CN106289242A - Particle filtering method and device based on earth magnetism - Google Patents

Abstract

The invention discloses a particle filter positioning method and device based on geomagnetism. The method is as follows: obtain rough positioning results and positioning variance, obtain pedestrian step length and step direction, establish a magnetic field map through preset point intervals, and obtain magnetic field through fixed-point continuous acquisition value, according to the rough positioning results and positioning variance, calculate the first normal distribution, according to the step length and the step direction, correspondingly calculate the second normal distribution and the third normal distribution, and according to the magnetic field value, calculate the variance of the magnetic field measurement; through the particle filter to match the magnetic field, each time the pedestrian takes a step, the corresponding particles are re-arranged, and the position of the pedestrian is corrected according to the change of the magnetic field where the particles are located to obtain the actual position of the pedestrian. The geomagnetism-based particle filter positioning method and device of the present invention do not need to give the starting position of the pedestrian, but adopt the particle filter to correct the position of the pedestrian according to the change of the magnetic field where the particle is located, so as to obtain the actual position of the pedestrian.

Description

Particle filter positioning method and device based on geomagnetism

technical field

The invention relates to the field of indoor positioning, in particular to a geomagnetic-based particle filter positioning method and device.

Background technique

The geomagnetic field originates from the interior of the earth and is relatively stable. Generally, it is less affected by the outside world. Theoretically speaking, any location on the earth should have different geomagnetic data, and even the same location has different altitudes, so the geomagnetic data is also different. The theoretical basis of geomagnetic navigation is provided. Modern buildings are mostly reinforced concrete or steel structures, which bend the geomagnetic field in local space, but are stable in time, and have different magnetic field elements in different directions.

The geomagnetic positioning technology mainly uses the geomagnetic sensor device to collect three-dimensional geomagnetic data indoors, and generates a geomagnetic reference map after computer processing, which is used for geomagnetic indoor positioning. Because the magnetometer on the mobile phone itself is affected by other magnets and soft iron effects, there is an error, and there is a certain error between the result after the coordinate change and the actual measurement value, which will easily lead to system failure.

In the prior art, indoor positioning usually uses the dead reckoning PDR algorithm for pedestrians. Through the PDR algorithm, the number of steps, step length, and direction of the line are obtained for measurement and statistics, and information such as the walking track and position of the pedestrian is calculated. However, this method The positioning error will accumulate over time, and the long-term positioning accuracy is low; and in the prior art, it is usually necessary to know the initial position of the pedestrian, and then calculate the indoor positioning result of the pedestrian based on the pedestrian's step length, number of steps and direction .

Therefore, the defect in the prior art is that the actual position of the measured target cannot be located according to the observation of the measured target (pedestrian) to the surrounding magnetic field environment during motion without an accurate initial position.

Contents of the invention

The technical problem to be solved by the present invention is to provide a particle filter positioning method based on geomagnetism. The particle filter method does not need to give the starting position of the pedestrian. Every time the pedestrian takes a step, the corresponding particles are re-collected, and according to the magnetic field where the particles are located, the corresponding particles are collected again. change to get the actual position of the pedestrian.

In order to solve the problems of the technologies described above, the technical solution provided by the invention is:

In the first aspect, the present invention provides a geomagnetic-based particle filter positioning method, including:

Step S1, obtain rough positioning results and positioning variance, obtain pedestrian step length and step direction, establish a magnetic field map through preset point intervals, and obtain magnetic field values through fixed-point continuous acquisition. The magnetic field value is the value of the magnetic field regardless of the direction of the magnetic field ;

Step S2: Calculate the first normal distribution according to the rough positioning result and the positioning variance, and calculate the second normal distribution and the third normal distribution according to the step length and the step direction distribution, and according to the magnetic field value, calculate the magnetic field measurement variance;

Step S3, laying out a plurality of particles according to the first normal distribution, and according to the second normal distribution and the third normal distribution, each of the particles is distributed according to the second normal distribution and the third normal distribution. The third normal distribution moves, and obtains the magnetic field value of each particle in the position of the magnetic field diagram after the movement;

Step S4, according to the magnetic field value of each particle in the magnetic field map after the movement, combined with the magnetic field measurement variance, calculate the probability density of the measured magnetic field value according to the normal distribution, and the probability density is used as each The weight of the particle, obtaining the relationship between the position of each particle and its weight;

Step S5, according to the relationship between the position of each particle and its weight, rearrange new particles, the number of the new particles is the same as the number of the previously arranged particles;

Step S6, according to the new particles, perform weighted average processing on their positions in the magnetic field map to obtain the actual positioning position of the pedestrian after taking a step.

