CN103248955B - Personal identification method and device based on intelligent remote control system - Google Patents
Personal identification method and device based on intelligent remote control system Download PDFInfo
- Publication number
- CN103248955B CN103248955B CN201310140986.8A CN201310140986A CN103248955B CN 103248955 B CN103248955 B CN 103248955B CN 201310140986 A CN201310140986 A CN 201310140986A CN 103248955 B CN103248955 B CN 103248955B
- Authority
- CN
- China
- Prior art keywords
- similarity
- user behavior
- identity
- behavior data
- data
- 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.)
- Active
Links
Abstract
The invention discloses a kind of personal identification method and device based on intelligent remote control system, its method includes step：Receive at least one user behavior data that sensor is obtained；The information gain rate of each user behavior data is calculated, and each user behavior data is sorted out according to the information gain rate of each user behavior data using decision tree, same category of user behavior data is constituted a Sub Data Set；Each Sub Data Set is modeled using Gauss model and/or SVM models, identification result is obtained；The identification result is shown, for user's calibration validation.Present invention employs sensor identification technology, Decision Tree Technologies, modeling and identification technology, realize that user identity is accurately recognized, effectively increase the accuracy of user identity in remote control system.
Description
Technical field
The present invention relates to TV technology, personal identification method and dress based on intelligent remote control system are related specifically to
Put.
Background technology
At present, the service that intelligent remote control system is provided is increasingly dependent on its interacting with user, such as server and intelligence
Can television terminal program push given to the user etc. for having demand by user mutual, therefore realize system and user interact by
Gradually turn into basic function indispensable in intelligent remote control system.But intelligent remote control system of the prior art all can not basis
The specific identity of user realizes interaction, and this undoubtedly seriously limits the service content that intelligent remote control system is provided.Such as one
Generally there are the piece identities such as grandfather, father, child, mother in family, but server and Intelligent television terminal but can not bases
These piece identities realize suitable program push, such as：The program that takes care of health is recommended according to grandfather's identity, electricity is recommended according to father's identity
Film festival mesh, juvenile's program is recommended according to child's identity, and shopping program etc. is recommended according to mother's identity.Even if present some remote controls
Sensor technology is applied to the user recognition technology of television system by system, still, the identification that these technologies can only be rough
The general identity of user, it is impossible to accomplish accurate identification, the situation for frequently resulting in identification mistake occurs.
The content of the invention
The main object of the present invention is known to provide a kind of identity based on intelligent remote control system for accurately identifying user identity
Other method and apparatus.
The present invention proposes a kind of personal identification method based on intelligent remote control system, including step：
Step A, receives at least one user behavior data that sensor is obtained；
Step B, calculates the information gain rate of each user behavior data, and using decision tree according to each described user
The information gain rate of behavioral data is sorted out to each user behavior data, same category of user behavior data is constituted one
Sub Data Set；
Step C, using Gauss model and/or SVM（Support Vector Machine, SVMs）Model is to each
Individual Sub Data Set modeling, obtains identification result；
Step D, shows the identification result, for user's calibration validation.
Preferably, the step B is specifically included：
Determine the comentropy of each user behavior data；
According to described information entropy, the ratio of profit increase of each user behavior data is determined；
Decision tree is set up according to the ratio of profit increase of each user behavior data；
User behavior data of the ratio of profit increase in the range of same predetermined gain ratio step is categorized in using decision tree same
In individual Sub Data Set.
Preferably, the step C is specifically included：
Each Sub Data Set is modeled using Gauss model and/or SVM models, the similarity of each identity is obtained respectively；
The similarity for each identity that sorts, regard similarity highest identity as identification result.
Preferably, in the step C, when being modeled using Gauss model and SVM models to each Sub Data Set, the row
The similarity of each identity of sequence, is specifically included the step of using similarity highest identity as identification result：
The similarity for each identity obtained by Gauss model that sorts respectively and the phase of each identity obtained by SVM models
Like degree；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity obtained by SVM models
When S2 is identical, directly the similarity highest identity is regard as identification result；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity obtained by SVM models
When S2 is different, compare S1 and S2 similarity；
When S1 is identical with S2 similarity, while regarding S1 and S2 as identification result；
When S1 is different from S2 similarity, the higher identity of similarity in S1 and S2 is regard as identification result.
Preferably, also include after the step A：
The user behavior data belonged in preset reference range is extracted in the user behavior data that the sensor is obtained,
And using the user behavior data belonged in preset reference range as being used in step B calculate information gain rate and confession
The data that decision tree is sorted out.
