CN107545221A - Baby kicks quilt recognition methods, system and device - Google Patents
Baby kicks quilt recognition methods, system and device Download PDFInfo
- Publication number
- CN107545221A CN107545221A CN201610495768.XA CN201610495768A CN107545221A CN 107545221 A CN107545221 A CN 107545221A CN 201610495768 A CN201610495768 A CN 201610495768A CN 107545221 A CN107545221 A CN 107545221A
- Authority
- CN
- China
- Prior art keywords
- feature
- fixed reference
- baby
- image
- reference feature
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of baby to kick quilt recognition methods, system and device, wherein, methods described includes：Gather the original image of baby's monitored space；The original image collected is pre-processed, obtains image to be analyzed；Characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；By the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database respectively compared with, if the special characteristic figure of setting quantity is consistent with corresponding fixed reference feature figure, judge that baby does not kick quilt son；If special characteristic figure does not meet corresponding fixed reference feature figure, then judge that baby kicks quilt son.The current state of dynamic access baby of the present invention, it is not necessary to which prompt message can be sent, accurately and timely in the very first time for kicking open quilt to efficiently identify out whether the quilt of baby is kicked open to parent by carrying out the comparison of great amount of images；Any article need not be dressed to baby, discomfort will not be caused to baby.
Description
Technical field
The present invention relates to monitoring processing technology field, specifically, it is related to a kind of baby and kicks quilt recognition methods, system and dress
Put.
Background technology
With the progress of modern society, child turns into the center of gravity of one family, and baby does not understand how to deprotect oneself, often
Chang Yinwei, which sleeps disorderly to turn at night, disorderly to be kicked and suffers from cold.Due to the policy issue of baby, medicine using again be greatly limited, in
It is parent's just guarding beside baby day and night, causes the fatigue strength of parent increasingly to increase.It is that baby suffers from cold, be raw to kick quilt son
One great inducement of disease, thus, people devise various schemes to prevent baby from kicking quilt son.
A kind of simplest scheme is to use baby bunting.Baby is directly wrapped in sleeping bag, although preventing baby
Quilt is kicked open into the problem of suffering from cold, but significantly limit the activity of baby, is unfavorable for the development of baby.
Other scheme is to wear alarm to baby, and these alarms are divided into two kinds, one kind to gather the difference of signal
It is the alarm using temperature detector.The temperature signal for detecting to obtain according to the temperature detector with baby, if baby
Quilt is kicked open, because warming without quilt, then the Wen Duxiajiang of baby, now alarm equipment alarm, notifies caregiver.This side
It although method can detect that baby kicks quilt son, can not be alarmed in the very first time, only decline alarm device just in infant temperature
It can work, and now child may kick open quilt, the baby of constitution difference may now suffer from cold.Another kind is using weight
Force snesor gathers gravitational cue, and when baby starts to turn from side to side, the gravitational cue of the gravity sensor sensing of wearing changes,
Then alarm.Both modes will give the utensil of some on baby harness, the comfort of baby can be influenceed, so as to cry and scream.
The content of the invention
The technical problem to be solved in the present invention is, in view of the shortcomings of the prior art, there is provided a kind of baby kicks quilt identification side
Method, system and device, need not be in the case that baby wears any utensil, accurately and timely to identify that baby kicks quilt son
State.
In order to solve above-mentioned technical problem, the invention provides a kind of baby to kick quilt recognition methods, wherein, including：
Gather the original image of baby's monitored space；
The original image collected is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；
By the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database respectively compared with, if setting
The special characteristic figure of quantity is consistent with corresponding fixed reference feature figure, then judges that baby does not kick quilt son；If special characteristic figure
Corresponding fixed reference feature figure is not met, then judges that baby kicks quilt son.
Preferably, it is described to be compared the multiple characteristic pattern respectively with the corresponding fixed reference feature figure stored in database
Compared with the step of include：
Calculate the characteristic value of the multiple characteristic pattern；
The characteristic value of the multiple characteristic pattern is respectively compared whether positioned at the fixed reference feature value scope for corresponding to fixed reference feature figure
It is interior；
Whether it is consistent by following steps judging characteristic figure with corresponding fixed reference feature figure：
If the characteristic value of characteristic pattern is located in the range of corresponding fixed reference feature value, characteristic pattern and corresponding fixed reference feature figure
It is consistent；If the characteristic value of characteristic pattern exceedes corresponding fixed reference feature value scope, the characteristic pattern does not meet corresponding with reference to special
Sign figure.
Preferably, described the step of carrying out characterization to the image to be analyzed, generating multiple characteristic patterns, includes：
In the image of predetermined quantity, identify that feature shines from each two field picture, obtain multiple feature and shine；
Calculate picture rate of each feature according to the picture frame shone with obtaining this feature；
The picture rate shone according to each feature, the multiple feature is shone and is combined into a complete feature
According to；
According to default characteristic pattern quantity, the feature is shone and is divided into multiple characteristic patterns.
Preferably, after being combined into a complete feature and shining, the complete feature is shone and is smoothed；
It is described according to default characteristic pattern quantity, be specially according to the step of being divided into multiple characteristic patterns by the feature：
According to default characteristic pattern quantity, the feature Jing Guo smoothing processing is shone and is divided into multiple characteristic patterns.
