CN104581939A - Queuing behavior detection method and system based on multiple heterogeneous sensors - Google Patents
Queuing behavior detection method and system based on multiple heterogeneous sensors Download PDFInfo
- Publication number
- CN104581939A CN104581939A CN201510001905.5A CN201510001905A CN104581939A CN 104581939 A CN104581939 A CN 104581939A CN 201510001905 A CN201510001905 A CN 201510001905A CN 104581939 A CN104581939 A CN 104581939A
- Authority
- CN
- China
- Prior art keywords
- queuing
- mobile phone
- smart mobile
- state
- motion state
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title abstract description 9
- 230000033001 locomotion Effects 0.000 claims abstract description 98
- 238000006073 displacement reaction Methods 0.000 claims abstract description 73
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 41
- 238000000034 method Methods 0.000 claims abstract description 37
- 230000006399 behavior Effects 0.000 claims abstract description 35
- 238000010801 machine learning Methods 0.000 claims abstract description 21
- 238000005259 measurement Methods 0.000 claims abstract description 19
- 230000008859 change Effects 0.000 claims description 59
- 230000001133 acceleration Effects 0.000 claims description 29
- 230000003542 behavioural effect Effects 0.000 claims description 19
- 238000000605 extraction Methods 0.000 claims description 14
- 230000009466 transformation Effects 0.000 claims description 13
- 238000003066 decision tree Methods 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 description 12
- 238000004364 calculation method Methods 0.000 description 11
- 101001041673 Homo sapiens Peroxisomal 2,4-dienoyl-CoA reductase [(3E)-enoyl-CoA-producing] Proteins 0.000 description 8
- 102100021404 Peroxisomal 2,4-dienoyl-CoA reductase [(3E)-enoyl-CoA-producing] Human genes 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 7
- 238000012706 support-vector machine Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 230000008447 perception Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000014509 gene expression Effects 0.000 description 3
- 230000002452 interceptive effect Effects 0.000 description 3
- 239000002699 waste material Substances 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000013468 resource allocation Methods 0.000 description 2
- 238000011524 similarity measure Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/025—Services making use of location information using location based information parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/025—Services making use of location information using location based information parameters
- H04W4/026—Services making use of location information using location based information parameters using orientation information, e.g. compass
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/025—Services making use of location information using location based information parameters
- H04W4/027—Services making use of location information using location based information parameters using movement velocity, acceleration information
Abstract
The invention relates to a queuing behavior detection method and system based on multiple heterogeneous sensors. The method comprises the following steps that original measurement data of a smart phone are collected by using the multiple heterogeneous sensors internally installed in the smart phone; according to the collected original measurement data, the motion state, the direction state, the displacement state and the relative distance of a user using the smart phone are predicted; according to the predicted motion state, the predicted direction state, the predicted displacement state and the predicted relative distance, queuing characteristic values which the user carries out queuing behavior are extracted; a queuing model is built according to the queuing characteristic values and a machine learning algorithm; the current queuing characteristic values of the smart phone are collected by using the multiple heterogeneous sensors internally installed in the smart phone, the current queuing characteristic values are input in the queuing model, and whether the current phone is in a queuing state or not is judged according to an output result. The queuing behavior detection method and system based on the multiple heterogeneous sensors can automatically detect the queuing behavior of the user and have robustness.
Description
Technical field
The present invention relates to the method for multiple heterogeneous sensor data fusion, particularly a kind of queuing behavioral value method and system based on multiple heterogeneous sensor.
Background technology
Along with fast developments such as embedded device, radio sensing network, mobile computing, the pervasive intelligent system of integrated perception, calculating and communication capacity is widely deployed, and is progressively dissolved into the everyday environments of the mankind.Wherein the most significantly, smart mobile phone universal, reached a new milestone in global smart mobile phone market, till 2014, shipment is up to billions of shipment amounts, and nearly the people of 63% have one or multi-section smart mobile phone.And the smart mobile phone in modern times, possess many built-in sensing equipments, wireless communication interface and high performance processing power.Three-dimensional accelerometer, electronic steering wheel, gyroscope, photosensitive sensors, nearfield sensor, microphone, pre-post is made a video recording first-class sensing equipment, can the perception anywhere or anytime of perception smart mobile phone with the information of surrounding environment detected, be the omnipresent perception of general fit calculation and serving.Built-in various wireless communication technology, WiFi, Bluetooth (bluetooth), Cellular (3G, 2G, 4G), smart mobile phone can carry out radio communication anywhere or anytime, serves for the immanent communication of general fit calculation.High performance calculating and processing power, 4 core 1.2Ghz, 8 core 1.5Ghz chips constantly update, and smart mobile phone more and more can process complicated calculations and provide high-quality service.Particularly, general fit calculation technology is being embed point, its sensing technology and context aware technology with smart mobile phone, strengthens us collect, analyze and utilize the range of data and dark in unprecedented mode.
Be embed point with smart mobile phone, people live in is merged in the environment of the mixing general fit calculation formed mutually by Sensor Network, communication network, internet etc., for people provide immanent service, and supports the social activities of the mankind.By the help of smart mobile phone to people, our daily life quality can be improved, as reduced traffic congestion, restriction transmission and optimizing public resource scheduling.In general fit calculation, the detection of queuing behavior and providing of queue waiting time, can improve the Consumer's Experience of people when queuing up, and also improve the efficiency of the resource allocation scheduling of businessman, avoid wasting of resources etc. situation.