The present invention also proposes a kind of identity recognition device based on intelligent remote control system, including：
Receiving module, at least one user behavior data for receiving sensor acquisition；
Sort module, the information gain rate for calculating each user behavior data, and using decision tree according to described
The information gain rate of each user behavior data is sorted out to each user behavior data, makes same category of user behavior data
Constitute a Sub Data Set；
Identification module, for being modeled using Gauss model and/or SVM models to each Sub Data Set, obtains identification
As a result；
Display module, for showing the identification result, for user's calibration validation.
Preferably, the sort module specifically for：
Determine the comentropy of each user behavior data；
According to described information entropy, the ratio of profit increase of each user behavior data is determined；
Decision tree is set up according to the ratio of profit increase of each user behavior data；
User behavior data of the ratio of profit increase in the range of same predetermined gain ratio step is categorized in using decision tree same
In individual Sub Data Set.
Preferably, the identification module specifically for：
Each Sub Data Set is modeled using Gauss model and/or SVM models, the similarity of each identity is obtained respectively；
The similarity for each identity that sorts, regard similarity highest identity as identification result.
Preferably, when the identification module is modeled using Gauss model and SVM models to each Sub Data Set, the knowledge
Other module specifically for：
The similarity for each identity obtained by Gauss model that sorts respectively and the phase of each identity obtained by SVM models
Like degree；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity obtained by SVM models
When S2 is identical, directly the similarity highest identity is regard as identification result；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity obtained by SVM models
When S2 is different, compare S1 and S2 similarity；
When S1 is identical with S2 similarity, while regarding S1 and S2 as identification result；
When S1 is different from S2 similarity, the higher identity of similarity in S1 and S2 is regard as identification result.
Preferably, the identity recognition device based on intelligent remote control system also includes screening module, specifically in institute
The user behavior data for extracting and belonging in preset reference range in the user behavior data of sensor acquisition is stated, and is sent to described
Sort module.
Present invention employs sensor identification technology, Decision Tree Technologies, modeling and identification technology, user identity is realized
Accurate identification, effectively increases the accuracy of user identity in remote control system.
Brief description of the drawings
Fig. 1 is the flow chart of the first embodiment of the personal identification method of the invention based on intelligent remote control system；
Fig. 2 is the flow chart of the second embodiment of the personal identification method of the invention based on intelligent remote control system；
Fig. 3 is the flow chart of the 3rd embodiment of the personal identification method of the invention based on intelligent remote control system；
Fig. 4 is the flow chart of the fourth embodiment of the personal identification method of the invention based on intelligent remote control system；
Fig. 5 is the flow chart of the 5th embodiment of the personal identification method of the invention based on intelligent remote control system；
Fig. 6 is the structural representation of the first embodiment of the identity recognition device of the invention based on intelligent remote control system；
Fig. 7 is the structural representation of the second embodiment of the identity recognition device of the invention based on intelligent remote control system.
The realization, functional characteristics and advantage of the object of the invention will be described further referring to the drawings in conjunction with the embodiments.
Embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
As shown in figure 1, Fig. 1 is the flow of the first embodiment of the personal identification method of the invention based on intelligent remote control system
Figure.The personal identification method based on intelligent remote control system that the present embodiment is mentioned, including：
Step A, receives at least one user behavior data that sensor is obtained；
In the present embodiment, due to the behavioural habits of each user be it is different, such as user pick up remote control speed,
Remote control has difference in terms of the angle of remote control systems receiver, and the number of the behavior act of user is obtained using sensor
According to recognizing the identity of user according to the concrete behavior data of user.The sensor that the present embodiment is used can pass for acceleration
Sensor, direction sensor, gyro sensor etc., acceleration transducer can obtain acceleration when user picks up or put down remote control
Degree can obtain user and pick up remote control deflection angle, and gyro sensor can obtain angle of rotation speed when user tilts remote control
Degree.For example, father picks up the acceleration magnitude of remote control：X=5, y=20, z=40, direction tilt value：X=0, y=45 degree, z=90 degree；
Grandfather picks up the acceleration magnitude of remote control：X=1, y=4, z=10, direction tilt value：x=0、y=0、z=90.Obtain and use in sensor
Before the behavioral data of family, first each sensor function can be configured, for example, the sensor receive capabilities for needing to use are opened, screen
The sensor receive capabilities that need not be used are covered, can be directly by configuring biography when the sensor of a certain classification need not be used
The mode of sensor masks it in the software end of television system, the hardware configuration without changing remote control.Concrete configuration mode can
With reference to following table：
| Data function | Function is selected（1- open functions, 0- closing functions） |
| Gyroscope X-axis data | 1 |
| Gyroscope Y-axis data | 1 |
| Gyroscope Z axis data | 1 |
| Direction X-axis data | 0 |
| Direction Y-axis data | 0 |
| Direction Z axis data | 0 |
| Acceleration X-axis data | 1 |
| Acceleration Y-axis data | 1 |
| Acceleration Z axis data | 1 |
According to the configuration of sensor function, user behavior data is obtained using the sensor for having turned on function.