Preferably, fixed reference feature figure and fixed reference feature the value scope is present in database, is obtained by following steps：
Gather original image during infant sleep；
The original image is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple fixed reference feature figures for embodying different characteristic；
Calculate the characteristic value of the multiple fixed reference feature figure；
The repeatedly characterization to image to be analyzed and the calculating process to characteristic value, obtain it is multiple meet it is same
The different characteristic value of the fixed reference feature figure of feature, obtain fixed reference feature value scope.
Preferably, the step of described pair of original image collected pre-processes includes：
The original image is subjected to gray processing processing, obtains gray level image.
Correct the direction of the gray level image；
The picture main body to be used is intercepted from the gray level image after the correction, as image to be analyzed.
Preferably, when gathering the original image, using the search window Multiple-Scan monitored space baby of different proportion,
To obtain the sizeable original image of target object.
Present invention also offers a kind of baby to kick quilt identifying system, wherein, including：
Image capture module, for gathering the original image of baby's monitored space；
Image pre-processing module, for being pre-processed to the original image collected, obtaining image to be analyzed；With
Module is characterized, for carrying out characterization to the image to be analyzed, generates multiple characteristic patterns；
Comparison module, for the multiple characteristic pattern to be carried out respectively with the corresponding fixed reference feature figure stored in database
Compare；
Judge module, for judging baby with whether corresponding fixed reference feature figure is consistent according to the special characteristic figure of setting quantity
Whether youngster kicks quilt son.
Preferably, described image pretreatment module includes：
Gray processing processing unit, for the original image to be carried out into gray processing processing, obtain gray level image；
Image flame detection unit, for correcting the direction of the gray level image；With
Picture interception unit, for intercepting the picture main body to be used from the gray level image after the correction, as
Image to be analyzed.
Preferably, the characterization module includes：
Feature shines acquiring unit, in the original image of predetermined quantity, one to be identified from each two field picture
Feature shines, and is shone so as to obtain multiple feature；
Picture rate computing unit, for calculating picture of each feature according to the picture frame shone with obtaining this feature
Ratio；
Assembled unit, for the picture rate shone according to each feature, the multiple feature is shone and is combined into one
Individual complete feature is shone；
Cutting unit, for according to default characteristic pattern quantity, the feature being shone and is divided into multiple characteristic patterns.
Preferably, the comparison module includes：
Computing unit, for calculating the characteristic value of the multiple characteristic pattern；
Whether comparison unit, the characteristic value for more the multiple characteristic pattern are special positioned at the reference for corresponding to fixed reference feature figure
In the range of value indicative；
Whether the judge module is according to the characteristic value of characteristic pattern positioned at the fixed reference feature value scope for corresponding to fixed reference feature figure
Whether interior judging characteristic figure is consistent with fixed reference feature figure.
Preferably, the judge module is judging that obtaining baby kicks quilt the period of the day from 11 p.m. to 1 a.m, sends a signal to corresponding warning device.
Preferably, described image acquisition module includes ratio adjustment unit, adjusts original image for the collection
The size of search window.
Preferably, in addition to：
Database, for storing fixed reference feature figure and corresponding fixed reference feature value scope.
Present invention also offers a kind of baby to kick quilt identification device, wherein, including：
Camera module, for gathering the original image of baby's monitored space；
Memory, for store instruction and related data；
Processor, configured according to the instruction as follows：
Gather the original image of baby's monitored space；
The original image collected is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；
By the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database respectively compared with, if setting
The special characteristic figure of quantity is consistent with corresponding fixed reference feature figure, then judges that baby does not kick quilt son；If special characteristic figure
Corresponding fixed reference feature figure is not met, then judges that baby kicks quilt son.
The multiple characteristic pattern is entered respectively with the corresponding fixed reference feature figure stored in database preferably for described
The instruction that row compares, the processor are further configured to：
Calculate the characteristic value of the multiple characteristic pattern；
The characteristic value of the multiple characteristic pattern is respectively compared whether positioned at the fixed reference feature value scope for corresponding to fixed reference feature figure
It is interior；
Whether it is consistent by following steps judging characteristic figure with corresponding fixed reference feature figure：
If the characteristic value of characteristic pattern is located in the range of corresponding fixed reference feature value, characteristic pattern and corresponding fixed reference feature figure
It is consistent；If the characteristic value of characteristic pattern exceedes corresponding fixed reference feature value scope, the characteristic pattern does not meet corresponding with reference to special
Sign figure.
Characterization is carried out to the image to be analyzed preferably for described, generates the instruction of multiple characteristic patterns, institute
Processor is stated to be further configured to：
In the image of predetermined quantity, identify that feature shines from each two field picture, obtain multiple feature and shine；
Calculate picture rate of each feature according to the picture frame shone with obtaining this feature；
The picture rate shone according to each feature, the multiple feature is shone and is combined into a complete feature
According to；
According to default characteristic pattern quantity, the feature is shone and is divided into multiple characteristic patterns.
The present invention is based on image recognition technology, the current state of dynamic access baby, it is not necessary to carries out the ratio of great amount of images
Compared with can be sent and carried to parent in the very first time for kicking open quilt with effectively identifying whether the quilt of baby is kicked open
Show information, accurately and timely；Any article need not be dressed to baby, discomfort will not be caused to baby.