The technical scheme of the particle filter positioning method based on geomagnetism in the present invention is: first obtain the rough positioning result and the positioning variance, and there is no limit to the method used to obtain it. For example, it can be obtained according to the magnetometer on the pedestrian's mobile phone, and the pedestrian's step length and step direction can be obtained. , to establish a magnetic field map through preset point intervals, that is, to establish a magnetic field map in advance according to the interval distance, and at the same time obtain the magnetic field value through fixed-point continuous acquisition, and the magnetic field value is the magnetic field value regardless of the magnetic field direction; The absolute value of the magnetic field at each fixed point is taken, that is, regardless of the direction of the corresponding magnetic field, after multiple continuous acquisitions, the obtained magnetic field value is more accurate and stable.

Then, according to the rough positioning result and the positioning variance, the rough positioning value and the positioning variance are brought into the normal distribution formula to obtain the first normal distribution; in the same way, according to the step length and the step direction, Correspondingly calculating the second normal distribution and the third normal distribution, and calculating the magnetic field measurement variance according to the magnetic field value;

Then according to the first normal distribution, a plurality of particles are arranged; that is, a plurality of particles are randomly arranged within a certain range of the rough positioning result, and then according to the second normal distribution and the third normal distribution, Each particle moves according to the second normal distribution and the third normal distribution, and obtains the magnetic field value of each particle in the position of the magnetic field diagram after the movement; each particle moves according to the second normal distribution The normal distribution and the third normal distribution move in the preset step size and direction, and each particle moves in a different step size and direction.

After moving, each particle is changed to a new magnetic field environment, and according to the magnetic field value of each particle in the magnetic field diagram after the movement, combined with the magnetic field measurement variance, calculate the measured magnetic field value according to the normal distribution. Probability density, the probability density is used as the weight of each of the particles, and the relationship between the position of each of the particles and its weight is obtained; then re-sampling of the same number of new particles; according to the new particles, the The position in the magnetic field map is weighted and averaged to obtain the actual positioning position of the pedestrian after taking a step. Because the pedestrian movement is continuous, according to this method, when the pedestrian takes another step, the particles of this step are used as the old particles. When the environment (magnetic field environment) around the old particles changes, new particles are re-arranged. The number of particles is the same as that of the previous layout; in this way, the actual position of the pedestrian in motion is obtained. The principle of particle rearrangement is that according to the particle weight obtained by calculation, the higher the weight, the higher the probability of being selected again when re-arranging. Correction, and finally realize the positioning of pedestrians according to the particles resampled each time.

The particle filter positioning method based on the geomagnetism of the present invention adopts the particle filter method, and does not need to give the starting position of the pedestrian. Every time the pedestrian takes a step, the corresponding particles are collected again, and the position of the pedestrian is corrected according to the change of the magnetic field where the particles are located. Get the actual location of the pedestrian.

Further, the interval between the preset points is 0.01 meters. Within the scope of indoor space, establish a magnetic field map according to the preset point interval, set a fixed point every 0.01 meters, and get the distribution data of the indoor geomagnetic field through the magnetic field map, that is, the magnetic field size data; therefore, it has been verified by experiments that the interval should not be set too high If the distance is too large, a lot of magnetic field distribution data will be missed, so set the interval to 0.01 meters, and the obtained magnetic field data can make the indoor positioning results more accurate.

Further, in the step 2, obtaining the second normal distribution and the third normal distribution also includes:

Step S21, according to the step length and the step direction, respectively calculate the variance of the step length and the variance of the step direction through multiple measurement statistics, and respectively calculate the average value of the step length and the step direction;

Step S22, according to the step length variance, the step direction variance, the mean value of the step length and the mean value of the step direction, correspondingly obtain a second normal distribution and a third normal distribution.