Step B, calculates the information gain rate of each user behavior data, and using decision tree according to each user behavior
The information gain rate of data is sorted out to each user behavior data, same category of user behavior data is constituted a subnumber
According to collection；
The present embodiment employs mining algorithm and analyzes and sort out user behavior to subsequently set up Model Identification user identity
Data, are to select attribute for user behavior data, obtain the optimal parameter for creating mining model.The present embodiment can be used
Decision tree and information gain rate realize sorted users behavioral data in mining algorithm, and repetition is removed using the ratio of profit increase of comentropy
Similar data, different data are taken out, and then do classification action by decision tree, i.e., set up decision-making according to the ratio of profit increase of entropy
Tree, finds out data of the ratio of profit increase in same scope using decision tree from each user behavior data and forms a sub data
Collection, finally according to multiple different gains rate scopes, forms multiple Sub Data Sets, and Gauss model or SVM models are utilized for subsequently middle
It is identified.
Step C, is modeled to each Sub Data Set using Gauss model and/or SVM models, obtains identification result；
According to above-mentioned classification results, select each Sub Data Set Gauss model or selection SVM models to be modeled, go forward side by side
Row identification.The user identity that above-mentioned identification can be obtained is all as identification result, and output to display module is shown,
Or only using similarity highest user identity as identification result, output to display module is shown.
Step D, shows identification result, for user's calibration validation.
Display module is shown after identification result on a display screen, can carry out calibration validation by user, if display
Recognition result only one of which, and the recognition result is correct, then user confirms that the recognition result is effective；If display
Recognition result has multiple, then user needs to select correct identity；If the recognition result of display is incorrect, by user voluntarily
The correct identity of typing.After user's calibration validation, remote control system can store this user behavior data and the model of use
Into the data file of corresponding identity subordinate, as reference data when being easy to recognize next time.
The present embodiment employs sensor identification technology, Decision Tree Technologies, modeling and identification technology, realizes user's body
The accurate identification of part, effectively increases the accuracy of user identity in remote control system.
As shown in Fig. 2 Fig. 2 is the flow of the second embodiment of the personal identification method of the invention based on intelligent remote control system
Figure.Also include after step A：
Step E, extracts the user's row belonged in preset reference range in the user behavior data that the sensor is obtained
For data.
Belong to after above-mentioned extraction the user behavior data in preset reference range as in step B be used for calculate comentropy increase
Beneficial rate and the data sorted out for decision tree.In the user behavior data that sensor is obtained, in order to avoid some Outliers
Significant error is caused to whole identification process, it is necessary to extract correct reliable data to the user behavior data of acquisition, these are just
Really reliable data can be such as following table in the behavioral data term of reference of kinsfolk：
| Identity | Acceleration X-axis term of reference | Acceleration Y-axis term of reference | Acceleration Z axis term of reference |
| Grandfather | 0-30 | 0-30 | 0-30 |
| Grandmother | 0-25 | 0-25 | 0-25 |
| Father | 35-75 | 35-75 | 35-75 |
| Mother | 35-75 | 35-75 | 35-75 |
| Child | 80-100 | 80-100 | 80-100 |
The behavioral data term of reference of all kinsfolks is 0-100 i.e. in this family, should be taken out in this term of reference
Take authentic data.Above-mentioned term of reference is a relative value relative to normal data, it would however also be possible to employ specific tape unit
Data are as value range, for example, grandfather's acceleration X-axis term of reference is 0-20m/s2。
Because the present embodiment is using the operations such as correct reliable data progress follow-up classification, identification are extracted, pole can be avoided
The misleading that end data is caused to whole identification process, makes recognition result more accurate, further increases user in remote control system
The accuracy of identity.
As shown in figure 3, Fig. 3 is the flow of the 3rd embodiment of the personal identification method of the invention based on intelligent remote control system
Figure.Step B is specifically included：
Step B01, determines the comentropy of each user behavior data；
Assuming that data set S is the data set of an acceleration transducer, Ci is some user action data, freq
(Ci, S) is exactly the number of times that Ci occurs in S, | S | it is the total number of user action data in data set S, then S comentropy can
Be expressed as Info (S)=- Σ [(freq (Ci, S)/| S |] log2[freq (Ci, S)/| S |)].Simultaneously, it is assumed that number
It is the data set of the X-axis of acceleration transducer according to collection T, then Info may be calculated according to the comentropy after X branchesx(T) =
Σ [(|Ti|/ |T|) · Info(Ti)]。
Step B02, according to comentropy, determines the ratio of profit increase of each user behavior data；
It is Gain (X)=Info (S)-Info to calculate information gain ratex(T) each branch, is made all to maximize gain
Rate.