Brief description of the drawings
By referring to description of the following drawings to the embodiment of the present invention, above-mentioned and other purpose of the invention, feature and
Advantage will be apparent from, in the accompanying drawings：
Fig. 1 is the overall procedure schematic diagram that baby of the present invention kicks quilt the embodiment of recognition methods one；
Fig. 2 is the handling process schematic diagram of characterization of the present invention；
Fig. 3 is characterized the forming process schematic diagram of photograph and characteristic pattern；
Fig. 4 be judging characteristic figure of the present invention whether the schematic flow sheet being consistent with fixed reference feature figure；
Fig. 5 is fixed reference feature map generalization schematic flow sheet in database of the present invention；
Fig. 6 is the schematic flow sheet of the present invention to original image pretreatment；
Fig. 7 A-7B are that baby of the present invention kicks quilt comparison diagram of the image before and after correction in recognition methods；
Fig. 8 is the principle assumption diagram that baby of the present invention kicks quilt identifying system；
Fig. 9 is the principle assumption diagram of image pre-processing module of the present invention；
Figure 10 is the principle assumption diagram of characterization module of the present invention；
Figure 11 is the principle assumption diagram of comparison module of the present invention；
Figure 12 is the principle assumption diagram that baby of the present invention kicks quilt identification device；
Figure 13 is the structural principle structure chart of an Application Example of the invention.
Embodiment
Below based on embodiment, present invention is described, but the present invention is not restricted to these embodiments.Under
It is detailed to describe some specific detail sections in the literary detailed description to the present invention.Do not have for a person skilled in the art
The description of these detail sections can also understand the present invention completely.In order to avoid obscuring the essence of the present invention, known method, mistake
Journey, flow do not describe in detail.What other accompanying drawing was not necessarily drawn to scale.
Flow chart, block diagram in accompanying drawing illustrate the possible system frame of the system of the embodiment of the present invention, method, apparatus
Frame, function and operation, the square frame on flow chart and block diagram can represent a module, program segment or only one section of code, institute
It is all the executable instruction for realizing regulation logic function to state module, program segment and code.It should also be noted that described realize rule
Determining the executable instruction of logic function can reconfigure, so as to generate new module and program segment.Therefore accompanying drawing square frame with
And square frame order is used only to the process and step of preferably diagram embodiment, without should be in this, as the limit to invention itself
System.
As shown in figure 1, the overview flow chart of the embodiment of recognition methods one is kicked quilt for a kind of baby provided by the invention, at this
In embodiment, methods described includes：
Step S1, gather the original image of baby's monitored space；
Step S2, the original image collected is pre-processed, obtains image to be analyzed；
Step S3, characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；
Step S4, one is taken out from multiple characteristic patterns of generation；
Whether step S5, characteristic pattern fixed reference feature figure corresponding with database are consistent, if be not consistent, hold
Row step S6, if be consistent, perform step S8；
Step S6, judge whether the characteristic pattern is specific characteristic pattern, i.e., whether the existing feature of this feature chart is generation
Table may determine that whether baby kicks quilt the position of son, such as her chest, belly, leg, and rule of thumb, this kind of position can be with
As judging whether to kick quilt the feature of son, and the feature quantity for determining that baby covers quilt is set, when all meeting these features,
Then may determine that and do not kick quilt son, as long as and there is one not meet, just it is considered that kick quilt son.If with fixed reference feature figure not phase
The characteristic pattern of symbol is these specific characteristic patterns, then judges that baby has kicked quilt son in step S7.If not specific characteristic pattern,
Need to be further continued for comparing, then go to step S8；
Step S8, judge whether the special characteristic figure of completeer predetermined quantity, such as 5 special characteristic figures, if do not had
Have, then return to step S4, continue to take characteristic pattern to be compared, if having reached predetermined quantity, judge that baby does not have in step S9
Kick quilt son.
In abovementioned steps, the present invention is using principle component analysis (Principal Component Analysis, abbreviation
PCA) image to be analyzed is analyzed and processed.The principle component analysis is a kind of data analysing method, by linearly becoming
Initial data of changing commanders is transformed to the expression of one group of each dimension linear independence, available for the principal character component of extraction data, commonly uses
In the dimensionality reduction of high dimensional data.In the present invention, the image collected is divided into multiple characteristic patterns according to principle component analysis, carried out
When picture contrasts, schemed by contrast characteristic, not only considerably reduce data processing amount, and reached enough precision.
As shown in Fig. 2 to carry out characterization to the image to be analyzed in step S3, the tool of multiple characteristic patterns is generated
Body handling process：
Step S311, it in the picture frame of predetermined quantity, will identify that feature shines from each two field picture.Wherein, exist
When identification feature shines, as shown in figure 3, taking out a two field picture n1, identified since the body region of the two field picture, i.e., first
Then individual feature removes a two field picture n2, in the two field picture, is removing main body according to the body region that m1 is current image frame
A secondary regions behind region identify that second feature shines m2 ..., by that analogy, obtain multiple feature and shine m1,
M2, m3 ....
Step S312, calculate picture rate of each feature according to the picture frame shone with obtaining this feature；
Step S313, the picture rate shone according to each feature are complete according to being combined into one by the multiple feature
Whole feature shines IM1.
For example, it is probably into its component relationship is as follows by 100 group of picture that a feature, which is shone,：
(baby)+(11.5%of baby1)-(36.3%of baby2)+(7.2%of baby3)+...+(0.1%of
baby99)
Percentage therein is picture rate, and babyn is that n-th feature shines.