To calculate the normal distribution, according to the formula of the normal distribution, the mean value and variance of the corresponding step length and the mean value and variance of the corresponding step direction must be obtained, and in order to make the obtained mean value variance more accurate, it is necessary to go through multiple measurements and statistics. The more you get the step length value, the more stable the mean value and the more accurate the variance. Correspondingly, the more you get the step direction, the more accurate the mean value and variance are.

Further, in the step S3, each of the particles moves according to the second normal distribution and the third normal distribution, further comprising:

Step S31, according to the step size and step direction, each of the particles moves according to a specific step size and a specific direction, and the specific step size satisfies the second normal of the step size and the step size variance distribution, the particular direction satisfies a third normal distribution of the step direction and the variance of the step direction;

Step S32, each of the particles moves according to the second normal distribution and the third normal distribution.

The particles move according to the obtained second normal distribution and the third normal distribution. In essence, the particles move according to a random step size, which is a preset step size from the second normal distribution. , correspondingly, the moving direction of the particles is moving according to the stepping direction preset by the third normal distribution, but the moving direction and step size of each particle are different, so they move according to the normal distribution. That is to say, the step length of particle movement satisfies the second normal distribution, and the moving direction of particles satisfies the third normal distribution.

Further, the pedestrian's stride length and stride direction are obtained through the pedestrian's dead reckoning PDR model. PDR (Pedestrian Dead Reckoning) is called Pedestrian Dead Reckoning, and it is called Pedestrian Dead Reckoning. It mainly calculates and obtains the position of people walking indoors. And statistics, deduce information such as the walking track and position of the pedestrian. The algorithm is simple.

In the second aspect, the present invention provides a particle filter positioning device based on geomagnetism, including:

The data acquisition module is used to obtain rough positioning results and positioning variance, obtain pedestrian step length and step direction, establish a magnetic field map through preset point intervals, and obtain magnetic field values through fixed-point continuous acquisition. The magnetic field value does not count the direction of the magnetic field Magnetic field value;

The normal distribution calculation module is used to calculate and obtain the first normal distribution according to the rough positioning result and the positioning variance, and to obtain the second normal distribution and the corresponding calculation according to the step length and the step direction. The third normal distribution, and according to the magnetic field value, calculate the magnetic field measurement variance;

A particle moving module, configured to lay out a plurality of particles according to the first normal distribution, and according to the second normal distribution and the third normal distribution, each of the particles according to the second normal distribution The state distribution and the third normal distribution are moved, and the magnetic field value of each particle in the position of the magnetic field diagram is obtained after the movement;

The weight value calculation module is used to calculate the probability density of the measured magnetic field value according to the normal distribution according to the magnetic field value of each particle in the position of the magnetic field diagram after the movement, and the probability density of the measured magnetic field value in combination with the magnetic field measurement variance. The density is used as the weight of each particle, and the relationship between the position of each particle and its weight is obtained;

The new particle acquisition module is used to rearrange new particles according to the relationship between the position of each particle and its weight, and the number of the new particles is the same as the number of the particles previously arranged;

The positioning position acquisition module is used to perform weighted average processing on the position of the new particle in the magnetic field map to obtain the actual positioning position of the pedestrian after taking a step.

The technical scheme of the particle filter positioning device based on geomagnetism in the present invention is as follows: firstly, through the data acquisition module, the rough positioning result and positioning variance are obtained, and the method is not limited. Step length and step direction, establish the magnetic field map through the preset point interval, that is, establish the magnetic field map according to the interval distance in advance, and obtain the magnetic field value through fixed-point continuous acquisition. The magnetic field value is the value of the magnetic field regardless of the magnetic field direction; The most important thing is that the absolute value of the magnetic field value of each fixed point in the magnetic field diagram is taken, that is, the direction of the corresponding magnetic field size is ignored. After multiple continuous acquisitions, the obtained magnetic field value is more accurate and stable.