Step B03, decision tree is set up according to the ratio of profit increase of each user behavior data；
Decision tree is set up according to the size of ratio of profit increase, user behavior data ratio of profit increase is bigger, and the user behavior data is more leaned on
Nearly decision-making root vertex.
User behavior data of the ratio of profit increase in the range of same predetermined gain ratio step, is categorized in by step B04 using decision tree
In same Sub Data Set.
The present embodiment realizes that user behavior data is classified using comentropy and decision tree, using sorted Sub Data Set as
Follow-up modeling parameters, make recognition result more accurate, further increase the accuracy of user identity in remote control system.
As shown in figure 4, Fig. 4 is the flow of the fourth embodiment of the personal identification method of the invention based on intelligent remote control system
Figure.Step C is specifically included：
Step C01, is modeled to each Sub Data Set using Gauss model and/or SVM models, the phase of each identity is obtained respectively
Like degree；
The present embodiment configures any model and above-mentioned Sub Data Set is identified, and can be configured in remote control system,
Or selected according to default setting, or use two kinds of models to be identified simultaneously.Obtained according to Gauss model or selection SVM models
Obtain identity similarity.For example, after being modeled using Gauss model to each Sub Data Set, obtaining following similarity：Grandfather's similarity
50%th, grandmother's similarity 40%, father's similarity 10%；After being modeled using SVM models to each Sub Data Set, obtain following similar
Degree：Grandfather's similarity 70%, grandmother's similarity 30%.
Step C02, the similarity for each identity that sorts, regard similarity highest identity as identification result.
Above-mentioned similarity is ranked up, similarity highest identity is the result of this identification.For example, in above-mentioned reality
Apply in example, after Gauss model and SVM Model Identifications, similarity highest identity is all grandfather, then " grandfather " is used as into body
The result of part identification is shown on a display screen, is calibrated for user.The present embodiment screens recognition result by sequencing of similarity, uses
Family identification is more accurate.
As shown in figure 5, Fig. 5 is the flow of the 5th embodiment of the personal identification method of the invention based on intelligent remote control system
Figure.In step C01, when being modeled using Gauss model and SVM models to each Sub Data Set, step C02 is specifically included：
Step C021, the similarity for each identity obtained by Gauss model that sorts respectively and by SVM models obtain it is each
The similarity of identity；
Step C022, judge the similarity highest identity S1 obtained by Gauss model to by SVM models obtain it is similar
Whether identical spend highest identity S2, that is, judge whether S1=S2=S；If it is, performing step C023；If it is not, then performing
Step C024；
Step C023, directly regard similarity highest identity S as identification result；
The present embodiment after being modeled using Gauss model to each Sub Data Set, obtains following similar by taking following data as an example
Degree：Grandfather's similarity 50%, grandmother's similarity 40%, father's similarity 10%；After being modeled using SVM models to each Sub Data Set,
Obtain following similarity：In grandfather's similarity 70%, grandmother's similarity 30%, the recognition result that two models are obtained, " grandfather " row
Name all highests, " grandfather " is shown on a display screen as the result of identification, calibrated for user.
Step C024, compares S1 and S2 similarity；
Step C025, when S1 is identical with S2 similarity, while regarding S1 and S2 as identification result；
By taking following data as an example, use the highest similarity identity that Gauss model is obtained for " grandfather ", similarity 50% is adopted
The highest similarity identity obtained with SVM models is " father ", similarity 50%, while " grandfather " and " father " is shown in aobvious
In display screen, calibrated for user.
Step C026, when S1 similarity is more than S2 similarity, regard S1 as identification result；
Step C027, when S1 similarity is less than S2 similarity, regard S2 as identification result.
By taking following data as an example, use the highest similarity identity that Gauss model is obtained for " grandfather ", similarity 50% is adopted
The highest similarity identity obtained with SVM models is " father ", similarity 70%, by " father " display on a display screen, for user
Calibration.
The present embodiment is identified simultaneously using two kinds of models, it is to avoid the inaccurate feelings of recognition result that single model is caused
Condition, effectively increases identification accuracy.
As shown in fig. 6, Fig. 6 is the structure of the first embodiment of the identity recognition device of the invention based on intelligent remote control system
Schematic diagram.The identity recognition device based on intelligent remote control system that the present embodiment is mentioned, including：
Receiving module 10, at least one user behavior data for receiving sensor acquisition；
Sort module 20, the information gain rate for calculating each user behavior data, and using decision tree according to institute
The information gain rate for stating each user behavior data is sorted out to each user behavior data, makes same category of user behavior number
According to one Sub Data Set of composition；
Identification module 30, for being modeled using Gauss model and/or SVM models to each Sub Data Set, is obtained identity and known
Other result；
Display module 40, for showing identification result, for user's calibration validation.