Wherein, ratio comes several features of foremost according to showing that some main sons that kick quilt are easier to expose
The place come, and if sub- license-master's picture noise of last feature etc..
Handled by combinations thereof, the current sleep shape of child can be reappeared, the processing of image one by one is avoided, not
In the case of influenceing accuracy rate, reduce data processing amount.
Step S314, to the feature combined according to being smoothed, obtain smooth feature and shine IM2；
Step S315, according to default characteristic pattern quantity, the feature Jing Guo smoothing processing is divided into multiple characteristic patterns according to IM2
C1, c2, c3 ....
In step s 5, as shown in figure 4, judging that the characteristic pattern is corresponding with database with reference to special by following steps
Whether sign figure is consistent：
Step S51, calculate the characteristic value of the multiple characteristic pattern；
Step S52, take the characteristic value of a characteristic pattern；
Step S53, whether the characteristic value of the characteristic pattern is positioned at the fixed reference feature value scope for corresponding to fixed reference feature figure
It is interior；I.e. compared with two endpoint values of the fixed reference feature value scope, if the characteristic value is more than or equal to small end point
Value, less than or equal to larger endpoint value, that is to say, that the characteristic value of characteristic pattern is located in the range of corresponding fixed reference feature value, then
It can determine whether that the characteristic value of the characteristic pattern is located in the range of the fixed reference feature value of corresponding fixed reference feature figure, judge institute in step S54
Characteristic pattern fixed reference feature figure corresponding with database is stated to be consistent；If the characteristic value is less than small end point value, or, greatly
In larger endpoint value, that is to say, that the characteristic value of characteristic pattern judges beyond corresponding fixed reference feature value scope in step S55
Characteristic pattern fixed reference feature figure corresponding with database is not consistent.
In the present invention, in order to realize that the image to obtain in real time provides contrasting foundation when being compared in monitoring,
Database is provided with, wherein storing fixed reference feature photograph, according to being made up of multiple fixed reference feature figures, each joins the fixed reference feature
The corresponding fixed reference feature value scope of characteristic pattern is examined, is contrasted for the characteristic pattern with being obtained during monitoring in real time, to determine baby
The current lid of youngster is by subcase.Specifically, as shown in figure 5, the outline flowchart generated for fixed reference feature figure in database, the number
Shone according to the fixed reference feature in storehouse, fixed reference feature figure and its range of characteristic values are obtained by following steps：
Step S1a, original image during infant sleep is gathered, including cover the image of quilt and do not cover the figure of quilt
Picture.
Step S2a, the original image is pre-processed, obtains image to be analyzed.
Step S3a, characterization as shown in Figure 2 is carried out to the image to be analyzed, obtain fixed reference feature and shine and pre-
The fixed reference feature figure of fixed number amount.
Wherein, each described fixed reference feature figure embodies a feature, such as some body part of baby.According to original
Image, what described fixed reference feature figure embodied is probably the feature for being covered with quilt and not being covered with quilt, if what is embodied is lid
The feature of quilt, then is effective fixed reference feature figure.
Step S4a, calculate the characteristic value of the effective fixed reference feature figure.
Step S5a, judges whether the calculating to a fixed reference feature value reaches predetermined number of computations.Due to the change of action
Change, the reason such as the difference of picture collection, for same feature characteristic value can difference, it is then desired to be directed to same feature,
Multiple characteristic patterns are generated, so as to obtain multiple characteristic values, and then a range of characteristic values is obtained, essence can be just provided for real time contrast
True contrasting foundation.Predetermined quantity described here, it is the more reliable quantity of one obtained by repetition test.If
Predetermined quantity is not reaching to, then return to step S1a, a new wheel image processing process is re-started, to obtain for same
Feature, obtain characteristic value.If now having reached predetermined quantity, step S6a is performed, according to by multiple characterization
The different characteristic value of obtained multiple fixed reference feature figures for meeting same feature, it is determined that corresponding fixed reference feature value scope.
Step S7a, by described fixed reference feature figure and its storage of fixed reference feature value scope into database.
Wherein, in fixed reference feature figure, it is these spies showed with particular reference to characteristic pattern with particular reference to characteristic pattern to have some
Sign represent may determine that baby whether kick quilt son position, such as her chest, belly, leg, rule of thumb, this kind of position
Can be as the feature for judging whether to kick quilt son, and cover according to baby the situation of quilt, it may be determined that such feature number
Amount, for example, 5 in one embodiment, what is set certainly is more, compares in real-time monitoring process more, also more smart
Really, it is but also time-consuming longer, thus, it by setting appropriate quantity, both can quickly identify, and also can guarantee that recognition accuracy.
The total quantity of fixed reference feature figure can be set as it is multiple, and by they by judgment accuracy ascending order arrange, will be specific
Before fixed reference feature figure comes, such as the first fixed reference feature figure is that can most judge whether baby currently covers the characteristic pattern of quilt, the
Two fixed reference feature figures take second place, and sequentially successively decrease.During monitoring, if drawn according to monitoring image with the first fixed reference feature figure
Corresponding characteristic pattern does not meet the first fixed reference feature value scope, then may determine that now baby does not cover quilt.In setting number
The characteristic value of the special characteristic figure (characteristic pattern as corresponding to preceding 5 fixed reference feature figures) of amount is all in the spy of corresponding fixed reference feature figure
In the range of value indicative, then it may determine that baby does not kick quilt son.
