CN106092093A - A kind of indoor orientation method based on earth magnetism fingerprint matching algorithm - Google Patents

Abstract

A kind of indoor orientation method based on earth magnetism fingerprint matching algorithm, including: step 1, gather the earth magnetism fingerprint time series of region, required location critical path；Step 2, is mapped to new sequence by the earth magnetism fingerprint time series collected and map reference and preserves to earth magnetism fingerprint database；Step 3, gathers earth magnetism subsequence to be matched, and filters out corresponding candidate matching sequence according to the variation tendency of this sequence；Step 4, calculates the dynamic time warping distance of earth magnetism subsequence to be matched and all candidate matches sequences, and is finally mapped to corresponding map reference.The present invention takes full advantage of earth's magnetic field local anomaly and feature steady in a long-term in building, it is achieved without network infrastructure, real-time height, the indoor orientation method of accurate positioning.

Description

A kind of indoor orientation method based on earth magnetism fingerprint matching algorithm

Technical field

The present invention relates to a kind of indoor orientation method, especially relate to a kind of indoor based on earth magnetism fingerprint matching algorithm fixed Method for position, belongs to indoor positioning technologies field.

Background technology

In recent years, along with the development of mobile network Yu intelligent terminal, the indoor positioning practicality in some specific occasions and Necessity increasingly significant.The such as intelligent shopping guide in megastore, supermarket；A meeting is found in strange office building View room；Parked vehicle etc. is found in huge parking lot.

Global positioning system (Global Positioning System, GPS) is widely used in outdoor positioning, but Due to indoor signal decay and the impact of multipath effect, this system can only operate in outdoor opening, it is impossible to realizes indoor Reliable location under complex environment.Indoor positioning navigation system based on WIFI fingerprint and bluetooth fingerprint has a wide range of applications, But they are all highly dependent upon expensive radio infrastructure (wireless aps and Bluetooth base. station), are fitted without wireless base for those The building of Infrastructure is just felt simply helpless.Meanwhile, both signals are all easily subject to the shadow of signal attenuation, human body interference etc. in indoor Ring.

Summary of the invention

The problem above of present invention prior art to be overcome, it is provided that a kind of indoor positioning based on earth magnetism fingerprint matching algorithm Method, the inventive method takes full advantage of earth's magnetic field local anomaly and feature steady in a long-term in building, it is achieved without network Infrastructure, real-time are high, the indoor orientation method of accurate positioning.

In order to achieve the above object, the technical solution adopted in the present invention is a kind of room based on earth magnetism fingerprint matching algorithm Inner position method, comprises the following steps:

Step 1, gathers the earth magnetism fingerprint time series of region, required location critical path.Use low-pass filtering algorithm logarithm According to carrying out smoothing denoising process, and calculate the conjunction absolute force of locality magnetic data, as the earth magnetism fingerprint matching time Sequence；

Step 2, uses cubic spline interpolation algorithm that earth magnetism fingerprint matching vector carries out interpolation processing, simultaneously and building Mapping relations are set up in path corresponding to plane graph.Earth magnetism fingerprint matching time series and map maps relation are stored earth magnetism In fingerprint database；

Step 3, positioning stage, acquisition length is about the earth magnetism subsequence to be matched of 2s, similarly, uses cubic spline to insert Value-based algorithm carries out interpolation processing to this sequence.And the variation tendency (growth trend, minimax value etc.) of this subsequence is carried out Judge, in earth magnetism fingerprint database, screen corresponding candidate matching sequence according to its variation tendency；

Step 4, travels through all candidate matches sequences, calculate earth magnetism subsequence to be matched and their dynamic time warping away from From, find out the candidate matches sequence minimum with earth magnetism subsequence dynamic time warping distance to be matched, and by the terminal of this sequence It is mapped as corresponding map reference as position location.

Further, described step 1 specifically includes:

Step 11, certain the particular path hand-held intelligent mobile phone along region, location at the uniform velocity walks, mobile phone towards fixing, with The frequency collection of 2Hz is along road geomagnetic data.Use low-pass filtering algorithm that the geomagnetic data read is carried out smoothing denoising process, its Computing formula is:

B_i=B_i-1*(1-lowPassCoef)+currentB*lowPassCoef.