In the present embodiment, due to the behavioural habits of each user be it is different, such as user pick up remote control speed,
Remote control has difference in terms of the angle of remote control systems receiver, and the number of the behavior act of user is obtained using sensor
According to recognizing the identity of user according to the concrete behavior data of user.The sensor that the present embodiment is used can pass for acceleration
Sensor, direction sensor, gyro sensor etc., acceleration transducer can obtain acceleration when user picks up or put down remote control
Degree can obtain user and pick up remote control deflection angle, and gyro sensor can obtain angle of rotation speed when user tilts remote control
Degree.For example, father picks up the acceleration magnitude of remote control：X=5, y=20, z=40, direction tilt value：X=0, y=45 degree, z=90 degree；
Grandfather picks up the acceleration magnitude of remote control：X=1, y=4, z=10, direction tilt value：x=0、y=0、z=90.Obtain and use in sensor
Before the behavioral data of family, first each sensor function can be configured, for example, the sensor receive capabilities for needing to use are opened, screen
The sensor receive capabilities that need not be used are covered, can be directly by configuring biography when the sensor of a certain classification need not be used
The mode of sensor masks it in the software end of television system, the hardware configuration without changing remote control.Concrete configuration mode can
With reference to following table：
| Data function | Function is selected（1- open functions, 0- closing functions） |
| Gyroscope X-axis data | 1 |
| Gyroscope Y-axis data | 1 |
| Gyroscope Z axis data | 1 |
| Direction X-axis data | 0 |
| Direction Y-axis data | 0 |
| Direction Z axis data | 0 |
| Acceleration X-axis data | 1 |
| Acceleration Y-axis data | 1 |
| Acceleration Z axis data | 1 |
According to the configuration of sensor function, user behavior data is obtained using the sensor for having turned on function.
The present embodiment employs mining algorithm and analyzes and sort out user behavior to subsequently set up Model Identification user identity
Data, are to select attribute for user behavior data, obtain the optimal parameter for creating mining model.The present embodiment can be used
Decision tree and information gain rate realize sorted users behavioral data in mining algorithm, and repetition is removed using the ratio of profit increase of comentropy
Similar data, different data are taken out, and then do classification action by decision tree, i.e., set up decision-making according to the ratio of profit increase of entropy
Tree, finds out data of the ratio of profit increase in same scope using decision tree from each user behavior data and forms a sub data
Collection, finally according to multiple different gains rate scopes, forms multiple Sub Data Sets, and Gauss model or SVM models are utilized for subsequently middle
It is identified.According to above-mentioned classification results, Gauss model or selection SVM models is selected each Sub Data Set to be modeled, and
Carry out identification.The user identity that above-mentioned identification can be obtained all is used as identification result, output to display module 40
It has been shown that, or only using similarity highest user identity as identification result, output to display module 40 is shown.
Display module 40 shows after identification result on a display screen, can by wanting user to carry out calibration validation, if
The recognition result only one of which of display, and the recognition result is correct, then and user confirms that the recognition result is effective；If aobvious
The recognition result shown has multiple, then user needs to select correct identity；If the recognition result of display is incorrect, by user
The voluntarily correct identity of typing.After user's calibration validation, remote control system can be by this user behavior data and the model of use
In the data file for storing corresponding identity subordinate, as reference data when being easy to recognize next time.
The present embodiment employs sensor identification technology, Decision Tree Technologies, modeling and identification technology, realizes user's body
The accurate identification of part, effectively increases the accuracy of user identity in remote control system.
In the embodiment of the present invention, sort module 20 specifically for：
Determine the comentropy of each user behavior data；
According to comentropy, the ratio of profit increase of each user behavior data is determined；
Decision tree is set up according to the ratio of profit increase of each user behavior data；
User behavior data of the ratio of profit increase in the range of same predetermined gain ratio step is categorized in by same height using decision tree
In data set.
In the present embodiment, it is assumed that data set S is the data set of an acceleration transducer, Ci is some user action number
According to, freq (Ci, S) is exactly the number of times that Ci occurs in S, | S | be the total number of user action data in data set S, then S
Comentropy can be expressed as Info (S)=- Σ [(freq (Ci, S)/| S |] log2[freq (Ci, S)/| S |)].Together
When, it is assumed that data set T is the data set of the X-axis of acceleration transducer, then be may be calculated according to the comentropy after X branches
Infox(T) = Σ [(|Ti|/ |T|)· Info(Ti)].Calculate information gain rate be Gain (X)=Info (S)-
Infox(T) each branch, is made all to maximize ratio of profit increase.Decision tree, user behavior data gain are set up according to the size of ratio of profit increase
Rate is bigger, and the user behavior data is closer to decision-making root vertex.The present embodiment realizes user's row using comentropy and decision tree
Classify for data, using sorted Sub Data Set as follow-up modeling parameters, make recognition result more accurate, further increase
The accuracy of user identity in remote control system.