In foregoing monitoring process and fixed reference feature figure generating process in real time, it is directed to and the original image collected is entered
The step of row pretreatment, specifically as shown in fig. 6, comprising the following steps：
Step S21, the original image is subjected to gray processing processing, obtains gray level image.
Step S22, correct the direction of the gray level image.Certain angle is had due to the original image of collection, it is necessary to
Rotated, you can with 360 rotations, while rotation, be recorded in the projection of x (or Y) direction of principal axis, maximized in the shadow that sets
When, stop the rotation, as illustrated in figures 7 a-b, Fig. 7 A are the image before correction, and Fig. 7 B are the image after correction.
Step S23, the picture main body to be used is intercepted from the gray level image after the correction, as figure to be analyzed
Picture.Unwanted part in image range is cropped, only intercepts the region needed to use.
The invention also discloses a kind of baby to kick quilt identifying system, as shown in figure 8, the system includes image capture module
1st, image pre-processing module 2, characterization module 3, comparison module 4 and judge module 5.Wherein, described image acquisition module 1 is used for
Gather the original image of baby's monitored space；Described image pretreatment module 2 is used to locate the original image to collecting in advance
Reason, such as the adjustment of angle, size, obtain image to be analyzed；Module 3 is characterized to characterize the image to be analyzed
Processing, generates multiple characteristic patterns；Comparison module 4 is by the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database
It is compared respectively；Whether judge module 5 is consistent to sentence according to the special characteristic figure of setting quantity with corresponding fixed reference feature figure
Whether disconnected baby kicks quilt son.
In addition, the system also includes database 6, for storing various data, such as following involved fixed reference feature
The data such as figure and its range of characteristic values.
Wherein, described image pretreatment module 2, as shown in figure 9, including gray processing processing unit 21, image flame detection unit
22 and picture interception unit 23.Wherein, the gray processing processing unit 21, for will carry out the original image at gray processing
Reason, obtains gray level image；Described image correcting unit 22 is used for the direction for correcting the gray level image；The picture interception unit
23 are used to intercept the picture main body to be used from the gray level image after the correction, as image to be analyzed.
The characterization module 3 is as shown in Figure 10, including feature according to acquiring unit 31, picture rate computing unit 32,
Assembled unit 33 and cutting unit 34.Wherein, feature according to acquiring unit 31 in the original image of predetermined quantity, from every
Identify that feature shines in one two field picture, shone so as to obtain multiple feature；The picture rate computing unit 32 calculates
Picture rate of each feature according to the picture frame shone with obtaining this feature；The assembled unit 33 is according to each feature
The picture rate that son shines, the multiple feature is shone according to a complete feature is combined into；The cutting unit 34 is according to pre-
If characteristic pattern quantity, by the feature according to being divided into multiple characteristic patterns.
Comparison module 4 of the present invention is as shown in figure 11, including computing unit 41 and comparison unit 42, wherein, it is described
Computing unit 41 calculates the characteristic value of the multiple characteristic pattern；The characteristic value of the 42 more the multiple characteristic pattern of comparison unit
Whether in the range of the fixed reference feature value of corresponding fixed reference feature figure；The judge module 5 according to the characteristic value of characteristic pattern whether
Whether judging characteristic figure is consistent with fixed reference feature figure in the range of the fixed reference feature value of corresponding fixed reference feature figure.It is specific treated
Each flow of journey method as previously described, will not be repeated here.
Described image acquisition module is in addition to the device that normal image gathers, in addition to ratio adjustment unit, adjusts and is used for
Gather the size of the search window of original image.When carrying out IMAQ, with hop algorithm, the big of search window can be adjusted
It is small, by the scaling of proper proportion, the suitable original image of target object size can be disposably collected, after collection
The size of image is adjusted again, more effectively.
Based on the above method and system, the invention also discloses a kind of baby to kick quilt identification device, as shown in figure 12, bag
Include：Camera module, memory and processor, wherein, the original image of camera module collection baby's monitored space.Such as hand
Cam device in machine, the image of target object is obtained by a series of optical element, circuit element.Can be multiple quiet
State image (for example with camera continuous shooting module, and sets the shooting cycle, several still images are gathered every a cycle), also may be used
To be dynamic video image, collection is video image.
Memory is used to store identification instruction and related data.The instruction includes being used to realize whether baby kicks quilt son
Recognizer instructs, and it is the instruction algorithm for realizing aforementioned identification method.The related data includes the image of collection, place
Intermediate data during reason, the fixed reference feature figure for contrast and fixed reference feature value etc..
Processor is configured according to the instruction, completes programmed instruction, realizes the identification that son whether is kicked quilt to baby.Tool
Body, configuration is as follows：
Gather the original image of baby's monitored space；
The original image collected is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；
By the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database respectively compared with, if setting
The special characteristic figure of quantity is consistent with corresponding fixed reference feature figure, then judges that baby does not kick quilt son；If special characteristic figure
Corresponding fixed reference feature figure is not met, then judges that baby kicks quilt son.
Wherein, the multiple characteristic pattern is carried out respectively with the corresponding fixed reference feature figure stored in database for described
The instruction compared, the processor are further configured to：
Calculate the characteristic value of the multiple characteristic pattern；
The characteristic value of the multiple characteristic pattern is respectively compared whether positioned at the fixed reference feature value scope for corresponding to fixed reference feature figure
It is interior；
Whether it is consistent by following steps judging characteristic figure with corresponding fixed reference feature figure：
If the characteristic value of characteristic pattern is located in the range of corresponding fixed reference feature value, characteristic pattern and corresponding fixed reference feature figure
It is consistent；If the characteristic value of characteristic pattern exceedes corresponding fixed reference feature value scope, the characteristic pattern does not meet corresponding with reference to special
Sign figure.
Characterization is carried out to the image to be analyzed for described, generates the instruction of multiple characteristic patterns, the processing
Device is further configured to：
In the image of predetermined quantity, identify that feature shines from each two field picture, obtain multiple feature and shine；
Calculate picture rate of each feature according to the picture frame shone with obtaining this feature；
The picture rate shone according to each feature, the multiple feature is shone and is combined into a complete feature
According to；
According to default characteristic pattern quantity, the feature is shone and is divided into multiple characteristic patterns.
System of the present invention can match use with current any warning device, as shown in figure 13, including this hair
Bright described baby kicks quilt identifying system 101 and a device that alarm sound can be sent with receive information, such as mobile phone 200, its
In there is buzzer 201.Wherein, baby kicks quilt identifying system 101 and may be mounted on another mobile phone 100.When baby kicks quilt
Identifying system 101 obtains baby according to preceding method and process monitoring and kicks quilt the period of the day from 11 p.m. to 1 a.m, sends corresponding signal to mobile phone 200, hand
After machine 200 receives the signal, control buzzer 201 pipes, for reminding child parent to kick quilt son.So as to be kicked quilt in child
The very first time of son notifies its ward.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention, for those skilled in the art
For, the present invention can have various changes and change.All any modifications made within spirit and principles of the present invention, it is equal
Replace, improve etc., it should be included in the scope of the protection.
Claims (17)
1. a kind of baby kicks quilt recognition methods, wherein, including：
Gather the original image of baby's monitored space；
The original image collected is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；
By the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database respectively compared with, if setting quantity
Special characteristic figure be consistent with corresponding fixed reference feature figure, then judge that baby does not kick quilt son；If special characteristic figure is not inconsistent
Corresponding fixed reference feature figure is closed, then judges that baby kicks quilt son.
2. baby as claimed in claim 1 kicks quilt recognition methods, wherein, it is described by the multiple characteristic pattern with being deposited in database
The step of corresponding fixed reference feature figure of storage is compared respectively includes：
Calculate the characteristic value of the multiple characteristic pattern；
The characteristic value of the multiple characteristic pattern is respectively compared whether in the range of the fixed reference feature value of corresponding fixed reference feature figure；
Whether it is consistent by following steps judging characteristic figure with corresponding fixed reference feature figure：
If the characteristic value of characteristic pattern is located in the range of corresponding fixed reference feature value, characteristic pattern and corresponding fixed reference feature figure phase
Symbol；If the characteristic value of characteristic pattern exceedes corresponding fixed reference feature value scope, the characteristic pattern does not meet corresponding fixed reference feature
Figure.
3. baby as claimed in claim 1 kicks quilt recognition methods, wherein, it is described that the image to be analyzed is carried out at characterization
The step of managing, generating multiple characteristic patterns includes：
In the image of predetermined quantity, identify that feature shines from each two field picture, shone so as to obtain multiple feature；
Calculate picture rate of each feature according to the picture frame shone with obtaining this feature；
The picture rate shone according to each feature, the multiple feature is shone according to a complete feature is combined into；
According to default characteristic pattern quantity, the feature is shone and is divided into multiple characteristic patterns.
4. baby as claimed in claim 3 kicks quilt recognition methods, wherein, it is right after being combined into a complete feature and shining
The complete feature, which is shone, to be smoothed；
It is described according to default characteristic pattern quantity, be specially according to the step of being divided into multiple characteristic patterns by the feature：
According to default characteristic pattern quantity, the feature Jing Guo smoothing processing is shone and is divided into multiple characteristic patterns.
5. baby as claimed in claim 1 kicks quilt recognition methods, wherein, the fixed reference feature figure and reference in the database are special
Value indicative scope is obtained by following steps：
Gather original image during infant sleep；
The original image is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple fixed reference feature figures for embodying different characteristic；
Calculate the characteristic value of the multiple fixed reference feature figure；
The repeatedly characterization to image to be analyzed and the calculating process to characteristic value, obtain multiple meeting same feature
Fixed reference feature figure different characteristic value, so as to obtain fixed reference feature value scope.
6. the baby as described in claim 1 or 5 kicks quilt recognition methods, wherein, the described pair of original image collected carries out pre-
The step of processing, includes：
The original image is subjected to gray processing processing, obtains gray level image；
Correct the direction of the gray level image；
The picture main body to be used is intercepted from the gray level image after the correction, as image to be analyzed.
7. the baby as described in claim 1 or 6 kicks quilt recognition methods, wherein, when gathering the original image, using difference
The search window Multiple-Scan monitored space of ratio, to obtain the sizeable original image of target object.
8. a kind of baby kicks quilt identifying system, wherein, including：
Image capture module, for gathering the original image of baby's monitored space；
Image pre-processing module, for being pre-processed to the original image collected, obtain image to be analyzed；
Module is characterized, for carrying out characterization to the image to be analyzed, generates multiple characteristic patterns；
Comparison module, for the multiple characteristic pattern to be compared respectively with the corresponding fixed reference feature figure stored in database
Compared with；With
Judge module, for judging that baby is with whether corresponding fixed reference feature figure is consistent according to the special characteristic figure of setting quantity
It is no to kick quilt son.
9. baby as claimed in claim 8 kicks quilt identifying system, wherein, described image pretreatment module includes：
Gray processing processing unit, for the original image to be carried out into gray processing processing, obtain gray level image；
Image flame detection unit, for correcting the direction of the gray level image；With
Picture interception unit, for intercepting the picture main body to be used from the gray level image after the correction, as treating point
Analyse image.
10. baby as claimed in claim 8 kicks quilt identifying system, wherein, the characterization module includes：
Feature shines acquiring unit, in the original image of predetermined quantity, identifying a feature from each two field picture
Son shines, and is shone so as to obtain multiple feature；
Picture rate computing unit, for calculating picture ratio of each feature according to the picture frame shone with obtaining this feature
Rate；
Assembled unit, it is for the picture rate shone according to each feature, the multiple feature is complete according to being combined into one
Whole feature is shone；With
Cutting unit, for according to default characteristic pattern quantity, the feature being shone and is divided into multiple characteristic patterns.
11. baby as claimed in claim 8 kicks quilt identifying system, wherein, the comparison module includes：
Computing unit, for calculating the characteristic value of the multiple characteristic pattern；
Comparison unit, whether the characteristic value for more the multiple characteristic pattern is positioned at the fixed reference feature value for corresponding to fixed reference feature figure
In the range of；With
The judge module is sentenced according to whether the characteristic value of characteristic pattern is located in the range of the fixed reference feature value of corresponding fixed reference feature figure
Whether disconnected characteristic pattern is consistent with fixed reference feature figure.
12. baby as claimed in claim 8 kicks quilt identifying system, wherein, the judge module is judging that obtaining baby kicks quilt
The period of the day from 11 p.m. to 1 a.m, send a signal to corresponding warning device.
13. baby as claimed in claim 8 kicks quilt identifying system, wherein, it is single that described image acquisition module includes ratio adjustment
Member, adjust the size of the search window of original image for the collection.
14. baby as claimed in claim 8 kicks quilt identifying system, wherein, in addition to：
Database, for storing fixed reference feature figure and corresponding fixed reference feature value scope.
15. a kind of baby kicks quilt identification device, wherein, including：
Camera module, for gathering the original image of baby's monitored space；
Memory, for store instruction and related data；With
Processor, configured according to the instruction as follows：
Gather the original image of baby's monitored space；
The original image collected is pre-processed, obtains image to be analyzed；
Characterization is carried out to the image to be analyzed, generates multiple characteristic patterns；
By the multiple characteristic pattern and the corresponding fixed reference feature figure that is stored in database respectively compared with, if setting quantity
Special characteristic figure be consistent with corresponding fixed reference feature figure, then judge that baby does not kick quilt son；If special characteristic figure is not inconsistent
Corresponding fixed reference feature figure is closed, then judges that baby kicks quilt son.
16. baby as claimed in claim 15 kicks quilt identification device, wherein, for described by the multiple characteristic pattern and data
The instruction that the corresponding fixed reference feature figure stored in storehouse is compared respectively, the processor are further configured to：
Calculate the characteristic value of the multiple characteristic pattern；
The characteristic value of the multiple characteristic pattern is respectively compared whether in the range of the fixed reference feature value of corresponding fixed reference feature figure；
Whether it is consistent by following steps judging characteristic figure with corresponding fixed reference feature figure：
If the characteristic value of characteristic pattern is located in the range of corresponding fixed reference feature value, characteristic pattern and corresponding fixed reference feature figure phase
Symbol；If the characteristic value of characteristic pattern exceedes corresponding fixed reference feature value scope, the characteristic pattern does not meet corresponding fixed reference feature
Figure.