Queuing phenomena is that people live very common phenomenon often, and especially in public domain, such as supermarket checkout stations, doorway, restaurant is waited for and being entered, and bank service window queuing waits for, transport hub, public place of entertainment etc. are to be serviced, park play facility etc.These public places, provide public service, and people are in order to enjoy this service, and behavior of must ranking, could accept public service according to order.Especially in the big city of some population concentrations, such as Beijing, Shanghai, the phenomenon waited in line is common especially.But long queue waiting time, can reduce the Consumer's Experience of people greatly.If people take a large amount of time in queuing behavior, the reduction of the dejected of people's mood and Consumer's Experience can be caused.Such as 2010 Shanghai World's Fair, each turnover guild hall, people often need the wait queue waiting time of 1 hour to 3 hours, and this is a thing very gloomy.Simultaneously, for business businessman, by the time waste of a large amount of users in queuing, instead of is placed on all the other commodity of purchase the time of user, buy itself and service is above everything else, this is a kind of waste of resource and the unreasonable of resource distribution.The more important thing is, a large number of users is in queuing, and business businessman needs to employ manpower to go to maintain queuing order, can not cause bad effect.Therefore, no matter wait in line, be for Consumer's Experience, or for business businessman, be all a thing needing to consider emphatically.If people are before queuing, be aware of the number waited in line and the time waited in line in advance, so people reasonably can utilize the time of oneself, and can not go queuing blindly.Business businessman is aware of the queue number of people in advance, and so they can reasonable efficient resource allocation, provides more service, avoids the waste of resource.Therefore, the service that queue waiting time provides, be one common and easily by service that people ignores.General fit calculation, provide immanent service, queuing time provides, and general fit calculation provides immanent service just.
Nowadays the development of smart mobile phone, integrated communicaton, perception and calculate in one, in immanent general fit calculation service, play more and more important important.Three-dimensional accelerometer, can detect people's behavior state, runs, and walks, or has a rest.Electronic steering wheel, the current residing direction of people can be detected, GPS can detect geographical location information (the Sensing meets mobile social networks:the design of people, implementation and evaluation of the cencemeapplication, " in SenSys 2008).We propose, and utilize smart mobile phone automatically to detect the queuing behavior of people, thus provide auxiliary for queuing time prognoses system.
Existing queuing behavioral value, is mainly divided into three major types.The first kind, adopts camera video to detect queuing behavior, thus is prediction queue waiting time raising subsidiary function.But this method, has following shortcoming.First, video surveillance queuing behavior, needs to utilize image processing techniques and machine learning algorithm, thus finds the model that can detect queuing behavior.This class model often needs to be optimized according to specific queuing scene, thus improves Detection accuracy.A general video queuing detection model, accuracy rate is very low.Setting up video queuing behavioral value model for all scenes, is very unpractical situation.Secondly, video surveillance queuing behavior, need to arrange enough cameras, cover the length of queuing troop, this needs extra facility to arrange, needs the man power and material of at substantial.
Equations of The Second Kind, is utilize electronics automatic call distributor, detects the monitoring of the behavior of waiting in line.Each user, before needs service, before first coming electronic queuing machine, gets a queuing sequence number ticket.Ticket identifies, the number of queuing up above, when machine is polled to user by the time, user is going to information desk.But this method, has following defect.First, electronic queuing machine needs extra-pay to go to buy.Secondly, queuing sequence number is inaccurate, thus queue waiting time accurately cannot be provided to estimate.User leaves because of certain reason after can having taken queuing sequence number ticket.Queuing sequence number is not often the number truly waited for above.
3rd class is the method utilizing manpower, monitoring queuing troop, thus by the number of troop and the stand-by period of pre-estimation, broadcast is gone out.This traditional method, needs to spend a large amount of manpower supports.
The present invention, utilizes the sensor that smart mobile phone is built-in, three-dimensional accelerometer, electronic steering wheel, bluetooth, WiFi, collect individual information walking, static, direction, distance, the behavioural information of the cellphone subscribers such as displacement, in conjunction with machine learning algorithm, support vector machine (Supported Vector Machine), thus judge whether active user is in queuing behavior, as the subsidiary function of queue waiting time prediction.
Summary of the invention
Technical matters to be solved by this invention is to provide a kind of method and system of the detection queuing behavior based on multiple heterogeneous sensor of queuing behavior of carrying out data fusion based on the multiple heterogeneous sensor in smart mobile phone and then automatically can detect user.
The technical scheme that the present invention solves the problems of the technologies described above is as follows: a kind of queuing behavioral value method based on multiple heterogeneous sensor, comprises the following steps:
Step 1: the multiple heterogeneous sensor utilizing smart mobile phone built-in gathers the raw measurement data of smart mobile phone respectively;
Step 2: according to the raw measurement data gathered, predicts the motion state of the user using smart mobile phone, direction state, displacement state and relative distance;
Step 3: according to the motion state of prediction, direction state, displacement state and relative distance, extracts user and to rank the queuing eigenwert of behavior;
Step 4: set up queuing model according to queuing eigenwert and machine learning algorithm;
Step 5: the multiple heterogeneous sensor utilizing smart mobile phone built-in gathers the current queuing eigenwert of smart mobile phone, by current queuing eigenwert input queue model, judges whether current phone is in queueing condition according to Output rusults.
The invention has the beneficial effects as follows: (1) manually inputs currently whether be in queuing without the need to user, and smart mobile phone detects the queuing behavior of user automatically; (2) without the need to any additional facilities, compare and three class queuing detection methods above, based on video, based on electronic queuing machine, based on manually, the present invention only requires the user of queuing, this prerequisite of carrying mobile phone of going out; (3) the present invention can be adapted to most of queuing scene, possesses robustness.
On the basis of technique scheme, the present invention can also do following improvement.
Further, described step 2 is specially:
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize the motion state of decision tree optimization algorithm predicts user;
Directional information residing for current according to the smart mobile phone gathered and the angular velocity information of relative displacement, utilize integral formula prediction direction state;
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize Wavelet Transformation Algorithm predictive displacement state;
According to the signal strength information of bluetooth equipment around the smart mobile phone gathered, multinomial fitting algorithm is utilized to predict relative distance.
Further, described queuing eigenwert comprises motion state similarity, direction state similarity, motion state rate of change, direction status variation rate, position skew, relative distance rate of change and alignment distance.
Further, described step 3 is specially,
Calculating formula of similarity is utilized to calculate motion state similarity according to the motion state of prediction;
Direction state according to prediction utilizes calculating formula of similarity calculated direction state similarity;
Rate of change computing formula is utilized to calculate motion state rate of change according to motion state similarity;
Rate of change computing formula calculated direction status variation rate is utilized according to direction state similarity;
Side-play amount computing formula displacement calculating side-play amount is utilized according to displacement state;
Rate of change computing formula is utilized to calculate relative distance rate of change according to relative distance;
Alignment distance is calculated with utilizing dynamic time adjustment algorithm according to motion state, direction state, displacement state and relative distance.