In the embodiment of the present invention, identification module 30 specifically for：
Each Sub Data Set is modeled using Gauss model and/or SVM models, the similarity of each identity is obtained respectively；
The similarity for each identity that sorts, regard similarity highest identity as identification result.
The present embodiment configures any model and above-mentioned Sub Data Set is identified, and can be configured in remote control system,
Or selected according to default setting, or use two kinds of models to be identified simultaneously.Obtained according to Gauss model or selection SVM models
Obtain identity similarity.For example, after being modeled using Gauss model to each Sub Data Set, obtaining following similarity：Grandfather's similarity
50%th, grandmother's similarity 40%, father's similarity 10%；After being modeled using SVM models to each Sub Data Set, obtain following similar
Degree：Grandfather's similarity 70%, grandmother's similarity 30%.Above-mentioned similarity is ranked up, similarity highest identity is this
The result of identification.For example, in the above-described embodiments, after Gauss model and SVM Model Identifications, similarity highest identity is all
For grandfather, then " grandfather " is shown on a display screen as the result of identification, calibrated for user.The present embodiment passes through similar
Degree sequence screening recognition result, makes user's identification more accurate.
The embodiment of the present invention is known when identification module 30 is modeled using Gauss model and SVM models to each Sub Data Set
Other module 30 specifically for：
The similarity for each identity obtained by Gauss model that sorts respectively and the phase of each identity obtained by SVM models
Like degree；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity obtained by SVM models
When S2 is identical, directly similarity highest identity is regard as identification result；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity obtained by SVM models
When S2 is different, compare S1 and S2 similarity；
When S1 is identical with S2 similarity, while regarding S1 and S2 as identification result；
When S1 is different from S2 similarity, the higher identity of similarity in S1 and S2 is regard as identification result.
The present embodiment after being modeled using Gauss model to each Sub Data Set, obtains following similar by taking following data as an example
Degree：Grandfather's similarity 50%, grandmother's similarity 40%, father's similarity 10%；After being modeled using SVM models to each Sub Data Set,
Obtain following similarity：In grandfather's similarity 70%, grandmother's similarity 30%, the recognition result that two models are obtained, " grandfather " row
Name all highests, " grandfather " is shown on a display screen as the result of identification, calibrated for user.By taking following data as an example,
The highest similarity identity that Gauss model is obtained is used for " grandfather ", similarity 50%, the highest obtained using SVM models is similar
It is " father ", similarity 50%, while by " grandfather " and " father " display on a display screen, being calibrated for user to spend identity.With following
Data instance, uses the highest similarity identity that Gauss model is obtained for " grandfather ", similarity 50%, is obtained using SVM models
Highest similarity identity is " father ", and similarity 70% by " father " display on a display screen, is calibrated for user.The present embodiment is adopted
It is identified simultaneously with two kinds of models, it is to avoid the inaccurate situation of recognition result that single model is caused, effectively increases identification
Accuracy.
As shown in fig. 7, Fig. 7 is the structure of the second embodiment of the identity recognition device of the invention based on intelligent remote control system
Schematic diagram.The present embodiment adds screening module 50 on the basis of embodiment illustrated in fig. 6, specifically for：
Judge the user behavior data of sensor acquisition whether in preset reference range；
The user behavior data belonged in preset reference range is extracted in the user behavior data that the sensor is obtained,
And send to sort module 20.
The user behavior data belonged in the present embodiment, after extraction in preset reference range is as sort module 20 based on
Calculate information gain rate and the data sorted out for decision tree.In the user behavior data that sensor is obtained, in order to avoid certain
A little Outliers cause misguidance to whole identification process, it is necessary to extract correct reliable number to the user behavior data of acquisition
According to these correct reliable data should be such as following table in the behavioral data term of reference of kinsfolk：
| Identity | Acceleration X-axis term of reference | Acceleration Y-axis term of reference | Acceleration Z axis term of reference |
| Grandfather | 0-30 | 0-30 | 0-30 |
| Grandmother | 0-25 | 0-25 | 0-25 |
| Father | 35-75 | 35-75 | 35-75 |
| Mother | 35-75 | 35-75 | 35-75 |
| Child | 80-100 | 80-100 | 80-100 |
The behavioral data term of reference of all kinsfolks is 0-100 i.e. in this family, should be taken out in this term of reference
Take authentic data.Above-mentioned term of reference is a relative value relative to normal data, it would however also be possible to employ specific tape unit
Data are as value range, for example, grandfather's acceleration X-axis term of reference is 0-20m/s2。
Because the present embodiment is using the operations such as correct reliable data progress follow-up classification, identification are extracted, pole can be avoided
The misleading that end data is caused to whole identification process, makes recognition result more accurate, further increases user in remote control system
The accuracy of identity.