Then, through the normal distribution calculation module, according to the rough positioning result and the positioning variance, the rough positioning value and the positioning variance are brought into the normal distribution formula to obtain the first normal distribution; in the same way, according to the step-by-step The length and the stepping direction are calculated correspondingly to obtain the second normal distribution and the third normal distribution, and according to the magnetic field value, the magnetic field measurement variance is calculated;

Then through the particle layout module, according to the first normal distribution, a plurality of particles are arranged; that is, a plurality of particles are randomly set within a certain range of the rough positioning result, and then through the particle movement module, according to the second normal distribution normal distribution and the third normal distribution, each particle moves according to the second normal distribution and the third normal distribution, and obtains the position of each particle in the magnetic field diagram after the movement Magnetic field value; each particle moves according to the preset step size and direction of the second normal distribution and the third normal distribution, and each particle moves with a different step size and direction.

After moving, each particle is changed to a new magnetic field environment. Through the weight value calculation module, according to the magnetic field value of each particle in the magnetic field diagram after the movement, combined with the magnetic field measurement variance, according to the normal distribution Calculate the probability density of the measured magnetic field value, and use the probability density as the weight of each particle to obtain the relationship between the position of each particle and its weight; then, through the new particle acquisition module, according to the weight of each particle The relationship between the position and its weight, re-arrange new particles, the number of new particles is the same as the number of particles arranged before; then by positioning the position acquisition module, according to the new particles, their positions in the magnetic field map are weighted and averaged processing to obtain the actual positioning position of the pedestrian after taking a step. Because the pedestrian movement is continuous, according to this method, when the pedestrian takes another step, the particles of this step are used as the old particles. When the environment (magnetic field environment) around the old particles changes, new particles with the same number as the old particles Particles circulate in this way to get the actual position of the pedestrian in motion. The principle of particle rearrangement is that according to the particle weight obtained by calculation, the higher the weight, the higher the probability of being selected again when re-arranging. Correction, and finally realize the positioning of pedestrians according to the particles resampled each time.

The particle filter positioning device based on the geomagnetism of the present invention adopts the method of particle filter, without giving the starting position of the pedestrian, every time the pedestrian takes a step, the corresponding particles are collected again, and the position of the pedestrian is corrected according to the change of the magnetic field where the particles are located. Get the actual location of the pedestrian.

Further, in the normal distribution calculation module, obtaining the second normal distribution and the third normal distribution also includes:

The mean variance calculation module is used to calculate the variance of the step length and the variance of the step direction through multiple measurement statistics according to the step length and the step direction, respectively calculate the step length and the step mean of direction;

The normal distribution obtaining module is used to obtain the second normal distribution and the third normal distribution correspondingly according to the variance of the step length, the variance of the direction of the step, the mean value of the step length and the mean value of the direction of the step .

To calculate the normal distribution, according to the formula of the normal distribution, the mean value and variance of the corresponding step length are obtained, and the mean value and variance of the corresponding step direction are obtained, so the second normal distribution and the third normal distribution obtain the mean value of the variance The calculation module and the normal distribution acquisition module are obtained together; and, in order to make the obtained mean variance more accurate, it needs to go through multiple measurement statistics, the more the step length value is obtained, the more stable the mean value and the more accurate the variance, correspondingly, step The more directions you get, the more accurate your mean and variance will be.

Further, in the particle moving module, moving each particle according to the second normal distribution and the third normal distribution also includes:

A particle movement condition module, configured to move each particle according to a specific step size and a specific direction according to the step size and the step direction, and the specific step size satisfies the variance of the step size and the step size variance a second normal distribution, said particular direction satisfying a third normal distribution of said step direction and said step direction variance;

The particle moving submodule is configured to move each particle according to the second normal distribution and the third normal distribution.

The particles move according to the obtained second normal distribution and the third normal distribution. In essence, the particles move according to a random step size, which is a preset step size from the second normal distribution. , correspondingly, the moving direction of the particles is moving according to the stepping direction preset by the third normal distribution, but the moving direction and step size of each particle are different, so they move according to the normal distribution. That is to say, the step length of particle movement satisfies the second normal distribution, and the moving direction of particles satisfies the third normal distribution.