Further, a kind of queuing behavioral value system based on multiple heterogeneous sensor, comprises acquisition module, prediction module, extraction module, sets up module and input judge module;
Described acquisition module, the multiple heterogeneous sensor for utilizing smart mobile phone built-in gathers the raw measurement data of smart mobile phone respectively;
Described prediction module, for according to the raw measurement data gathered, predicts the motion state of the user using smart mobile phone, direction state, displacement state and relative distance;
Described extraction module, for the motion state according to prediction, direction state, displacement state and relative distance, extracts user and to rank the queuing eigenwert of behavior;
Describedly set up module, for setting up queuing model according to queuing eigenwert and machine learning algorithm;
Described input judge module, the multiple heterogeneous sensor for utilizing smart mobile phone built-in gathers the current queuing eigenwert of smart mobile phone, by current queuing eigenwert input queue model, judges whether current phone is in queueing condition according to Output rusults.
Further, the raw measurement data that comprises of described acquisition module: the signal strength information of bluetooth equipment around the smart mobile phone that the angular velocity information of the smart mobile phone relative displacement that the current residing directional information of the smart mobile phone that the acceleration information of smart mobile phone three coordinate axis that three-dimensional accelerometer gathers, electronic steering wheel gather, gyroscope gather and bluetooth gather.
Further, described prediction module specifically comprises:
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize the motion state of decision tree optimization algorithm predicts user;
Directional information residing for current according to the smart mobile phone gathered and the angular velocity information of relative displacement, utilize integral formula prediction direction state;
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize Wavelet Transformation Algorithm predictive displacement state;
Multinomial fitting algorithm is utilized to predict relative distance according to the signal strength information of bluetooth equipment around the smart mobile phone gathered.
Further, described extraction module, comprises queuing eigenwert: motion state similarity, direction state similarity, motion state rate of change, direction status variation rate, position skew, relative distance rate of change and alignment distance.
Further, described extraction module specifically comprises:
Calculating formula of similarity is utilized to calculate motion state similarity according to the motion state of prediction;
Direction state according to prediction utilizes calculating formula of similarity calculated direction state similarity;
Rate of change computing formula is utilized to calculate motion state rate of change according to motion state similarity;
Rate of change computing formula calculated direction status variation rate is utilized according to direction state similarity;
Side-play amount computing formula displacement calculating side-play amount is utilized according to displacement state;
Rate of change computing formula is utilized to calculate relative distance rate of change according to relative distance;
Alignment distance is calculated with utilizing dynamic time adjustment algorithm according to motion state, direction state, displacement state and relative distance.
Accompanying drawing explanation
Fig. 1 is the inventive method process flow diagram;
Fig. 2 is present system structural drawing;
Fig. 3 is that the present invention is through wavelet transformation, the step number schematic diagram utilizing same number crest calculating people walking;
Fig. 4 is the fitting of a polynomial schematic diagram of Distance geometry signal intensity of the present invention;
Fig. 5 is present system data processing schematic diagram.
In accompanying drawing, the list of parts representated by each label is as follows:
1, acquisition module, 2, prediction module, 3, extraction module, 4, set up module, 5, input judge module.
Embodiment
Be described principle of the present invention and feature below in conjunction with accompanying drawing, example, only for explaining the present invention, is not intended to limit scope of the present invention.
Embodiment 1
Based on a queuing behavioral value method for multiple heterogeneous sensor, comprise the following steps:
Step 1: the multiple heterogeneous sensor utilizing smart mobile phone built-in gathers the raw measurement data of smart mobile phone respectively;
Step 2: according to the raw measurement data gathered, predicts the motion state of the user using smart mobile phone, direction state, displacement state and relative distance;
Step 3: according to the motion state of prediction, direction state, displacement state and relative distance, extracts user and to rank the queuing eigenwert of behavior;
Step 4: set up queuing model according to queuing eigenwert and machine learning algorithm;
Step 5: the multiple heterogeneous sensor utilizing smart mobile phone built-in gathers the current queuing eigenwert of smart mobile phone, by current queuing eigenwert input queue model, judges whether current phone is in queueing condition according to Output rusults.
The signal strength information of bluetooth equipment around the smart mobile phone that the angular velocity information of the smart mobile phone relative displacement that the current residing directional information of smart mobile phone that described raw measurement data comprises the acceleration information of smart mobile phone three coordinate axis that three-dimensional accelerometer gathers, electronic steering wheel gathers, gyroscope gather and bluetooth gather.
Described step 2 is specially:
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize the motion state of decision tree optimization algorithm predicts user;
Directional information residing for current according to the smart mobile phone gathered and the angular velocity information of relative displacement, utilize integral formula prediction direction state;
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize Wavelet Transformation Algorithm predictive displacement state;
According to the signal strength information of bluetooth equipment around the smart mobile phone gathered, multinomial fitting algorithm is utilized to predict relative distance.
Described queuing eigenwert comprises motion state similarity, direction state similarity, motion state rate of change, direction status variation rate, position skew, relative distance rate of change and alignment distance.
Described step 3 is specially,
Calculating formula of similarity is utilized to calculate motion state similarity according to the motion state of prediction;
Direction state according to prediction utilizes calculating formula of similarity calculated direction state similarity;
Rate of change computing formula is utilized to calculate motion state rate of change according to motion state similarity;
Rate of change computing formula calculated direction status variation rate is utilized according to direction state similarity;
Side-play amount computing formula displacement calculating side-play amount is utilized according to displacement state;
Rate of change computing formula is utilized to calculate relative distance rate of change according to relative distance;
Alignment distance is calculated with utilizing dynamic time adjustment algorithm according to motion state, direction state, displacement state and relative distance.
Based on a queuing behavioral value system for multiple heterogeneous sensor, comprise acquisition module 1, prediction module 2, extraction module 3, set up module 4 and input judge module 5;
Described acquisition module 1, the multiple heterogeneous sensor for utilizing smart mobile phone built-in gathers the raw measurement data of smart mobile phone respectively;
Described prediction module 2, for according to the raw measurement data gathered, predicts the motion state of the user using smart mobile phone, direction state, displacement state and relative distance;
Described extraction module 3, for the motion state according to prediction, direction state, displacement state and relative distance, extracts user and to rank the queuing eigenwert of behavior;
Describedly set up module 4, for setting up queuing model according to queuing eigenwert and machine learning algorithm;
Described input judge module 5, the multiple heterogeneous sensor for utilizing smart mobile phone built-in gathers the current queuing eigenwert of smart mobile phone, by current queuing eigenwert input queue model, judges whether current phone is in queueing condition according to Output rusults.
The raw measurement data that described acquisition module 1 comprises: the signal strength information of bluetooth equipment around the smart mobile phone that the angular velocity information of the smart mobile phone relative displacement that the current residing directional information of the smart mobile phone that the acceleration information of smart mobile phone three coordinate axis that three-dimensional accelerometer gathers, electronic steering wheel gather, gyroscope gather and bluetooth gather.
Described prediction module 2 specifically comprises:
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize the motion state of decision tree optimization algorithm predicts user;
Directional information residing for current according to the smart mobile phone gathered and the angular velocity information of relative displacement, utilize integral formula prediction direction state;
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize Wavelet Transformation Algorithm predictive displacement state;
Multinomial fitting algorithm is utilized to predict relative distance according to the signal strength information of bluetooth equipment around the smart mobile phone gathered.
Described extraction module 3, comprises queuing eigenwert: motion state similarity, direction state similarity, motion state rate of change, direction status variation rate, position skew, relative distance rate of change and alignment distance.
Described extraction module 3 specifically comprises:
Calculating formula of similarity is utilized to calculate motion state similarity according to the motion state of prediction;
Direction state according to prediction utilizes calculating formula of similarity calculated direction state similarity;
Rate of change computing formula is utilized to calculate motion state rate of change according to motion state similarity;
Rate of change computing formula calculated direction status variation rate is utilized according to direction state similarity;
Side-play amount computing formula displacement calculating side-play amount is utilized according to displacement state;
Rate of change computing formula is utilized to calculate relative distance rate of change according to relative distance;
Alignment distance is calculated with utilizing dynamic time adjustment algorithm according to motion state, direction state, displacement state and relative distance.
In concrete enforcement, the present invention proposes a kind of method that multiple heterogeneous sensor based on smart mobile phone detects queuing behavior automatically.Detection method mainly comprises four-stage.
First, utilize the sensor that modern smart mobile phone is built-in, three-dimensional accelerometer collects the acceleration information of smart mobile phone three coordinate axis, electronic steering wheel collects the current residing direction of smart mobile phone, the angular velocity of smart mobile phone relative displacement collected by gyroscope, and bluetooth collects the signal intensity of the bluetooth equipment around smart mobile phone.Secondly, utilize fast fourier transform, by seasonal effect in time series three-dimensional accelerometer data, rate of change data, in conjunction with the decision Tree algorithms of machine learning, judge the motion state of current phone user; Utilize statistical method (mean and variance), statistics electronic steering wheel and gyrostatic data, the direction of estimation current phone user; The motion state of the user obtained before utilization and direction, calculate the displacement of active user; Utilize the polynomial regression technology of machine learning, Bluetooth signal intensity is fitted to relative distance data, thus estimate the distance between mobile phone and mobile phone.Thirdly, we according to previous step say obtain, motion state, direction, displacement information, range information, recycle these data, the eigenwert of synthesis queuing model, comprise 7 eigenwerts: motion state similarity (Movement Similarity, hereafter write a Chinese character in simplified form into MS), direction state similarity (Direction Similarity, hereafter write a Chinese character in simplified form into DS), motion state rate of change (Movement Crossing Rate, hereafter write a Chinese character in simplified form into MCR), direction status variation rate (Direction Crossing Rate, hereafter write a Chinese character in simplified form into DCR), displacement bias (DisplacementDeviation, hereafter write a Chinese character in simplified form into DD), relative distance rate of change (Peer-Distance CrossingRate, hereafter write a Chinese character in simplified form into PDCR), alignment distance (Alignment Distance, hereafter write a Chinese character in simplified form into AD).Finally, by the support vector machine (SVM) of machine learning, by (MS, DS, MCR, DCR, DD, PDCR, AD × 3) as proper vector, obtain queuing model.
On smart mobile phone, realize the method that multiple heterogeneous sensor detects queuing behavior automatically, its key step comprises:
1. smart mobile phone built-in sensors gathers raw data, comprises three-dimensional accelerometer, electronic steering wheel, gyroscope and bluetooth.
2. utilize signal transacting, machine learning techniques process raw data, infer the motion state of user, direction, displacement and distance;
3. the motion state of the user utilizing step 2 to obtain, direction, displacement and distance, extract the eigenwert of queuing behavior, mainly comprise the eigenwert of three aspects, similarity eigenwert, rate of change feature and time series eigenwert, specifically comprise 7 clocks, motion state similarity (MS), direction state similarity (DS), motion state rate of change (MCR), direction status variation rate (DCR), displacement bias (DD), relative distance rate of change (PDCR), alignment distance (AD).
4. the eigenwert of integrating step 3 and the support vector machines of machine learning, set up queuing model, automatically detects current phone and whether be in queuing behavior.
Wherein, 1. smart mobile phone built-in sensors (three-dimensional accelerometer, electronic steering wheel, gyroscope and bluetooth) gathers raw data, is specially:
Three-dimensional accelerometer gathers raw data:
The three-dimensional accelerometer of modern smart mobile phone, with the frequency collection mobile phone acceleration information of 20Hz-200Hz.With the acceleration of the frequency collection three-dimensional accelerometer of 50Hz, each acceleration information has three value (x, y, z), wherein x represents the X-axis acceleration of mobile phone, and y represents the Y-axis acceleration of mobile phone, and z represents the Z axis acceleration of mobile phone, calculate the phase amplitude of the acceleration of mobile phone, utilize formula 1:
Abs (writing a Chinese character in simplified form of Absolute mean absolute average), as the acceleration absolute average of mobile phone X-axis, Y-axis and Z axis, can describe the amplitude of the change of current phone acceleration.The employing frequency of the three-dimensional accelerometer of mobile phone is 50Hz, we adopt 64 abs data, and (requirement of fast fourier transform input data scale must be the n power of 2, in order to the employing frequency of 50Hz can be covered, we adopt the input data scale of 6 powers of 2) as a Frame, be placed in a window, each window obtains the output of 64 frequency values, as the input of step 2 through fast fourier transform (FFT package).