The preferred embodiments of the present invention are these are only, are not intended to limit the scope of the invention, it is every to utilize this hair
Equivalent structure or equivalent flow conversion that bright specification and accompanying drawing content are made, or directly or indirectly it is used in other related skills
Art field, is included within the scope of the present invention.
Claims (6)
1. a kind of personal identification method based on intelligent remote control system, it is characterised in that including step：
Step A, receives at least one user behavior data that sensor is obtained；
Step B, calculates the information gain rate of each user behavior data, and using decision tree according to each described user behavior
The information gain rate of data is sorted out to each user behavior data, same category of user behavior data is constituted a subnumber
According to collection；
Step C, is modeled to each Sub Data Set using Gauss model and/or SVM models, obtains identification result；
Step D, shows the identification result, for user's calibration validation；
Wherein, the step C is specifically included：
Each Sub Data Set is modeled using Gauss model and SVM models, the similarity of each identity is obtained respectively；
The similarity for each identity obtained by Gauss model that sorts respectively and the similarity of each identity obtained by SVM models；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity S2 phases that are obtained by SVM models
Meanwhile, directly it regard the similarity highest identity as identification result；
When the similarity highest identity S1 that is obtained by the Gauss model and similarity highest identity S2 that is obtained by SVM models not
Meanwhile, compare S1 and S2 similarity；
When S1 is identical with S2 similarity, while regarding S1 and S2 as identification result；
When S1 is different from S2 similarity, the higher identity of similarity in S1 and S2 is regard as identification result.
2. the personal identification method according to claim 1 based on intelligent remote control system, it is characterised in that the step B
Specifically include：
Determine the comentropy of each user behavior data；
According to described information entropy, the ratio of profit increase of each user behavior data is determined；
Decision tree is set up according to the ratio of profit increase of each user behavior data；
User behavior data of the ratio of profit increase in the range of same predetermined gain ratio step is categorized in by same height using decision tree
In data set.
3. the personal identification method according to claim 1 based on intelligent remote control system, it is characterised in that the step A
Also include afterwards：
The user behavior data belonged in preset reference range is extracted in the user behavior data that the sensor is obtained, and will
The user behavior data belonged in preset reference range in step B as being used to calculate information gain rate and supply decision-making
Set the data sorted out.
4. a kind of identity recognition device based on intelligent remote control system, it is characterised in that including：
Receiving module, at least one user behavior data for receiving sensor acquisition；
Sort module, the information gain rate for calculating each user behavior data, and using decision tree according to it is described each
The information gain rate of user behavior data is sorted out to each user behavior data, constitutes same category of user behavior data
One Sub Data Set；
Identification module, for being modeled using Gauss model and/or SVM models to each Sub Data Set, obtains identification result；
Display module, for showing the identification result, for user's calibration validation；
Wherein, the identification module specifically for：
Each Sub Data Set is modeled using Gauss model and SVM models, the similarity of each identity is obtained respectively；
The similarity for each identity obtained by Gauss model that sorts respectively and the similarity of each identity obtained by SVM models；
As the similarity highest identity S1 obtained by Gauss model and the similarity highest identity S2 phases that are obtained by SVM models
Meanwhile, directly it regard the similarity highest identity as identification result；
When the similarity highest identity S1 that is obtained by the Gauss model and similarity highest identity S2 that is obtained by SVM models not
Meanwhile, compare S1 and S2 similarity；
When S1 is identical with S2 similarity, while regarding S1 and S2 as identification result；
When S1 is different from S2 similarity, the higher identity of similarity in S1 and S2 is regard as identification result.
5. the identity recognition device according to claim 4 based on intelligent remote control system, it is characterised in that the classification mould
Block specifically for：
Determine the comentropy of each user behavior data；
According to described information entropy, the ratio of profit increase of each user behavior data is determined；
Decision tree is set up according to the ratio of profit increase of each user behavior data；
User behavior data of the ratio of profit increase in the range of same predetermined gain ratio step is categorized in by same height using decision tree
In data set.
