U (en) * | 2012-01-06 | 2012-09-05 | 孔令华 | Multispectral microscopic chip and device for diagnosing cancers and infectious diseases |
| CN104083154A (en) * | 2014-07-24 | 2014-10-08 | 成都市晶林科技有限公司 | Population health detection system and method |
| CN107064089A (en) * | 2017-04-13 | 2017-08-18 | 浙江大学 | A kind of Hot Pepper Seedling epidemic disease early monitoring apparatus and method based on Internet of Things |
| CN108272437A (en) * | 2017-12-27 | 2018-07-13 | 中国科学院西安光学精密机械研究所 | Spectral detection system and sorter model construction method for skin disease diagnosis |
Non-Patent Citations (3)
| Title |
|---|
| 刘鑫等: "马铃薯叶片晚疫病的多光谱分类识别", 《光学仪器》 * |
| 杜剑等: "基于卷积神经网络与显微高光谱的胃癌组织分类方法研究", 《光学学报》 * |
| 虞佳佳等: "基于高光谱成像技术的番茄叶片灰霉病早期检测研究", 《光谱学与光谱分析》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Patil et al. | White blood cells image classification using deep learning with canonical correlation analysis | |
| Waghmare et al. | Detection and classification of diseases of grape plant using opposite colour local binary pattern feature and machine learning for automated decision support system | |
| Vijayakumar | Classification of brain cancer type using machine learning | |
| Song et al. | Simultaneous cell detection and classification in bone marrow histology images | |
| Akram et al. | Identification and classification of microaneurysms for early detection of diabetic retinopathy | |
| Devi et al. | Malaria infected erythrocyte classification based on a hybrid classifier using microscopic images of thin blood smear | |
| Anandhakrishnan et al. | Deep Convolutional Neural Networks for image based tomato leaf disease detection | |
| Baghel et al. | WBCs-Net: Type identification of white blood cells using convolutional neural network | |
| Roshini et al. | Automatic diagnosis of diabetic retinopathy with the aid of adaptive average filtering with optimized deep convolutional neural network | |
| Devi et al. | Hybrid classifier based life cycle stages analysis for malaria-infected erythrocyte using thin blood smear images | |
| Memeu | A rapid malaria diagnostic method based on automatic detection and classification of plasmodium parasites in stained thin blood smear images | |
| Kulkarni et al. | Rice leaf diseases detection using machine learning | |
| Lanjewar et al. | CNN and transfer learning methods with augmentation for citrus leaf diseases detection using PaaS cloud on mobile | |
| Rameen et al. | Leveraging supervised machine learning techniques for identification of malaria cells using blood smears | |
| Sunitha et al. | Modeling convolutional neural network for detection of plant leaf spot diseases | |
| Kukana | Hybrid Machine Learning Algorithm-Based Paddy Leave Disease Detection System | |
| Li et al. | A machine learning approach for detection plant disease: Taking orchid as example | |
| CN110021406A (en) | A kind of deep learning method based on multispectral camera | |
| CN109993110A (en) | A method of factory's monitoring is carried out based on spectral information | |
| Hamid et al. | An intelligent strabismus detection method based on convolution neural network | |
| Warke et al. | Novel approach of classification and detection of rice plant diseases | |
| Sajitha et al. | Banana Fruit Disease Detection and Categorization Utilizing Graph Convolution Neural Network (GCNN) | |
| Singh et al. | A Survey on Different Methods for Medicinal Plants Identification and Classification System | |
| Chaithanya et al. | Wheat Leaf Disease classification using modified ResNet50 Convolutional Neural Network model | |
| Prashasthi et al. | Image processing approach to diagnose eye diseases |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190716 |
|
| RJ01 | Rejection of invention patent application after publication |