Wherein, B_iFor the absolute force data obtained after low-pass filtering, B_i-1The absolute force number obtained for the last time According to, currentB is the geomagnetic data that the non-filtered that current device reads processes, and lowPassCoef is filter factor, takes 0.1。

Step 12, records each time point t_iCorresponding absolute force B_i(x_i,y_i,z_i), calculate its magnetic intensity accordingly Close absolute force | | B_i| |, wherein

Step 13, by continuous print time point { t on particular path₁,t₂,…,t_nCorresponding conjunction absolute force | | B₁| |,||B₂||,…,||B_n| | composition earth magnetism fingerprint matching time series TS={ (t₁,||B₁||),(t₂,||B₂||),…,(t_n, ||B_n||)}。

Further, described step 2 specifically includes:

Step 21, uses three Based on Interpolating Splines that earth magnetism fingerprint matching time series TS is carried out interpolation processing, obtains TS ' ={ (t '₁,||B₁||′),(t′₂,||B₂||′),…,(t′_m,||B_m||′)}.Wherein t '_i-t′_i-1=0.1s.

Step 22, sets up the time series TS ' after interpolation processing with the path coordinate corresponding to building plan and reflects Penetrate relation, form corresponding earth magnetism fingerprint reference sequence RS={ (t '₁,||B₁||′,x₁,y₁),(t′₂,||B₂||′,x₂, y₂),…,(t′_m,||B_m||′,x_m,y_m), wherein, x_i,y_iRepresent respective path point horizontal stroke in two dimensional surface, vertical seat respectively Mark, and preserve to earth magnetism fingerprint database.

Further, described step 3 also needs to judge to have existed in earth magnetism fingerprint database to need location path corresponding Earth magnetism fingerprint reference sequence, if not, need to gather corresponding reference sequences, from earth magnetism fingerprint database, otherwise take out phase Answer reference sequences for mating location.

Further, described step 3 specifically includes:

Step 31, gathers the earth magnetism finger print data in continuous time (generally 2 seconds) and absolute force is closed in calculating accordingly. Similarly, and use cubic spline interpolation algorithm that this sequence carries out interpolation processing, and then obtain the earth magnetism to be matched of a length of n Time series MS={ (t₁,||B₁||),(t₂,||B₂||),…,(t_n,||B_n||)}。

Step 32, calculates the trend vector T of earth magnetism time series MS to be matched, T={ (t₁,k₁),(t₂,k₂),(t₃, k₃),…,(t_n-1,k_n-1)}.Wherein

Step 33, takes out the reference sequences in corresponding path to be positioned from earth magnetism fingerprint database, similarly, calculates reference The trend vector RT of sequence.And vectorial with the trend of reference sequences based on earth magnetism time series to be matched, utilize sliding window vocal mimicry Art, the vectorial similarity distance with time earth magnetism time series trend to be matched vector of calculation window internal reference Sequence Trend, sieve Select similarity distance candidate matches sequence in certain threshold value ω.

Further, described step 4 specifically includes:

Step 41, travels through all of candidate matches sequence, calculates earth magnetism fingerprint and the time of earth magnetism time series MS to be matched Select the similarity between matching sequence earth magnetism fingerprint.Similarity uses dynamic time warping algorithm to calculate.Assume to be matched The earth magnetism fingerprint sequence of earth magnetism time series MS is X={x₁,x₂,…,x_n, candidate matches sequence earth magnetism fingerprint sequence is Y= {y₁,y₂,…,y_n, then the dynamic time warping distance of two sequences is

$D_{dtw}(X,Y)=\left\{\begin{array}{c}0,ifm=n=0\\ \mathrm{\∞},ifm=0orn=0\\ d(x_1,y_1)+\min \left\{\begin{array}{c}{D}_{dte}(X,\mathrm{Re}st(Y))\\ D_{dtw}\left(\mathrm{Re}st\right(X),Y)\\ D_{dtw}\left(\mathrm{Re}st\right(X),\mathrm{Re}st\left(Y\right))\end{array}\right.\\ d(x,y)=\left|\right|x-y\left|\right|\end{array}\right.$

Step 42, finds out the highest with earth magnetism time series similarity to be matched, and i.e. dynamic time warping is apart from little candidate Matching sequence, and the terminal of this sequence is mapped as corresponding map reference, this coordinate is positioning result, completes location.