6. the identity recognition device according to claim 4 based on intelligent remote control system, it is characterised in that also including screening
Module, specifically for extracting the user behavior belonged in preset reference range in the user behavior data that the sensor is obtained
Data, and send to the sort module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310140986.8A CN103248955B (en) | 2013-04-22 | 2013-04-22 | Personal identification method and device based on intelligent remote control system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310140986.8A CN103248955B (en) | 2013-04-22 | 2013-04-22 | Personal identification method and device based on intelligent remote control system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103248955A CN103248955A (en) | 2013-08-14 |
| CN103248955B true CN103248955B (en) | 2017-07-28 |
Family
ID=48928136
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310140986.8A Active CN103248955B (en) | 2013-04-22 | 2013-04-22 | Personal identification method and device based on intelligent remote control system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103248955B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103546789B (en) * | 2013-08-23 | 2017-07-11 | Tcl集团股份有限公司 | A kind of TV programme suggesting method, system and intelligent terminal |
| CN103543832B (en) * | 2013-10-29 | 2017-08-04 | Tcl集团股份有限公司 | A kind of user identification method and device based on unbalanced data |
| CN105357565B (en) * | 2014-08-22 | 2018-05-25 | 深圳Tcl新技术有限公司 | For the information-pushing method and system of smart television |
| CN106610991A (en) * | 2015-10-23 | 2017-05-03 | 北京国双科技有限公司 | Data processing method and device |
| CN109146574A (en) * | 2018-09-06 | 2019-01-04 | 深圳市木瓜移动科技有限公司 | Ad click cheating monitoring method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005106760A2 (en) * | 2003-10-29 | 2005-11-10 | Smiths Detection Inc. | Chemical and biological agent sensor array |
| CN102302370A (en) * | 2011-06-30 | 2012-01-04 | 中国科学院计算技术研究所 | Method and device for detecting tumbling |
| CN102368293A (en) * | 2011-09-07 | 2012-03-07 | 常州蓝城信息科技有限公司 | Personal identity authentication system based on stealth consciousness |
| CN103019376A (en) * | 2012-12-04 | 2013-04-03 | 深圳Tcl新技术有限公司 | Identity-recognition-based remote control function configuration method and system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102750286B (en) * | 2011-04-21 | 2016-01-20 | 常州蓝城信息科技有限公司 | A kind of Novel decision tree classifier method processing missing data |
| CN102831447B (en) * | 2012-08-30 | 2015-01-21 | 北京理工大学 | Method for identifying multi-class facial expressions at high precision |
-
2013
- 2013-04-22 CN CN201310140986.8A patent/CN103248955B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005106760A2 (en) * | 2003-10-29 | 2005-11-10 | Smiths Detection Inc. | Chemical and biological agent sensor array |
| CN102302370A (en) * | 2011-06-30 | 2012-01-04 | 中国科学院计算技术研究所 | Method and device for detecting tumbling |
| CN102368293A (en) * | 2011-09-07 | 2012-03-07 | 常州蓝城信息科技有限公司 | Personal identity authentication system based on stealth consciousness |
| CN103019376A (en) * | 2012-12-04 | 2013-04-03 | 深圳Tcl新技术有限公司 | Identity-recognition-based remote control function configuration method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103248955A (en) | 2013-08-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103248955B (en) | Personal identification method and device based on intelligent remote control system | |
| EP3398034B1 (en) | Electrical device for hand gestures detection | |
| Paul et al. | An effective approach for human activity recognition on smartphone | |
| CN104573706A (en) | Object identification method and system thereof | |
| CN108090499B (en) | Data active labeling method and system based on maximum information triple screening network | |
| CN102236984B (en) | Messaging device, questioning tendency setting method | |
| CN104239416A (en) | User identification method and system | |
| CN103902591B (en) | Construct the method and device of decision tree classifier | |
| CN104795067A (en) | Voice interaction method and device | |
| CN111582350B (en) | AdaBoost-based composite material damage identification method and system | |
| CN103559504A (en) | Image target category identification method and device | |
| CN103414930A (en) | Remote control system for identifying and sensing user and method thereof | |
| CN102129477B (en) | Multimode-combined image reordering method | |
| CN108108491A (en) | A kind of recommendation method and device of multi-medium data | |
| CN107179831B (en) | Start method, apparatus, storage medium and the terminal of application | |
| KR101872733B1 (en) | System for recommending social networking service following and method for recommending social networking service following using it | |
| CN107251048A (en) | Reliable finger tip and palm detection | |
| CN104680188A (en) | Method for constructing human body posture reference image library | |
| CN104991908A (en) | Method and device for recognition processing of target object | |
| CN108985533B (en) | Motion management system and method based on Android platform | |
| CN111382605A (en) | Video content auditing method and device, storage medium and computer equipment | |
| CN105898604A (en) | Live broadcast video interaction information configuration method and device based on mobile terminal | |
| CN106156167A (en) | The method of data collection/process, motion information acquisition equipment and system | |
| CN104765476B (en) | Handwriting tracks generation method and device | |
| CN112116981B (en) | Rehabilitation training method, device and system based on artificial intelligence and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |