CN106056116A - Fuzzy rough set-based sleeping posture pressure image recognition method - Google Patents
Fuzzy rough set-based sleeping posture pressure image recognition method Download PDFInfo
- Publication number
- CN106056116A CN106056116A CN201610378199.0A CN201610378199A CN106056116A CN 106056116 A CN106056116 A CN 106056116A CN 201610378199 A CN201610378199 A CN 201610378199A CN 106056116 A CN106056116 A CN 106056116A
- Authority
- CN
- China
- Prior art keywords
- image
- pressure
- tonogram
- picture
- region
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
- A61B5/4809—Sleep detection, i.e. determining whether a subject is asleep or not
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
- A61B5/4815—Sleep quality
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
Landscapes
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Evolutionary Computation (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Anesthesiology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a fuzzy rough set-based sleeping posture pressure image recognition method. The method is characterized by comprising the following steps: 1, data acquisition is carried out, and real-time pressure data detected by a pressure sensitivity sensor array located below a sleeping position are acquired; 2, image conversion is carried out and the real-time pressure data acquired in the first step are converted into a pressure image; 3, image pretreatment is carried out on the pressure image obtained in the second step; 4, image feature extraction is carried out, that is, feature extraction is carried out on the pressure image through image pretreatment, and the extracted feature values form a feature set for single pressure images; and 5, a fuzzy rough set method is adopted to process the image features extracted in the fourth step for realizing sleeping posture recognition.
Description
Technical field
The present invention relates to a kind of monitoring system, the method and system of a kind of automatic identification sleep attitude, particularly
A kind of sleeping position pressure image-recognizing method based on fuzzy coarse central and system.
Background technology
The time in all one's life of people has spent 1/3rd in sleep, and sleep quality is generally heavier than the time length of sleep
, it is related to psychology and the physiological health of people, and the people of poor sleeping quality easily produces anxiety, tired, absent minded, or
Drinking and eating irregularly.Be may certify that the symptoms such as diagnosis of cardiovascular diseases, diabetes and obesity by Sleep stages.Sleep stages and sleeping
Dormancy obstacle all can produce mental sickness, such as depression, excessive drinking and bipolar affective disorder.Determine the index of sleep quality, such as sleep
In the stage, dyskoimesis, sleeping posture, to being non-the normally off key for medical diagnosis.It is most commonly that grinding of sleep apnea
Studying carefully, in recent years, many is studied and is devoted to sleep apnea sleeping posture and carries out data analysis.Show sleep-respiratory according to the study
Sleeping posture good in time-out is conducive to alleviating the respiratory disorder degree of apnea patient.
In terms of medical treatment, Ambrogio et al. finds sleeping posture and the relation of chronic respiratory failure, and this directly results in sleeps
Dormancy asphyxia.The most prostrate meeting of bad sleeping position makes the body weight of major part fall at rib and intestines and internal organs of the body, thus press against diaphragm and pulmonary,
Impact is breathed, especially even more serious to the patient effect having apnea syndrome.Cervical region is then owing to must turn round to side
Transfer holding and breathe unimpeded, increase torsional buckling, easily cause wound.In sum, have apnea syndrome patient should avoid with
Lie on the back and front lying position is slept, preferably take lateral position to sleep, fall after can alleviating or prevent pharyngeal cavity portion soft tissue and the root of the tongue；
Alleviate cervical region and breast fat tissue to airway pressure, thus help and alleviate the sound of snoring, even prevent asphyxia.
The impact of bedding, desirably prevents the formation of pressure ulcer to the analysis of sleeping position, and notice is easily generated the patient of bedding
With bedding pressure spot on the horizon.Therefore, sleep mode automatically attitude monitoring is necessary.
Up to the present, research worker proposes the next automatic monitoring sleep posture of different methods.Tradition research sleep appearance
Gesture pattern is to use video camera and mike.Nakajima Yoshitaka et al. proposes view-based access control model signal analysis sleep-respiratory and postural change
System, but imaging is brought the biggest noise than relatively low by night, and video council brings serious privacy concern, very
It is unfavorable for the health treatment of patient in hospital.In such working set, posture changes rather than the health appearance of understanding before testing
Gesture.
It is therefore proposed that use the real-time monitoring sleep status of array of pressure sensors, it is provided that the algorithm of a kind of sleeping position identification and
System.Increase substantially the accuracy of sleeping posture identification, and more protect privacy.Slow down for the asphyxia in medical treatment, bed
Cotton-padded mattress prevention provides data support with the area research personnel such as sleep quality raising in treatment and intelligent home.
Summary of the invention
The present invention is directed to the key issue that prior art exists, it is provided that a kind of sleeping position tonogram picture based on fuzzy coarse central
Recognition methods and system, it is high that this system has accuracy of detection, and real-time is good, and recognition speed is fast, is applicable to different sexes, body
The features such as the sleeping position detection of high, body weight patient.According to current medical market to the sleep demand of attitude and intelligent home direction
Development, slows down and suffers from the patient of bedding impact to the respiratory disorder suffering from apnea syndrome patient, and to sleeping
The demand of dormancy quality is from infant to person in middle and old age, and the sleep quality of different groups is badly in need of solving, and prospect is considerable.
A kind of sleeping position pressure image-recognizing method based on fuzzy coarse central, it is characterised in that described method includes following step
Suddenly
First step data acquisition
Gather and be positioned at the real-time pressure data that the force-sensing sensor array detection below sleep positions obtains；
Second step image is changed
The real-time pressure data first step collected is converted into tonogram picture, particularly as follows: set up image coordinate and sensing
The image that device array distribution is consistent, is converted into pixel in the image coordinate of correspondence by the pressure value collected on each sensor
Gray value, thus obtain reflect sensor array upward pressure distribution tonogram picture；
3rd step Image semantic classification
The tonogram picture obtaining second step carries out Image semantic classification；
4th step image characteristics extraction
Tonogram picture through Image semantic classification is carried out feature extraction；
5th step use the 4th step based on Fuzzy and Rough diversity method pair obtain characteristics of image set carry out process realization sleep
Appearance identification
To through manual sort sleeping position tonogram picture carry out image characteristics extraction according to method described in the 4th step, and will
The characteristics of image set extracted and the classification composition training sample set of tonogram picture, then the classification with sleeping position tonogram picture
(hereinafter referred to as image category) sets up decision table as decision attribute；Characteristics of image as conditional attribute in decision table is carried out
Attribute reduction, is the minimum characteristics of image that can express image category, thus has obtained image through the conditional attribute of yojan
The decision rules of classification；Through the feature extraction of the tonogram picture to sleeping position identification, carry out degree of membership calculating with decision rules, with
The image category of degree of membership maximum is as the sleeping position classification identified.
Described a kind of based on fuzzy coarse central sleeping position pressure image-recognizing method, it is characterised in that described 3rd step bag
Include
1) image rectification, including 1.1) geometric transformation correction, 1.2) Threshold segmentation；2) region divides.
Described region is divided into the tonogram through image rectification along Y direction as 4 deciles, and divides successively along Y-axis
Go out 25% region, 50% region and 75% region.
Described a kind of based on fuzzy coarse central sleeping position pressure image-recognizing method, it is characterised in that described 4th step needs
Feature to be extracted includes: the pressure span that pressure span accounts in the ratio of whole tonogram picture, 25% region accounts for whole pressure
The pressure span that pressure span in the ratio in region, 50% region accounts in the ratio of whole pressure span and 75% region accounts for whole
The ratio of individual pressure span, the symmetry of tonogram picture, pressure span number, tonogram image space tropism, the balance of tonogram picture,
Shoulder regions area, shoulder coordinate, seat area area, buttocks coordinate, shoulder buttocks centroidal distance, features described above value constitutes single width
The characteristic set of tonogram picture.
Compared with prior art, its major advantage is the present invention:
(1) instant invention overcomes existing to the camera brightness problem at night in sleep monitor with the protection problem to privacy,
And then make monitored person there is no mental maladjustment during sleeping position is monitored, more naturally, chainlessly sleep, greatly eliminate
The psychological burden of detected person.
(2) image multiple features ensure that the integrity of image information to greatest extent so that more accurate in categorizing process
Really.
(3) processing of image uses method based on sequential pressure data, compares more traditional image acquisition more quick,
Especially in processing procedure, the size of single image is 64X128 pixel, substantially reduces image processing speed and deposits with reducing
Storage takes up room.