17. baby as claimed in claim 5 kicks quilt identification device, wherein, spy is carried out to the image to be analyzed for described
Signization processing, generates the instruction of multiple characteristic patterns, the processor is further configured to：
In the image of predetermined quantity, identify that feature shines from each two field picture, obtain multiple feature and shine；
Calculate picture rate of each feature according to the picture frame shone with obtaining this feature；
The picture rate shone according to each feature, the multiple feature is shone according to a complete feature is combined into；
According to default characteristic pattern quantity, the feature is shone and is divided into multiple characteristic patterns.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610495768.XA CN107545221A (en) | 2016-06-28 | 2016-06-28 | Baby kicks quilt recognition methods, system and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610495768.XA CN107545221A (en) | 2016-06-28 | 2016-06-28 | Baby kicks quilt recognition methods, system and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107545221A true CN107545221A (en) | 2018-01-05 |
Family
ID=60965890
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610495768.XA Pending CN107545221A (en) | 2016-06-28 | 2016-06-28 | Baby kicks quilt recognition methods, system and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107545221A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580466A (en) * | 2019-09-05 | 2019-12-17 | 深圳市赛为智能股份有限公司 | infant quilt kicking behavior recognition method and device, computer equipment and storage medium |
| CN113221802A (en) * | 2021-05-24 | 2021-08-06 | 新疆爱华盈通信息技术有限公司 | Quilt kicking identification method and device and electronic equipment |
| CN113465155A (en) * | 2021-06-23 | 2021-10-01 | 深圳市海清视讯科技有限公司 | Method, system, host and storage medium for reducing quilt kicking behavior of infants |
| CN113974933A (en) * | 2021-08-18 | 2022-01-28 | 宁波星巡智能科技有限公司 | Infant quilt kicking detection method and device, electronic equipment and medium |
| CN114360207A (en) * | 2021-12-15 | 2022-04-15 | 好孩子儿童用品有限公司 | Children play by detecting system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789672A (en) * | 2012-07-03 | 2012-11-21 | 北京大学深圳研究生院 | Intelligent identification method and device for baby sleep position |
| CN105243395A (en) * | 2015-11-04 | 2016-01-13 | 东方网力科技股份有限公司 | Human body image comparison method and device |
| CN105335469A (en) * | 2015-09-29 | 2016-02-17 | 武汉传神信息技术有限公司 | Method and device for image matching and retrieving |
| US20160140399A1 (en) * | 2014-11-17 | 2016-05-19 | Canon Kabushiki Kaisha | Object detection apparatus and method therefor, and image recognition apparatus and method therefor |
-
2016
- 2016-06-28 CN CN201610495768.XA patent/CN107545221A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789672A (en) * | 2012-07-03 | 2012-11-21 | 北京大学深圳研究生院 | Intelligent identification method and device for baby sleep position |
| US20160140399A1 (en) * | 2014-11-17 | 2016-05-19 | Canon Kabushiki Kaisha | Object detection apparatus and method therefor, and image recognition apparatus and method therefor |
| CN105335469A (en) * | 2015-09-29 | 2016-02-17 | 武汉传神信息技术有限公司 | Method and device for image matching and retrieving |
| CN105243395A (en) * | 2015-11-04 | 2016-01-13 | 东方网力科技股份有限公司 | Human body image comparison method and device |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580466A (en) * | 2019-09-05 | 2019-12-17 | 深圳市赛为智能股份有限公司 | infant quilt kicking behavior recognition method and device, computer equipment and storage medium |
| CN113221802A (en) * | 2021-05-24 | 2021-08-06 | 新疆爱华盈通信息技术有限公司 | Quilt kicking identification method and device and electronic equipment |
| CN113465155A (en) * | 2021-06-23 | 2021-10-01 | 深圳市海清视讯科技有限公司 | Method, system, host and storage medium for reducing quilt kicking behavior of infants |
| CN113465155B (en) * | 2021-06-23 | 2022-07-26 | 深圳市海清视讯科技有限公司 | Method, system, host and storage medium for reducing quilt kicking behavior of infants |
| CN113974933A (en) * | 2021-08-18 | 2022-01-28 | 宁波星巡智能科技有限公司 | Infant quilt kicking detection method and device, electronic equipment and medium |
| CN113974933B (en) * | 2021-08-18 | 2024-03-12 | 宁波星巡智能科技有限公司 | Infant quilt kicking detection method and device, electronic equipment and medium |
| CN114360207A (en) * | 2021-12-15 | 2022-04-15 | 好孩子儿童用品有限公司 | Children play by detecting system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107545221A (en) | Baby kicks quilt recognition methods, system and device | |
| Saleheen et al. | puffMarker: a multi-sensor approach for pinpointing the timing of first lapse in smoking cessation | |
| CN111881770B (en) | Face recognition method and system | |
| CA2891834C (en) | Method and system for making tailored garments | |
| CN109643448A (en) | Fine granularity object identification in robot system | |
| CN109214373A (en) | A kind of face identification system and method for attendance | |
| CN108095694B (en) | State monitoring method and device of wearable equipment and wearable equipment | |
| CN106650666A (en) | Method and device for detection in vivo | |
| WO2016160794A1 (en) | Setting different background model sensitivities by user defined regions and background filters | |
| CN107273726B (en) | Equipment owner's identity real-time identification method and its device based on acceleration cycle variation law | |
| WO2015133159A1 (en) | Image processing device, image processing method, and image processing program | |
| CN106778481A (en) | A kind of body heath's monitoring method | |
| CN110461215A (en) | Health mark is determined using portable device | |
| JP2015106197A (en) | Image processing device and image processing method | |
| CN111000402B (en) | Temperature adjusting method, storage medium and intelligent mattress | |
| CN108319986A (en) | The identification method and its system of image sources based on PRNU | |
| CN110443107A (en) | Image procossing for object detection | |
| CN113807289A (en) | Human body posture detection method and device, electronic equipment and storage medium | |
| CN108875519A (en) | Method for checking object, device and system and storage medium | |
| JP2017517916A (en) | Image collection method and apparatus | |
| CN206849035U (en) | A kind of image processing apparatus and face identification system | |
| CN114638980A (en) | Dish type identification processing method and device | |
| JP2010203845A5 (en) | ||
| CN107436622A (en) | A kind of electronic installation and night-light control method with night-light function | |
| JPWO2021166062A5 (en) | GATE DEVICE, GATE DEVICE CONTROL METHOD AND PROGRAM |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180105 |