The present invention, by using technique scheme, compared with prior art, has the advantage that

One, earth's magnetic field local anomaly widely distributed in the present invention takes full advantage of building and feature steady in a long-term, Location can be realized without any infrastructure network；

Two, the present invention employs low-pass filtering algorithm at geomagnetic data gatherer process data is carried out pretreatment, it is possible to have Effect reduces the impact of noise data；

Three, the present invention has taken into full account earth magnetism seasonal effect in time series feature (be incremented by, successively decrease, maximum value minimum etc.), and with Trend vector form calculates, it is possible to be effectively improved the success rate of fingerprint matching algorithm.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of the inventive method.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage are clearer, below in conjunction with accompanying drawing by concrete real The present invention is described in more detail to execute example.Should be appreciated that specific embodiment described herein only in order to explain the present invention, It is not intended to limit the present invention.

As it is shown in figure 1, as a specific embodiment, a kind of based on earth magnetism fingerprint matching algorithm the indoor of the present invention Localization method, comprises the following steps:

Step 1, gathers the earth magnetism fingerprint time series of region, required location critical path.Use low-pass filtering algorithm logarithm According to carrying out smoothing denoising process, and calculate the conjunction absolute force of locality magnetic data, as the earth magnetism fingerprint matching time Sequence；

Step 2, uses cubic spline interpolation algorithm that earth magnetism fingerprint matching vector carries out interpolation processing, simultaneously and building Mapping relations are set up in path corresponding to plane graph.Earth magnetism fingerprint matching time series and map maps relation are stored earth magnetism In fingerprint database；

Step 3, positioning stage, acquisition length is about the earth magnetism subsequence to be matched of 2s, similarly, uses cubic spline to insert Value-based algorithm carries out interpolation processing to this sequence.And the variation tendency (growth trend, minimax value etc.) of this subsequence is carried out Judge, in earth magnetism fingerprint database, screen corresponding candidate matching sequence according to its variation tendency；

Step 4, travels through all candidate matches sequences, calculate earth magnetism subsequence to be matched and their dynamic time warping away from From, find out the candidate matches sequence minimum with earth magnetism subsequence dynamic time warping distance to be matched, and by the terminal of this sequence It is mapped as corresponding map reference as position location.

Described step 1 specifically includes:

Step 11, certain the particular path hand-held intelligent mobile phone along region, location at the uniform velocity walks, mobile phone towards fixing, with The frequency collection of 2Hz is along road geomagnetic data.Use low-pass filtering algorithm that the geomagnetic data read is carried out smoothing denoising process, its Computing formula is:

B_i=B_i-1*(1-lowPassCoef)+currentB*lowPassCoef.

Wherein, B_iFor the absolute force data obtained after low-pass filtering, B_i-1The absolute force number obtained for the last time According to, currentB is the geomagnetic data that the non-filtered that current device reads processes, and lowPassCoef is filter factor, takes 0.1。

Step 12, records each time point t_iCorresponding absolute force B_i(x_i,y_i,z_i), calculate its magnetic intensity accordingly Close absolute force | | B_i| |, wherein

Step 13, by continuous print time point { t on particular path₁,t₂,…,t_nCorresponding conjunction absolute force | | B₁| |,||B₂||,…,||B_n| | composition earth magnetism fingerprint matching time series TS={ (t₁,||B₁| |), (t₂,||B₂||),…,(t_n, ||B_n||)}。

Described step 2 specifically includes:

Step 21, uses three Based on Interpolating Splines that earth magnetism fingerprint matching time series TS is carried out interpolation processing, obtains TS ' ={ (t '₁,||B₁||′),(t′₂,||B₂||′),…,(t′_m,||B_m||′)}.Wherein t '_i-t′_i-1=0.1s.

Step 22, sets up the time series TS ' after interpolation processing with the path coordinate corresponding to building plan and reflects Penetrate relation, form corresponding earth magnetism fingerprint reference sequence RS={ (t '₁,||B₁||′,x₁,y₁),(t′₂,||B₂||′,x₂, y₂),…,(t′_m,||B_m||′,x_m,y_m), wherein, x_i,y_iRepresent respective path point horizontal stroke in two dimensional surface, vertical seat respectively Mark, and preserve to earth magnetism fingerprint database.

Described step 3 specifically includes:

Step 31, gathers the earth magnetism finger print data in continuous time (generally 2 seconds) and absolute force is closed in calculating accordingly. Similarly, and use cubic spline interpolation algorithm that this sequence carries out interpolation processing, and then obtain the earth magnetism to be matched of a length of n Time series MS={ (t₁,||B₁||),(t₂,||B₂||),…,(t_n,||B_n||)}。

Step 32, calculates the trend vector T of earth magnetism time series MS to be matched, T={ (t₁,k₁),(t₂,k₂),(t₃, k₃),…,(t_n-1,k_n-1)}.Wherein

Step 33, takes out the reference sequences in corresponding path to be positioned from earth magnetism fingerprint database, similarly, calculates reference The trend vector RT of sequence.And vectorial with the trend of reference sequences based on earth magnetism time series to be matched, utilize sliding window vocal mimicry Art, the vectorial similarity distance with time earth magnetism time series trend to be matched vector of calculation window internal reference Sequence Trend, sieve Select similarity distance candidate matches sequence in certain threshold value ω.

Described step 4 specifically includes:

Step 41, travels through all of candidate matches sequence, calculates earth magnetism fingerprint and the time of earth magnetism time series MS to be matched Select the similarity between matching sequence earth magnetism fingerprint.Similarity uses dynamic time warping algorithm to calculate.Assume to be matched The earth magnetism fingerprint sequence of earth magnetism time series MS is X={x₁,x₂,…,x_n, candidate matches sequence earth magnetism fingerprint sequence is Y= {y₁,y₂,…,y_n, then the dynamic time warping distance of two sequences is

$D_{dtw}(X,Y)=\left\{\begin{array}{c}0,ifm=n=0\\ \mathrm{\∞},ifm=0orn=0\\ d(x_1,y_1)+\min \left\{\begin{array}{c}{D}_{dte}(X,\mathrm{Re}st(Y))\\ D_{dtw}\left(\mathrm{Re}st\right(X),Y)\\ D_{dtw}\left(\mathrm{Re}st\right(X),\mathrm{Re}st\left(Y\right))\end{array}\right.\\ d(x,y)=\left|\right|x-y\left|\right|\end{array}\right.$

Step 42, finds out the highest with earth magnetism time series similarity to be matched, and i.e. dynamic time warping is apart from little candidate Matching sequence, and the terminal of this sequence is mapped as corresponding map reference, this coordinate is positioning result.

Claims (2)

1. an indoor orientation method based on earth magnetism fingerprint matching algorithm, it is characterised in that including:

Step 1, gathers the earth magnetism fingerprint time series of region, required location critical path；Use low-pass filtering algorithm that data are entered Row smoothing denoising processes, and calculate the conjunction absolute force of locality magnetic data, as earth magnetism fingerprint matching time series； Specifically include:

Step 11, at the uniform velocity walks along certain the particular path hand-held intelligent mobile phone positioning region, and mobile phone is towards fixing, with 2Hz's Frequency collection is along road geomagnetic data；Using low-pass filtering algorithm that the geomagnetic data read is carried out smoothing denoising process, it calculates Formula is:

B_i=B_i-1*(1-lowPassCoef)+currentB*lowPassCoef.

Wherein, B_iFor the absolute force data obtained after low-pass filtering, B_i-1The absolute force data obtained for the last time, CurrentB is the geomagnetic data that the non-filtered that current device reads processes, and lowPassCoef is filter factor, takes 0.1；

Step 12, records each time point t_iCorresponding absolute force B_i(x_i,y_i,z_i), calculate its accordingly magnetic intensity conjunction ground Magnetic intensity | | B_i| |, wherein

Step 13, by continuous print time point { t on particular path₁,t₂,…,t_nCorresponding conjunction absolute force | | B₁||,|| B₂||,…,||B_n| | composition earth magnetism fingerprint matching time series TS={ (t₁,||B₁||),(t₂,||B₂||),…,(t_n,||B_n| |)}；

Step 2, uses cubic spline interpolation algorithm that earth magnetism fingerprint matching vector carries out interpolation processing, simultaneously with building plane Mapping relations are set up in path corresponding to figure；Earth magnetism fingerprint matching time series and map maps relation are stored earth magnetism fingerprint In data base；Specifically include:

Step 21, uses three Based on Interpolating Splines that earth magnetism fingerprint matching time series TS is carried out interpolation processing, obtain TS '= {(t′₁,||B₁||′),(t′₂,||B₂||′),…,(t′_m,||B_m||′)}；Wherein t '_i-t′_i-1=0.1s；

Step 22, sets up to map with the path coordinate corresponding to building plan by the time series TS ' after interpolation processing and closes System, forms corresponding earth magnetism fingerprint reference sequence RS={ (t '₁,||B₁||′,x₁,y₁),(t′₂,||B₂||′,x₂,y₂),…, (t′_m,||B_m||′,x_m,y_m), wherein, x_i,y_iRepresent respective path point horizontal stroke in two dimensional surface, vertical coordinate respectively, and preserve To earth magnetism fingerprint database；

Step 3, positioning stage, acquisition length is about the earth magnetism subsequence to be matched of 2s, similarly, uses cubic spline interpolation to calculate Method carries out interpolation processing to this sequence；And the variation tendency (growth trend, minimax value etc.) of this subsequence is judged, In earth magnetism fingerprint database, corresponding candidate matching sequence is screened according to its variation tendency；Specifically include:

Step 31, gathers the earth magnetism finger print data in continuous time (generally 2 seconds) and absolute force is closed in calculating accordingly；Equally Ground, and use cubic spline interpolation algorithm that this sequence carries out interpolation processing, and then obtain the earth magnetism time to be matched of a length of n Sequence MS={ (t₁,||B₁||),(t₂,||B₂||),…,(t_n,||B_n||)}；

Step 32, calculates the trend vector T of earth magnetism time series MS to be matched, T={ (t₁,k₁),(t₂,k₂),(t₃,k₃),…, (t_n-1,k_n-1)}；Wherein

Step 33, takes out the reference sequences in corresponding path to be positioned from earth magnetism fingerprint database, similarly, calculates reference sequences Trend vector RT；And vectorial with the trend of reference sequences based on earth magnetism time series to be matched, utilize sliding window technique, meter Calculate the vectorial similarity distance with time earth magnetism time series trend to be matched vector of window internal reference Sequence Trend, filter out phase Like degree distance candidate matches sequence in certain threshold value ω；

Step 4, travels through all candidate matches sequences, calculates the earth magnetism subsequence to be matched dynamic time warping distance with them, Find out the candidate matches sequence minimum with earth magnetism subsequence dynamic time warping distance to be matched, and the terminal of this sequence is mapped For corresponding map reference as position location, specifically include:

Step 41, travels through all of candidate matches sequence, calculates earth magnetism fingerprint and the candidate of earth magnetism time series MS to be matched Join the similarity between sequence earth magnetism fingerprint；Similarity uses dynamic time warping algorithm to calculate；Assume earth magnetism to be matched The earth magnetism fingerprint sequence of time series MS is X={x₁,x₂,…,x_n, candidate matches sequence earth magnetism fingerprint sequence is Y={y₁, y₂,…,y_n, then the dynamic time warping distance of two sequences is

$D_{dtw}(X,Y)=\left\{\begin{array}{c}0,ifm=n=0\\ \mathrm{\∞},ifm=0orn=0\\ d(x_1,y_1)+\min \left\{\begin{array}{c}{D}_{dtw}(X,\mathrm{Re}st(Y\left)\right)\\ D_{dtw}\left(\mathrm{Re}st\right(X),Y)\\ D_{dtw}\left(\mathrm{Re}st\right(X),\mathrm{Re}st(Y\left)\right)\end{array}\right.\\ d(x,y)=\left|\right|x-y\left|\right|\end{array}\right.$

Step 42, finds out the highest with earth magnetism time series similarity to be matched, and i.e. dynamic time warping is apart from little candidate matches Sequence, and the terminal of this sequence is mapped as corresponding map reference, this coordinate is positioning result.

A kind of indoor orientation method based on earth magnetism fingerprint matching algorithm the most according to claim 1, it is characterised in that institute State step 3 also to need to judge earth magnetism fingerprint database has existed to need location path corresponding earth magnetism fingerprint reference sequence, If not, need to gather corresponding reference sequences, from earth magnetism fingerprint database, otherwise take out corresponding reference sequences fixed for mating Position.