(4) blocking body pressure due to imperfection and the oneself of body shape, video image is compared in pressure graphical analysis
Analyze more challenge, and fuzzy coarse central is to carry out data based on to the uncertainty of unknown message or conflicting data
Processing and speculate, essence is the mode of a kind of machine self-teaching, provides effective inspection for processing to obscure with uncertain knowledge
Survey instrument.
Present invention sleeping position based on fuzzy coarse central recognition methods, compared to prior art, comprises more sleeping position characteristics of image
Under more quickly, Real time identification, slow down for the asphyxia in medical treatment, bedding prevention and sleep quality in treatment and intelligent home
The area research personnel such as raising provide accurate data support.
Detailed description of the invention
In order to make the purpose of the present invention more highlight, below to the detailed description of the invention.
Embodiment
A kind of sleeping position pressure image-recognizing method based on fuzzy coarse central, comprises the following steps
First step data acquisition
Gather and be positioned at the real-time pressure data that the force-sensing sensor array detection below sleep positions obtains
Described flexible force sensitive sensor array (hereinafter referred to as sensor array) is the rectangular array of 64 × 128, data
Frequency acquisition is 10Hz, and the span of each flexible force sensitive sensor is 0-512.
Flexible force sensitive sensor array distribution can cover the maximal projection area of sleeper just, ensures to greatest extent
The integrity of whole body pressure data.The real-time pressure data gathered contains the size of real-time body pressure.
Second step image is changed
The real-time pressure data first step collected is converted into tonogram picture, particularly as follows: set up image coordinate and sensing
The image that device array distribution is consistent, is converted into pixel in the image coordinate of correspondence by the pressure value collected on each sensor
Gray value, thus obtain reflect sensor array upward pressure distribution tonogram picture.
In the present embodiment, each pixel of described tonogram picture and the sensor one_to_one corresponding of sensor array, described pressure
Trying hard to as being 8 gray level images, the value after being halved by the pressure data that each sensor records is as the gray value of respective pixel.
The Y-axis of the image coordinate of described tonogram picture and X-axis correspond respectively to y direction and the X direction of sensor array.
The spinal column direction of detected person is almost parallel with the y direction of sensor array.
3rd step Image semantic classification
The tonogram picture obtaining second step carries out Image semantic classification, including
1) image rectification, described image rectification includes
1.1) geometric transformation correction,
Described geometric transformation correction specifically includes and translates image, rotates, scale to correct reset pressure image
Geometric distortion；Owing to human body lies on pliable pressure sensor array mattress, especially monitored person at night be likely to occur body dynamic,
The action such as stand up, cause the pressure span reflected distribution to occur in that the shift in position relative to tonogram picture.Pass through geometry
Conversion correction can make the geometric distortion of the tonogram picture collected well be corrected；
1.2) Threshold segmentation,
First the pixel of tonogram picture is carried out threshold value setting, crosses other pixels filtered outside pressure span according to setting value,
And then make pressure span become apparent from；Then the gray value of reverse image pixel, obtains the tonogram picture through image rectification.Institute
Stating for white beyond the pressure span of image rectification, in pressure span, the region that pressure is the biggest, color of image more levels off to
Black, is white outside pressure span.(in existing gray value method for expressing, intensity value ranges is typically from 0 to 255, and white is
255, black is 0)
2) region divides, and as 4 deciles and is marked off successively along Y-axis by the tonogram through image rectification along Y direction
25% region, 50% region and 75% region, complete Image semantic classification.
4th step image characteristics extraction
Tonogram picture through Image semantic classification is carried out feature extraction, needs the feature extracted to include
Pressure span accounts for the pressure span in the ratio of whole tonogram picture, 25% region and accounts for the ratio of whole pressure span
The pressure span that pressure span in example, 50% region accounts in the ratio of whole pressure span and 75% region accounts for whole pressure area
The ratio in territory, the symmetry of tonogram picture, pressure span number, tonogram image space tropism, the balance of tonogram picture, shoulder regions
Area, shoulder coordinate, seat area area, buttocks coordinate, shoulder buttocks centroidal distance, features described above value constitutes single width tonogram picture
Characteristic set.
Described pressure span accounts for the ratio of whole image, and this refers to total pixel of pressure span that detected person's health formed
Number accounts for the percentage ratio of whole pressure image pixel, it is possible to reflect build and the sleeping position situation of change of tester, same test
Person's described ratio under different sleeping position is not quite similar；
The change of the ratio of whole pressure span is accounted for according to the pressure span in 25% region, 50% region and 75% region
Changing, the dynamic change of body and the sleeping position that can extract this tester change, and especially tester is that sleeping position is close to pressure transducer battle array
The top half of row or the latter half, be very helpful to the sleep habit correcting sleep monitor person；
The symmetry of tonogram picture, refers to extract, by the pressure span of described Image semantic classification, the tonogram picture obtained,
And compare under side-lying position, lie on the back and prostrate state more can show symmetry, symmetric value scope is 0-1, sample I and
Symmetry value under prostrate state is closer to 1, and the sleeping position symmetry value under other states is both less than 0.5；
Pressure span number, refers to the number of regions that tonogram picture is formed, and compares and more prostrate more can substantially show during supine position
Illustrating shoulder and seat area, this is the structure reason relatively highlighted due to shoulder and the buttocks of human body, makes full use of people
The structure of body, more can extract effective feature and reflect the sleeping position of current tester；
Tonogram image space tropism, refers to detect the curvature direction of the trunk portion of body pressure image, is lying on the back and bowing
Under sleeping state, human body shows more straight tonogram picture, and the step extracting this part includes: change into corresponding from tonogram picture dress
Skeleton drawing, removes excess pixel in skeleton drawing so that skeleton drawing more directly reflects the shape under state of lying on one's side, along skeleton
Graph discovery angular bisector, curvature direction is the angular bisector summation in y-axis durection component, the y-axis direction i.e. long side direction of sheet；
The balance of tonogram picture, refers to that it is different from the symmetry of image, and more calculating tonogram as which side comprises more
Pressure, this feature is more suitable for making a distinction between lying on one's side；
Shoulder regions area, refers to the region that shoulder pressure span is formed, and again calculate is the picture in pressure span
Element number；Shoulder coordinate, refers to, with the barycentric coodinates of shoulder pressure span as reference value, demarcate the position coordinates of shoulder；
Seat area area, refers to the region that buttocks pressure span is formed, again calculate the picture in buttocks pressure span
Element number；Buttocks coordinate, refers to, with the barycentric coodinates of buttocks pressure span as reference value, demarcate the position coordinates of buttocks；
Shoulder buttocks centroidal distance, refers to the distance of described shoulder barycentric coodinates and buttocks barycentric coodinates.
5th step the 4th step based on Fuzzy and Rough diversity method pair obtain characteristics of image set carry out process realize sleeping position know
Not,
To through manual sort sleeping position tonogram picture carry out image characteristics extraction according to method described in the 4th step, and will
The characteristics of image set extracted and the classification composition training sample set of tonogram picture, then the classification with sleeping position tonogram picture
(hereinafter referred to as image category) sets up decision table as decision attribute；Characteristics of image as conditional attribute in decision table is carried out
Attribute reduction, is the minimum characteristics of image that can express image category, thus has obtained image through the conditional attribute of yojan
The decision rules of classification；Through the feature extraction of the tonogram picture to sleeping position identification, carry out degree of membership calculating with decision rules, with
The image category of degree of membership maximum is as the sleeping position classification identified.
Specifically include
1) composition training sample set,
According to the type of lying on the back, prostrate type, left side fetal type, left side trunk type, right side fetal type, the big class pair of right side trunk type 6
Sleeping position tonogram picture as training sample carries out manual sort, described in lie on the back type, prostrate type, left side fetal type, left side trunk
Type, right side fetal type, right side trunk type, be briefly referred to as S, P, LF, LL, RF, RL；To the sleeping position tonogram through manual sort
Picture, carries out image characteristics extraction according to method described in the 4th step, and by the characteristics of image set extracted and the class of tonogram picture
Zu Cheng training sample set
2) decision table is set up
It is that an information system in Fuzzy and Rough diversity method sets up decision table, described decision table by training sample set cooperation
The information system of knowledge representation is i.e. carried out by training sample,
Using tonogram picture as the object set of decision table, it is decision-making by the set of image characteristics cooperation extracted through the 4th step
Conditional attribute in table, using the classification of image as decision attribute.
One information system S can be expressed as four-tuple S={U, A, V, f}, wherein
U={x1,x2,···,xnIt is the nonempty finite set of object, the i.e. collection of tonogram picture in n dimension theorem in Euclid space
Close；
A is the set of all fuzzy equivalence relations on U, A=C ∪ D,C is conditional attribute subset, i.e. image
Characteristic set, andD is the class categories set of decision attribute subset, i.e. image, andUsing information system T as
One decision table, is denoted as T={U, A, C, D}.
V=UP∈AVP,VPIt it is the territory of attribute P.
F:U × A → V is image classification functions to each image xi∈ U, each feature q ∈ A, have f (x, q) ∈ Vq。
Classification samples training set can be expressed as two-dimentional decision table, wherein a U={x1,x2,···,x6, C=
{c1,c2,···,c12, D={d1,d2,···,d6}。
In information system S={U, A, V, f}, ifIt is the subset in individual universe,Then X's is lower and upper
Approximate set and borderline region are respectively as follows: PX isThe set of upper those characteristic elements being necessarily classified, is i.e. included in X at interior maximum definable collection；PX is
The set of upper those characteristic elements that may be classified of U, i.e. comprises the minimum definable collection of X；Bndp(X) it is can not beOn be classified, the set of those characteristic elements can not being classified on U-X again.The border of image feature value scope
Part determines its ownership, needs in carrying out great amount of samples training, finds and reasonably select boundary.
3) attribute reduction
It follows that in attribute reduction part, by the foundation of information system S of above-mentioned training sample, define and comprise figure
As object, characteristics of image, image category and the set of classificating knowledge function, information system S is by SC={ U, C, V, f} and SD=U,
Two information system compositions of D, V, f}, they are by U, and V, f combine closely.Owing to the codomain of characteristics of image is the spy of continuum
Point, can produce a fuzzy decision table the simplest by by decision table carries out equivalence.In decision table, different condition
Attribute is different to the importance of decision attribute, in order to investigate the situation of change of classification, needs to remove a certain attribute in decision table
Or community set, thus Rule of judgment characteristic attribute and the correlation degree of decision-making characteristic attribute.In information system S={U, C ∪ D,
In V, f}, if D={d1,d2,···,di, conditional attribute subset" positive region " about decision attribute D is defined as
POSB(D)=∪PX:X∈D}；
Wherein about the positive region of D, B represents that those just can be divided into all objects of correct classification according to attribute set B.Bar
Part attribute setIt is defined as with the degree of correlation of decision attribute D:
The number of element during wherein cad (X) represents X set；Obviously, 0≤k (B, D)≤1, k (B, D) are that design conditions belong to
Degree of correlation between property B and decision attribute D provides means.At unessential attribute, i.e. close to 0 attribute, this is permissible
Cast out, reach the purpose of attribute reduction, reduce the operation time of attributes match simultaneously.
4) decision rules is produced
Each decision table, has a corresponding decision rules, and these relevant decision ruless are referred to as decision-making
Algorithm.T={U, A, C, D} are decision tables, make X ∈ U/C, Y ∈ U/D.
Note: des (X)=f (x, c) x ∈ X ∨ c ∈ C},
Des (Y)={ f (x, d) x ∈ X ∨ d ∈ C}
IfThen define and can be shown that decision rules is r by X, Yxy: des (X) → des (Y), define this rule
Definitiveness isWhen μ (X, Y)=1 isTime, this rule determines that, otherwise claims this rule
It is uncertain.The evaluation attributes subset quality to classification, by attribute setThe classification quality of the classification Ψ determined is:
Classification quality represents the ratio of all objects in the number of objects and information system correctly classified by attribute set P
Value, this is one of the key index of importance of evaluation attributes subset.
3) pressure classification of images
According to the result of attribute reduction, the sleeping position tonogram picture that will carry out sleeping position identification is carried out feature extraction, extract
Conditional attribute after yojan.
Calculate sleeping position tonogram picture and the degree of membership of image category in decision rules, specific as follows: the condition setting image P belongs to
C property value in property be the span of the c attribute of n, D classification image be [g, h]/m, with f (c) represent P c attribute belong to D
The c attribute degree of membership of classification:
Carrying out in image Auto-matching categorizing process, the tonogram of extraction input, as yojan attribute, can obtain this pressure
The image degree of membership to each classification, determines the classification of this image according to degree of membership value.
Describe the ultimate principle sticking with paste Rough Sets Classification method in the present invention above in association with specific embodiment, the present invention sleeps
Appearance identification technical scheme and traditional bottom semantic image sorting technique scheme based on content show on image classification performance
Go out the method and there is good accuracy and effectiveness, can preferably realize classification of images.
The present invention does not addresses part and is applicable to known technology.
Enforcement to the present invention has been described in detail above, but the preferable implementation process that described content is the present invention, no
Can be considered for limiting the application scope of the claims.All equalizations done with the present patent application right
Change and improvement, within all should belonging to the application scope of the claims.
Claims (4)
1. a sleeping position pressure image-recognizing method based on fuzzy coarse central, it is characterised in that described method includes following step
Rapid:
First step data acquisition
Gather and be positioned at the real-time pressure data that the force-sensing sensor array detection below sleep positions obtains；
Second step image is changed
The real-time pressure data first step collected is converted into tonogram picture, particularly as follows: set up image coordinate and sensor array
The image that column distribution is consistent, is converted into the ash of pixel in the image coordinate of correspondence by the pressure value collected on each sensor
Angle value, thus obtain reflecting the tonogram picture of sensor array upward pressure distribution；
3rd step Image semantic classification
The tonogram picture obtaining second step carries out Image semantic classification；
4th step image characteristics extraction
Tonogram picture through Image semantic classification is carried out feature extraction；The eigenvalue extracted constitute single width tonogram as
Characteristic set
5th step uses to extract the 4th step based on Fuzzy and Rough diversity method and obtains characteristics of image and carry out process and realize sleeping position identification
To through manual sort sleeping position tonogram picture carry out image characteristics extraction according to method described in the 4th step, and will extract
The classification composition training sample set of the characteristics of image set arrived and tonogram picture, then using the classification of sleeping position tonogram picture as certainly
Plan attribute sets up decision table；Characteristics of image as conditional attribute in decision table is carried out attribute reduction, belongs to through the condition of yojan
Property be the minimum characteristics of image that can express image category, thus obtained the decision rules of image classification；Through to sleeping
The feature extraction of the tonogram picture of appearance identification, carries out degree of membership calculating with decision rules, makees with the image category that degree of membership is maximum
For the sleeping position classification identified.
A kind of sleeping position pressure image-recognizing method based on fuzzy coarse central, it is characterised in that institute
State the 3rd step and include 1) image rectification, including 1.1) geometric transformation correction, and 1.2) Threshold segmentation；2) region divides.
A kind of sleeping position pressure image-recognizing method based on fuzzy coarse central, it is characterised in that institute
State region to be divided into the tonogram through image rectification along Y direction as 4 deciles, and mark off successively along Y-axis 25% region,
50% region and 75% region.
A kind of sleeping position pressure image-recognizing method based on fuzzy coarse central, it is characterised in that institute
Stating the 4th step needs the feature extracted to include: pressure span accounts for the pressure span in the ratio of whole tonogram picture, 25% region
Account for the pressure span in the ratio of whole pressure span, 50% region and account for the pressure in the ratio of whole pressure span and 75% region
Power region accounts for the ratio of whole pressure span, the symmetry of tonogram picture, pressure span number, tonogram image space tropism, tonogram
The balance of picture, shoulder regions area, shoulder coordinate, seat area area, buttocks coordinate, shoulder buttocks centroidal distance, features described above
Value constitutes the characteristic set of single width tonogram picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610378199.0A CN106056116A (en) | 2016-05-31 | 2016-05-31 | Fuzzy rough set-based sleeping posture pressure image recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610378199.0A CN106056116A (en) | 2016-05-31 | 2016-05-31 | Fuzzy rough set-based sleeping posture pressure image recognition method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106056116A true CN106056116A (en) | 2016-10-26 |
Family
ID=57173089
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610378199.0A Pending CN106056116A (en) | 2016-05-31 | 2016-05-31 | Fuzzy rough set-based sleeping posture pressure image recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106056116A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106963166A (en) * | 2017-03-30 | 2017-07-21 | 南京信息工程大学 | A kind of adaptively changing and the intelligent bed and its method of work of carrier contact portion stress |
| CN109925136A (en) * | 2019-03-20 | 2019-06-25 | 河北工业大学 | A kind of intelligent bed system and its application method based on movement intention assessment |
| CN110432724A (en) * | 2019-08-28 | 2019-11-12 | 南通金露智能设备有限公司 | A kind of change with sleeping position and the pillow of automatic controlled height |
| CN110477669A (en) * | 2019-08-16 | 2019-11-22 | 福建农林大学 | A kind of infrared massaging mattress based on position pressure measurement |
| CN110546473A (en) * | 2017-04-10 | 2019-12-06 | 霓达株式会社 | Judgment system and judgment program |
| CN110811630A (en) * | 2019-10-31 | 2020-02-21 | 瞬联软件科技（北京）有限公司 | Pregnant woman sleeping posture detection method and device |
| CN110989459A (en) * | 2019-12-23 | 2020-04-10 | 西南交通大学 | Intelligent bed body control system and use method thereof |
| CN111000528A (en) * | 2019-12-23 | 2020-04-14 | 西南交通大学 | Method for detecting sleeping posture of human body |
| CN111310599A (en) * | 2020-01-20 | 2020-06-19 | 重庆大学 | Sleep action recognition system capable of quickly adapting to various change factors |
| CN112001286A (en) * | 2020-08-14 | 2020-11-27 | 燕山大学 | Neck pillow height adjusting method and device for sleep posture recognition processing based on pressure image |
| CN113273998A (en) * | 2021-07-08 | 2021-08-20 | 南京大学 | Human body sleep information acquisition method and device based on RFID label matrix |
| CN113962995A (en) * | 2021-12-21 | 2022-01-21 | 北京鹰瞳科技发展股份有限公司 | Cataract model training method and cataract identification method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120323501A1 (en) * | 2011-05-20 | 2012-12-20 | The Regents Of The University Of California | Fabric-based pressure sensor arrays and methods for data analysis |
| CN104732250A (en) * | 2015-03-25 | 2015-06-24 | 安徽农业大学 | Non-sleep-interference sleeping posture and sleeping behavior testing identification method |
-
2016
- 2016-05-31 CN CN201610378199.0A patent/CN106056116A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120323501A1 (en) * | 2011-05-20 | 2012-12-20 | The Regents Of The University Of California | Fabric-based pressure sensor arrays and methods for data analysis |
| CN104732250A (en) * | 2015-03-25 | 2015-06-24 | 安徽农业大学 | Non-sleep-interference sleeping posture and sleeping behavior testing identification method |
Non-Patent Citations (2)
| Title |
|---|
| JASON J. LIU 等: ""Sleep posture analysis using a dense pressure sensitive bedsheet"", 《PERVASIVE AND MOBILE COMPUTING》 * |
| 周雪: ""基于模糊粗糙集的识别技术及其应用研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106963166A (en) * | 2017-03-30 | 2017-07-21 | 南京信息工程大学 | A kind of adaptively changing and the intelligent bed and its method of work of carrier contact portion stress |
| CN110546473A (en) * | 2017-04-10 | 2019-12-06 | 霓达株式会社 | Judgment system and judgment program |
| CN109925136A (en) * | 2019-03-20 | 2019-06-25 | 河北工业大学 | A kind of intelligent bed system and its application method based on movement intention assessment |
| CN109925136B (en) * | 2019-03-20 | 2021-08-10 | 河北工业大学 | Intelligent bed system based on action intention recognition and use method thereof |
| CN110477669B (en) * | 2019-08-16 | 2021-09-07 | 福建农林大学 | Infrared massage mattress based on body position pressure measurement |
| CN110477669A (en) * | 2019-08-16 | 2019-11-22 | 福建农林大学 | A kind of infrared massaging mattress based on position pressure measurement |
| CN110432724A (en) * | 2019-08-28 | 2019-11-12 | 南通金露智能设备有限公司 | A kind of change with sleeping position and the pillow of automatic controlled height |
| CN110811630A (en) * | 2019-10-31 | 2020-02-21 | 瞬联软件科技（北京）有限公司 | Pregnant woman sleeping posture detection method and device |
| CN110811630B (en) * | 2019-10-31 | 2022-07-22 | 瞬联软件科技（北京）有限公司 | Pregnant woman sleeping posture detection method and device |
| CN110989459A (en) * | 2019-12-23 | 2020-04-10 | 西南交通大学 | Intelligent bed body control system and use method thereof |
| CN111000528A (en) * | 2019-12-23 | 2020-04-14 | 西南交通大学 | Method for detecting sleeping posture of human body |
| CN111310599A (en) * | 2020-01-20 | 2020-06-19 | 重庆大学 | Sleep action recognition system capable of quickly adapting to various change factors |
| CN112001286B (en) * | 2020-08-14 | 2022-07-08 | 燕山大学 | Method and device for adjusting height of neck pillow based on pressure image for sleep posture recognition processing |
| CN112001286A (en) * | 2020-08-14 | 2020-11-27 | 燕山大学 | Neck pillow height adjusting method and device for sleep posture recognition processing based on pressure image |
| CN113273998A (en) * | 2021-07-08 | 2021-08-20 | 南京大学 | Human body sleep information acquisition method and device based on RFID label matrix |
| CN113273998B (en) * | 2021-07-08 | 2022-07-05 | 南京大学 | Human body sleep information acquisition method and device based on RFID label matrix |
| CN113962995B (en) * | 2021-12-21 | 2022-04-19 | 北京鹰瞳科技发展股份有限公司 | Cataract model training method and cataract identification method |
| CN113962995A (en) * | 2021-12-21 | 2022-01-21 | 北京鹰瞳科技发展股份有限公司 | Cataract model training method and cataract identification method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106056116A (en) | Fuzzy rough set-based sleeping posture pressure image recognition method | |
| US11712176B2 (en) | Systems and methods for video-based non-contact tidal volume monitoring | |
| Islam et al. | Multichannel lung sound analysis for asthma detection | |
| CN104834946B (en) | A kind of contactless sleep monitor method and system | |
| CN104887258B (en) | Diagnosis support system | |
| CN104680524B (en) | A kind of leafy vegetable disease screening method | |
| CN108670200A (en) | A kind of sleep sound of snoring classification and Detection method and system based on deep learning | |
| CN109493954A (en) | A kind of SD-OCT image retinopathy detection system differentiating positioning based on classification | |
| CN107330352A (en) | Sleeping position pressure image-recognizing method based on HOG features and machine learning | |
| CN111931733B (en) | Human body posture detection method based on depth camera | |
| CN105869144B (en) | A kind of contactless monitoring of respiration method based on depth image data | |
| WO2020019358A1 (en) | Physiological parameter measuring system and intelligent seat having same | |
| CN110276271A (en) | Merge the non-contact heart rate estimation technique of IPPG and depth information anti-noise jamming | |
| WO2012097436A1 (en) | Method and device for swallowing impairment detection | |
| CN109993068A (en) | A kind of contactless human emotion's recognition methods based on heart rate and facial characteristics | |
| CN108564113A (en) | A kind of tongue fur constitution recognition methods perceived based on deep neural network and complexity | |
| Li et al. | Sleep posture classification with multi-stream CNN using vertical distance map | |
| Ganfure | Using video stream for continuous monitoring of breathing rate for general setting | |
| CN106419938A (en) | Attention deficit hyperactivity disorder (ADHD) detection method and system based on kinetic energy release estimation | |
| Onkaew et al. | Automatic extraction of retinal vessels based on gradient orientation analysis | |
| CN101564323A (en) | Auxiliary equipment for diagnosing galactophore nidus based on galactophore X-ray photograph | |
| CN110693510A (en) | Attention deficit hyperactivity disorder auxiliary diagnosis device and using method thereof | |
| Sari et al. | Best performance comparative analysis of architecture deep learning on ct images for lung nodules classification | |
| CN107169978A (en) | Ultrasonoscopy edge detection method and system | |
| CN101346103A (en) | Method and system for regional assessment of lung physiology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161026 |