Electronic steering wheel gathers raw data:
The electronic steering wheel of modern smart mobile phone, with the frequency collection mobile phone acceleration information of 20Hz-200Hz.We adopt the data of the frequency collection electronic steering wheel of 50Hz, and in 3-D walls and floor (x, y, z), wherein y value represents the direction of current phone, x and z Value Data interfering data.(calculated rate of mobile phone motion state is needed to be consistent with 64 Frames data, ability step 2 calculates phase eigenwert in the same time) as a window, be placed on inside a Frame, ask draw value μ wherein and standard deviation sigma, as the input of step 2.
Gyroscope gathers raw data:
The gyroscope of modern smart mobile phone is uncertain with its frequency acquisition.We gather gyrostatic data as quickly as possible, and each gyrostatic data represents, smart mobile phone in x-axis, the angular velocity of y-axis and z-axis.By three-dimensional array, as a window, to be placed on inside a Frame, as the input of step 2 with 64 Frames (need the calculated rate in mobile phone motion state and direction to be consistent, ability step 2 calculates phase eigenwert in the same time).
Bluetooth gathers raw data:
The bluetooth of modern smart mobile phone, we go to collect bluetooth around and signal intensity with the frequency of 1Hz, store blue-teeth data (ID with data to form, RSSI), wherein ID identifies the identifier of bluetooth, and RSSI represents the signal intensity of current Bluetooth ID, as the input of step 2.
Wherein, 2. utilize the raw data that step 1 gathers, in conjunction with machine learning techniques, infer the motion state of user, direction, displacement and distance, be specially:
The acquisition of motion state:
In step 1, we have obtained through Fourier Rapid Variable Design for each frame data (64 abs data), the eigenwert of 64 different frequencies of acquisition, using these 64 eigenwerts as input.Utilize the J.48 algorithm inside machine learning algorithm, be J.48 wherein classical decision tree optimization algorithm, be input as the proper vector of 64 dimensions, export as classification results.Its classification results is people's all possible motion states when queuing up, and comprising: walking (walk), stops (still), and advance at a quick pace (running) etc.
The acquisition of direction state:
The electronic steering wheel Frame (data of 64 y-axis) that step 1 obtains, average μ and standard deviation sigma.And the gyro data that step 1 obtains, (x, y, z) each value represents the angular velocity of corresponding coordinate axle.We are in conjunction with gyro data and electronic steering wheel data, infer the direction state of current phone user.Specific practice: the directional data first obtained by electronic steering wheel is as initial directional data, and gyroscope (x, y, z) is relative acceleration, adopts integral formula 2, obtains the angle changed in time [t1, t2] interior direction:
In the three-dimensional (x, y, z) of the gyroscope employing of mobile phone, wherein y value represents the direction of current phone, x and z Value Data interfering data, therefore only adopts Y-axis data to carry out changing the calculating of orientation angle Δ angle.Secondly, utilize Δ angle to correct the data error of electronic steering wheel.The angle that the electronic steering wheel of correcting method: previous moment t1 is measured is α 1, the angle that subsequent time electronic steering wheel is measured is α 2, calculate the difference (Δ=α 1-α 2) of two numerical value, if Δ and Δ angle inconsistent, so utilize Δ angle to correct, otherwise do not make and change.
The acquisition of displacement:
According to the Frame (64 abs data) of each window that step 1 obtains, utilize wavelet transformation smoothed data frame.Wavelet transformation is in classical signals process, utilizes the segments of filters comprising finite impulse response filter, is showed clearly by the signal meeting this wave band.The displacement of people, be realized by walking (walking), people's is the wave band with timing dependence on foot, and therefore, the wave band of people's walking can describe out by wavelet transformation clearly.The present invention calculates people's step number, pass through wavelet transformation, (64 abs data) of each window are filtered out interfering data (wave band of non-sequential correlativity), remaining window data (abs data) is exactly the time sequence window of walking describing people, then crest number is calculated, computing method: the abs data on a time point, what be both greater than the left side (early than this time point) abs data is also greater than the right (being later than this time point), so this time point is calculated as a crest value, if the size of this crest exceeded 1.7g (concrete data we see Medical Devices meter step standard, so step number adds 1, otherwise it is constant, wherein g represents 9.8m/s
2acceleration.As shown in Figure 3.
The acquisition of distance:
In step 1, say the blue-teeth data two tuple (ID, RSSI) obtained, calculate current distance.Computing method are as follows, utilize historical data to carry out multinomial matching.The present invention, acquires 70 diverse locations (distance), a pair smart mobile phone, records their range information and corresponding signal intensity respectively, as shown in Figure 4:
Adopt the fitting of a polynomial algorithm in machine learning, obtain the mapping relations of Distance geometry signal intensity, described by formula:
d＝a
3*r
3+a
2*r
2+a
1*r
1(2)
Wherein r represents writing a Chinese character in simplified form of signal intensity RSSI, and d represents distance, wherein (a
3, a
2, a
1) be respectively the three cubed parameter of signal intensity r, the parameter of signal intensity r quadratic power, the parameter (this is the fixed form that fitting of a polynomial adopts) of signal intensity r, obtains the concrete numerical value of these three parameters by the multinomial approximating method of machine learning.The calculating of later distance, collects (ID, RSSI) data according to bluetooth, by formula 2, can obtain that to calculate with current device number be the distance of ID.
3. the characteristics extraction of queuing behavior:
The feature of queuing behavior mainly comprises three aspects, similarity eigenwert, rate of change feature and time series eigenwert, specifically, enumerate 7 kinds: motion state similarity (Movement Similarity, hereafter write a Chinese character in simplified form into MS), direction state similarity (Direction Similarity, hereafter write a Chinese character in simplified form into DS), motion state rate of change (Movement Crossing Rate, hereafter write a Chinese character in simplified form into MCR), direction status variation rate (Direction Crossing Rate, hereafter write a Chinese character in simplified form into DCR), displacement bias (Displacement Deviation, hereafter write a Chinese character in simplified form into DD), relative distance rate of change (Peer-Distance Crossing Rate, hereafter write a Chinese character in simplified form into PDCR), alignment distance (AlignmentDistance, hereafter write a Chinese character in simplified form into AD).For each eigenwert, will be introduced and how to calculate below.
Motion state similarity (MS):
Motion state Similarity Measure, it is described that two individual sports states similarity degree within a certain period of time.People are in queuing process, and people or wait for together, or when troop's movement, people also can move together, therefore queuing behavior can have higher similarity.Calculating formula of similarity 3:
State number wherein in n certain hour.Wherein function f (a, b)
iwhether consistent represent in the motion state of moment i, a and b, wherein a and b represents and refers to concrete people's (a or b of appearance hereafter also represents the same meaning), if consistent f (a, b)
iget 1, otherwise f (a, b)
iget 0, wherein motion state is obtained by step 2.The MS that formula 2 calculates, illustrate two people a and b within a period of time, whether their motion state is consistent.MS (a, b) numeral is higher, illustrates that the motion state of a with b two people is more similar.
Direction state similarity (DS):
Direction Similarity Measure, it is described that two people direction similarity degrees within a certain period of time.People are in queuing process, and the direction most of the time of people is all consistent, and therefore queuing behavior can have higher direction similarity.Calculating formula of similarity 4:
State number wherein in n certain hour.Wherein function g (a, b)
irepresent whether consistent in the direction state of moment i, a and b, if consistent g (a, b)
iget 1, otherwise g (a, b)
iget 0.Whether direction is consistent, adopts formula 5 to judge:
Wherein avg represents average, and std represents standard deviation.
represent the direction value of a when moment i, it obtains in step 2.If formula 5 is set up, so represent that the direction of two people is consistent, function g (a, b)
iequal 1.On the basis of formula 5, the DS of the formula 4 of calculating, illustrates two people a and b within a period of time, and whether their direction of motion is consistent, and its numerical value is higher, and both explanations are more similar.
Motion state rate of change (MCR):
Motion state rate of change calculates, and it is described that two individual sports states difference degree within a certain period of time.People are in queuing process, if troop moves forward, coming outrunner can first move, and the people come below moves after the meeting, and therefore queuing behavior has regular hour diversity factor.The present invention utilizes formula 6 to calculate:
The variable implication of formula 6 is consistent with formula 3 variable implication.The MCR that formula 6 calculates, illustrate two people a and b motion state similarity intensity of variation, its numerical value is higher, and both explanations motion state similarity is unstable, and numerical value is lower, and both expressions motion state similarity is very stable.
Direction status variation rate (DCR):
Direction rate of change calculates, and it is described that two people direction states difference degree within a certain period of time.People are in queuing process, if troop moves forward, coming outrunner can first move, and the people come below moves after the meeting, and therefore queuing behavior has regular hour diversity factor.The present invention utilizes formula 7 to calculate:
Variable implication and formula 4, the 5 variable implication of formula 7 are consistent.The DCR that formula 7 calculates, illustrate two people a and b direction state similarity intensity of variation, its numerical value is higher, and both explanations direction state similarity is unstable, and numerical value is lower, and both expressions direction state similarity is very stable.
Displacement bias (DD):
Change in displacement rate calculates, and it is described that two people's displacement states difference degree within a certain period of time.People are in queuing process, if troop moves forward, coming outrunner can first move, and the people come below moves after the meeting, and therefore in queuing behavior, displacement has regular hour diversity factor.The present invention utilizes formula 8 to calculate:
Wherein f (a, b) '
irepresent that whether the displacement of a and b is consistent, the displacement of a and b obtains in step 2, and wherein i represents the concrete moment.If the displacement of a and b is equal, so f (a, b) '
iget 1 otherwise get 0.Remaining variable implication is consistent with formula 3 variable implication.The DD that formula 8 calculates, illustrate two people a and b displacement state similarity, its numerical value is higher, and both explanations are more similar.
Relative distance rate of change (PDCR):
Distance state rate of change calculates, and it is described that two personal distance's states difference degree within a certain period of time.People, in queuing process, have fixing distance between men, and this distance has 2 factors to determine, the position of current queuing, the distance between adjacent people, have left troop unless there are people, otherwise in queuing process, the status variation rate of distance keeps very low value.The present invention utilizes formula 9 to calculate:
Wherein d (a, b)
irepresent the distance of a and b between moment i, it is obtained by step 2.The PDCR that formula 9 calculates, illustrate the intensity of variation of two people a and relative distance, its numerical value is higher, and the conversion of both explanations relative distance is faster, more unstable, and numerical value is lower, and the relative distance of both expressions is very stable.
Alignment distance (AD):
Temporal aspect value calculates, and the state that it describes two people has certain sequential relationship.People, in queuing process, are motion states, direction or displacement, distance all have certain sequential relationship, the present invention adopt minimum to its distance to describe the sequential relationship between people.Algorithm Dynamic TimeWarping (DTW) can calculate within the unit interval, sequential relationship between the two, returned a value determined.
4. the eigenwert of integrating step 3 and the support vector machines of machine learning, set up queuing model.
In step 3, we obtain 7 kinds of eigenwerts, and wherein temporal aspect value Alignment Distance (AD), can comprise the sequential relationship of motion state, the sequential relationship in direction, the sequential relationship of displacement.Therefore we have characteristic vector space, and 9 dimensions (MS, DS, MCR, DCR, DD, PDCR, AD × 3), in conjunction with the algorithm of support vector machine of machine learning, set up queuing model.Wherein support vector machine, is classical machine learning algorithm, by vector space, can divides different classifications, and find a gap (border), make different classes of between border maximize.
Below in conjunction with Fig. 5, workflow of the present invention is described:
A) three-dimensional accelerometer collects the raw data of the X, Y, Z axis of mobile phone;
B) after (a), wavelet transformation and decision tree judgement are carried out to data;
C) the relative angle speed of the X, Y, Z axis of mobile phone collected by gyroscope;
D) after (c), integration is continued to data, obtain the angular transformation in certain hour;
E) gyroscope collect mobile phone X, Y, Z axis direction;
F) after (e), its mean and variance is asked;
G) bluetooth collects the signal intensity of bluetooth equipment around;
H) after (g), utilize regression model, obtain the distance between surrounding devices;
I) after (b), the motion state of current phone is obtained;
J) after (d) (f), the direction of current phone is obtained;
K) after (j) (b), the displacement of current phone is obtained;
L) after (h), the distance of current phone is obtained;
M) after (i) (j) (k) (l), the queuing eigenwert of current phone is obtained;
After (m), judge whether current phone is in queueing condition.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1., based on a queuing behavioral value method for multiple heterogeneous sensor, it is characterized in that, comprise the following steps:
Step 1: the multiple heterogeneous sensor utilizing smart mobile phone built-in gathers the raw measurement data of smart mobile phone respectively;
Step 2: according to the raw measurement data gathered, predicts the motion state of the user using smart mobile phone, direction state, displacement state and relative distance;
Step 3: according to the motion state of prediction, direction state, displacement state and relative distance, extracts user and to rank the queuing eigenwert of behavior;
Step 4: set up queuing model according to queuing eigenwert and machine learning algorithm;
Step 5: the multiple heterogeneous sensor utilizing smart mobile phone built-in gathers the current queuing eigenwert of smart mobile phone, by current queuing eigenwert input queue model, judges whether current phone is in queueing condition according to Output rusults.
2. the queuing behavioral value method based on multiple heterogeneous sensor according to claim 1, it is characterized in that, described raw measurement data comprises the acceleration information of smart mobile phone three coordinate axis that three-dimensional accelerometer gathers, the signal strength information of bluetooth equipment around smart mobile phone that the angular velocity information of smart mobile phone relative displacement that the current residing directional information of smart mobile phone that electronic steering wheel gathers, gyroscope gather and bluetooth gather.
3. the queuing behavioral value method based on multiple heterogeneous sensor according to claim 2, it is characterized in that, described step 2 is specially:
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize the motion state of decision tree optimization algorithm predicts user;
Directional information residing for current according to the smart mobile phone gathered and the angular velocity information of relative displacement, utilize integral formula prediction direction state;
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize Wavelet Transformation Algorithm predictive displacement state;
According to the signal strength information of bluetooth equipment around the smart mobile phone gathered, multinomial fitting algorithm is utilized to predict relative distance.
4. according to the arbitrary described queuing behavioral value method based on multiple heterogeneous sensor of claim 2 to 3, it is characterized in that, described queuing eigenwert comprises motion state similarity, direction state similarity, motion state rate of change, direction status variation rate, position skew, relative distance rate of change and alignment distance.
5. the queuing behavioral value method based on multiple heterogeneous sensor according to claim 4, it is characterized in that, described step 3 is specially,
Calculating formula of similarity is utilized to calculate motion state similarity according to the motion state of prediction;
Direction state according to prediction utilizes calculating formula of similarity calculated direction state similarity;
Rate of change computing formula is utilized to calculate motion state rate of change according to motion state similarity;
Rate of change computing formula calculated direction status variation rate is utilized according to direction state similarity;
Side-play amount computing formula displacement calculating side-play amount is utilized according to displacement state;
Rate of change computing formula is utilized to calculate relative distance rate of change according to relative distance;
Alignment distance is calculated with utilizing dynamic time adjustment algorithm according to motion state, direction state, displacement state and relative distance.
6. the queuing behavioral value system based on multiple heterogeneous sensor, it is characterized in that, comprise acquisition module (1), prediction module (2), extraction module (3), set up module (4) and input judge module (5);
Described acquisition module (1), the multiple heterogeneous sensor for utilizing smart mobile phone built-in gathers the raw measurement data of smart mobile phone respectively;
Described prediction module (2), for according to the raw measurement data gathered, predicts the motion state of the user using smart mobile phone, direction state, displacement state and relative distance;
Described extraction module (3), for the motion state according to prediction, direction state, displacement state and relative distance, extracts user and to rank the queuing eigenwert of behavior;
Describedly set up module (4), for setting up queuing model according to queuing eigenwert and machine learning algorithm;
Described input judge module (5), multiple heterogeneous sensor for utilizing smart mobile phone built-in gathers the current queuing eigenwert of smart mobile phone, by in current queuing eigenwert input queue model, judge whether current phone is in queueing condition according to Output rusults.
7. the queuing behavioral value system based on multiple heterogeneous sensor according to claim 6, it is characterized in that, the raw measurement data that described acquisition module (1) comprises: the signal strength information of bluetooth equipment around the smart mobile phone that the angular velocity information of the smart mobile phone relative displacement that the current residing directional information of the smart mobile phone that the acceleration information of smart mobile phone three coordinate axis that three-dimensional accelerometer gathers, electronic steering wheel gather, gyroscope gather and bluetooth gather.
8. the queuing behavioral value system based on multiple heterogeneous sensor according to claim 7, it is characterized in that, described prediction module (2) specifically comprises:
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize the motion state of decision tree optimization algorithm predicts user;
Directional information residing for current according to the smart mobile phone gathered and the angular velocity information of relative displacement, utilize integral formula prediction direction state;
According to the acceleration information of smart mobile phone three coordinate axis gathered, utilize Wavelet Transformation Algorithm predictive displacement state;
Multinomial fitting algorithm is utilized to predict relative distance according to the signal strength information of bluetooth equipment around the smart mobile phone gathered.
9. according to the arbitrary described queuing behavioral value system based on multiple heterogeneous sensor of claim 7 to 8, it is characterized in that, described extraction module (3), comprises queuing eigenwert: motion state similarity, direction state similarity, motion state rate of change, direction status variation rate, position skew, relative distance rate of change and alignment distance.
10. the queuing behavioral value system based on multiple heterogeneous sensor according to claim 9, it is characterized in that, described extraction module (3) specifically comprises:
Calculating formula of similarity is utilized to calculate motion state similarity according to the motion state of prediction;
Direction state according to prediction utilizes calculating formula of similarity calculated direction state similarity;
Rate of change computing formula is utilized to calculate motion state rate of change according to motion state similarity;
Rate of change computing formula calculated direction status variation rate is utilized according to direction state similarity;
Side-play amount computing formula displacement calculating side-play amount is utilized according to displacement state;
Rate of change computing formula is utilized to calculate relative distance rate of change according to relative distance;
Alignment distance is calculated with utilizing dynamic time adjustment algorithm according to motion state, direction state, displacement state and relative distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510001905.5A CN104581939B (en) | 2015-01-04 | 2015-01-04 | A kind of queuing behavioral value method and system based on a variety of heterogeneous sensors |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510001905.5A CN104581939B (en) | 2015-01-04 | 2015-01-04 | A kind of queuing behavioral value method and system based on a variety of heterogeneous sensors |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104581939A true CN104581939A (en) | 2015-04-29 |
| CN104581939B CN104581939B (en) | 2018-03-30 |
Family
ID=53096876
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510001905.5A Expired - Fee Related CN104581939B (en) | 2015-01-04 | 2015-01-04 | A kind of queuing behavioral value method and system based on a variety of heterogeneous sensors |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104581939B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104994360A (en) * | 2015-08-03 | 2015-10-21 | 北京旷视科技有限公司 | Video monitoring method and video monitoring system |
| CN106230849A (en) * | 2016-08-22 | 2016-12-14 | 中国科学院信息工程研究所 | A kind of smart machine machine learning safety monitoring system based on user behavior |
| CN109711678A (en) * | 2018-12-07 | 2019-05-03 | 江南机电设计研究所 | A kind of heterogeneous sensor intelligent task planing method based on machine learning |
| CN111127733A (en) * | 2019-10-29 | 2020-05-08 | 杭州智策略科技有限公司 | Mobile crowd sensing-based canteen queuing time detection system and method |
| CN112949350A (en) * | 2019-12-10 | 2021-06-11 | 晶睿通讯股份有限公司 | Queue analysis method and image monitoring equipment thereof |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009067256A2 (en) * | 2007-11-25 | 2009-05-28 | Trilliant Networks, Inc. | System and method for power outage and restoration notification in an advanced metering infrastructure network |
| CN102772211A (en) * | 2012-08-08 | 2012-11-14 | 中山大学 | Human movement state detection system and detection method |
| CN104091053A (en) * | 2014-06-26 | 2014-10-08 | 李南君 | Method and equipment for automatically detecting behavior pattern |
-
2015
- 2015-01-04 CN CN201510001905.5A patent/CN104581939B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009067256A2 (en) * | 2007-11-25 | 2009-05-28 | Trilliant Networks, Inc. | System and method for power outage and restoration notification in an advanced metering infrastructure network |
| CN102772211A (en) * | 2012-08-08 | 2012-11-14 | 中山大学 | Human movement state detection system and detection method |
| CN104091053A (en) * | 2014-06-26 | 2014-10-08 | 李南君 | Method and equipment for automatically detecting behavior pattern |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104994360A (en) * | 2015-08-03 | 2015-10-21 | 北京旷视科技有限公司 | Video monitoring method and video monitoring system |
| CN104994360B (en) * | 2015-08-03 | 2018-10-26 | 北京旷视科技有限公司 | Video frequency monitoring method and video monitoring system |
| CN106230849A (en) * | 2016-08-22 | 2016-12-14 | 中国科学院信息工程研究所 | A kind of smart machine machine learning safety monitoring system based on user behavior |
| CN106230849B (en) * | 2016-08-22 | 2019-04-19 | 中国科学院信息工程研究所 | A kind of smart machine machine learning safety monitoring system based on user behavior |
| CN109711678A (en) * | 2018-12-07 | 2019-05-03 | 江南机电设计研究所 | A kind of heterogeneous sensor intelligent task planing method based on machine learning |
| CN109711678B (en) * | 2018-12-07 | 2021-02-12 | 江南机电设计研究所 | Heterogeneous sensor intelligent task planning method based on machine learning |
| CN111127733A (en) * | 2019-10-29 | 2020-05-08 | 杭州智策略科技有限公司 | Mobile crowd sensing-based canteen queuing time detection system and method |
| CN112949350A (en) * | 2019-12-10 | 2021-06-11 | 晶睿通讯股份有限公司 | Queue analysis method and image monitoring equipment thereof |
| CN112949350B (en) * | 2019-12-10 | 2024-04-09 | 晶睿通讯股份有限公司 | Queue analysis method and image monitoring device thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104581939B (en) | 2018-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104581939A (en) | Queuing behavior detection method and system based on multiple heterogeneous sensors | |
| JP6698092B2 (en) | Machine learning with model filtering and model mixing for edge devices in heterogeneous environments | |
| Yoon et al. | Cityride: a predictive bike sharing journey advisor | |
| JP6225257B2 (en) | Interest point clustering method and related apparatus | |
| Do et al. | A probabilistic kernel method for human mobility prediction with smartphones | |
| CN104798420B (en) | Based on the recognizable radio signal source Estimated Time of Arrival for encountering surrounding along route | |
| JP5536485B2 (en) | Portable terminal, server, program, and method for estimating address / location as user moves | |
| CN102893589B (en) | Method and apparatus for providing context sensing and fusion | |
| CN103026780B (en) | For controlling the method and apparatus called of sensor | |
| CN107526090A (en) | Location positioning method, equipment and system, computer-readable recording medium | |
| Zhang et al. | Research on precision marketing model of tourism industry based on user’s mobile behavior trajectory | |
| CN109945855A (en) | The acquisition of fine grained location data | |
| CN105243148A (en) | Checkin data based spatial-temporal trajectory similarity measurement method and system | |
| CN110313190A (en) | Control device and method | |
| US20200019365A1 (en) | Location prediction systems and related methods | |
| CN108734502A (en) | A kind of data statistical approach and system based on user location | |
| US11343636B2 (en) | Automatic building detection and classification using elevator/escalator stairs modeling—smart cities | |
| Huo et al. | Short-term estimation and prediction of pedestrian density in urban hot spots based on mobile phone data | |
| JP6089764B2 (en) | Activity status processing apparatus and activity status processing method | |
| CN104904244A (en) | Mobile device for distinguishing user's movement, method therefore, and method for generating hierarchical tree model therefor | |
| Qureshi et al. | Integration of OMNI channels and machine learning with smart technologies | |
| CN111372188B (en) | Method and device for determining hot spot track in area, storage medium and electronic device | |
| WO2013024276A1 (en) | Optimised context-awareness on mobile devices | |
| KR20210078203A (en) | Method for profiling based on foothold and terminal using the same | |
| CN107750339B (en) | Detecting a context of a user using a mobile device based on wireless signal characteristics |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180330 |