Description of drawings

In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that are required in the description of the specific embodiments or the prior art.

FIG. 1 shows a flow chart of a geomagnetic-based particle filter positioning method provided by the first embodiment of the present invention;

Fig. 2 shows a structural block diagram of a geomagnetism-based particle filter positioning device provided by the first embodiment of the present invention.

detailed description

Embodiments of the technical solutions of the present invention will be described in detail below in conjunction with the accompanying drawings. The following examples are only used to illustrate the technical solution of the present invention more clearly, so they are only examples, and should not be used to limit the protection scope of the present invention.

Embodiment one

Fig. 1 shows a flow chart of a geomagnetism-based particle filter positioning method provided by the first embodiment of the present invention. As shown in FIG. 1, the geomagnetism-based particle filter positioning method according to the first embodiment of the present invention includes:

Step S1, obtain rough positioning results and positioning variance, obtain pedestrian step length and step direction, establish a magnetic field map through preset point intervals, obtain magnetic field values through fixed-point continuous acquisition, and the magnetic field value is the magnetic field value regardless of the magnetic field direction;

Step S2, calculate the first normal distribution according to the roughly measured positioning results and the positioning variance, and obtain the second normal distribution and the third normal distribution according to the step length and step direction, and obtain the second normal distribution and the third normal distribution according to the magnetic field value, calculate the magnetic field measurement variance;

Step S3, lay out multiple particles according to the first normal distribution, and move each particle according to the second normal distribution and the third normal distribution according to the second normal distribution and the third normal distribution , and get the magnetic field value of each particle in the magnetic field diagram after moving;

Step S4, according to the magnetic field value of each particle in the magnetic field map after the movement, combined with the variance of the magnetic field measurement, calculate the probability density of the measured magnetic field value according to the normal distribution, the probability density is used as the weight of each particle, and each particle is obtained The relationship between the position and its weight;

Step S5, according to the relationship between the position of each particle and its weight, rearrange new particles, the number of new particles is the same as the number of particles previously arranged;

Step S6, according to the new particle, its position in the magnetic field map is weighted and averaged to obtain the actual positioning position of the pedestrian after taking a step.

The technical scheme of the particle filter positioning method based on geomagnetism in the present invention is: first obtain the rough positioning result and the positioning variance, and there is no limit to the method used to obtain it. For example, it can be obtained according to the magnetometer on the pedestrian's mobile phone, and the pedestrian's step length and step direction can be obtained. , to establish a magnetic field map through preset point intervals, that is, to establish a magnetic field map in advance according to the interval distance, and at the same time obtain the magnetic field value through fixed-point continuous acquisition. The magnetic field value is the value of the magnetic field regardless of the magnetic field direction; The absolute value of the magnetic field at a fixed point is taken, that is, regardless of the direction of the corresponding magnetic field, after multiple continuous acquisitions, the obtained magnetic field value is more accurate and stable.

Then, according to the rough positioning results and positioning variance, the rough positioning value and positioning variance are brought into the normal distribution formula to obtain the first normal distribution; in the same way, according to the step length and step direction, the corresponding calculation is obtained to obtain the second Normal distribution and the third normal distribution, and according to the magnetic field value, calculate the magnetic field measurement variance;

Then according to the first normal distribution, a plurality of particles are arranged; that is, a plurality of particles are randomly set within a certain range of the rough positioning results, and then according to the second normal distribution and the third normal distribution, each Particles move according to the second normal distribution and the third normal distribution, and get the magnetic field value of each particle in the magnetic field map after moving; each particle moves according to the second normal distribution and the third normal distribution The step size and direction of each particle's movement are different.

After moving, each particle is changed to a new magnetic field environment. According to the magnetic field value of each particle in the magnetic field diagram after the movement, combined with the variance of the magnetic field measurement, the probability density of the measured magnetic field value is calculated according to the normal distribution, and the probability density is taken as According to the weight of each particle, the relationship between the position of each particle and its weight is obtained; then new particles are re-arranged, and the number of new particles is the same as the number of particles previously arranged; according to the position of the new particles in the magnetic field diagram Perform weighted average processing to obtain the actual positioning position of the pedestrian after taking a step. Because the pedestrian movement is continuous, according to this method, when the pedestrian takes another step, the particles of this step are used as the old particles. When the environment (magnetic field environment) around the old particles changes, new particles with the same number as the old particles Particles circulate in this way to get the actual position of the pedestrian in motion. The principle of particle rearrangement is that according to the particle weight obtained by calculation, the higher the weight, the higher the probability of being selected again when re-arranging. Correction, and finally realize the positioning of pedestrians according to the particles resampled each time.

The particle filter positioning method based on the geomagnetism of the present invention adopts the particle filter method, and does not need to give the starting position of the pedestrian. Every time the pedestrian takes a step, the corresponding particles are collected again, and the position of the pedestrian is corrected according to the change of the magnetic field where the particles are located. Get the actual location of the pedestrian.

Specifically, the idea of Particle Filter (PF: Particle Filter) is based on Monte Carlo methods, which use particle sets to represent probability, and can be used in any form of state space model. Its core idea is to express its distribution through random state particles extracted from the posterior probability, which is a sequential importance sampling method (Sequential Importance Sampling). In simple terms, the particle filter method refers to the process of approximating the probability density function by finding a group of random samples propagated in the state space, and replacing the integral operation with the sample mean value, so as to obtain the minimum variance distribution of the state. The samples here refer to the particles, and when the number of samples is N→∝, any form of probability density distribution can be approximated.

Although the probability distribution in the algorithm is only an approximation of the real distribution, due to the non-parametric characteristics, it gets rid of the constraint that the random quantity must satisfy the Gaussian distribution when solving the nonlinear filtering problem, and can express a wider distribution than the Gaussian model. It also has a stronger modeling ability for the nonlinear characteristics of variable parameters. Therefore, particle filtering can more accurately express the posterior probability distribution based on observations and control quantities, and can be used to solve SLAM instant localization and map construction (simultaneous localization and mapping) problems. In the present invention, particle filtering is combined with geomagnetism for indoor positioning, that is, every time a pedestrian takes a step, the particles in the magnetic field map also move one step, and then the probability density is calculated by the magnetic field value corresponding to the position after the particle moves, that is, by finding a set of The random particles moving in the indoor space approximate the probability density function, and the actual position is finally obtained by gradually increasing the probability density (particle weight).

Specifically, the interval between preset points is 0.01 meters. Within the scope of indoor space, establish a magnetic field map according to the preset point interval, set a fixed point every 0.01 meters, and get the distribution data of the indoor geomagnetic field through the magnetic field map, that is, the magnetic field size data; therefore, it has been verified by experiments that the interval should not be set too high If the distance is too large, a lot of magnetic field distribution data will be missed, so set the interval to 0.01 meters, and the obtained magnetic field data can make the indoor positioning results more accurate.

Specifically, in step 2, obtaining the second normal distribution and the third normal distribution also includes:

Step S21, according to the stride length and the stride direction, respectively calculate the variance of the stride length and the variance of the stride direction through multiple measurement statistics, and respectively calculate the mean values of the stride length and the stride direction;

Step S22, according to the variance of the step length, the variance of the step direction, the mean value of the step length and the mean value of the step direction, correspondingly obtain the second normal distribution and the third normal distribution.

To calculate the normal distribution, according to the formula of the normal distribution, the mean value and variance of the corresponding step length and the mean value and variance of the corresponding step direction must be obtained, and in order to make the obtained mean value variance more accurate, it is necessary to go through multiple measurements and statistics. The more you get the step length value, the more stable the mean value and the more accurate the variance. Correspondingly, the more you get the step direction, the more accurate the mean value and variance are.

Specifically, in step S3, moving each particle according to the second normal distribution and the third normal distribution also includes:

Step S31, according to the step size and step direction, each particle moves according to a specific step size and a specific direction, the specific step size satisfies the second normal distribution of the step size and the variance of the step size, and the specific direction satisfies the step direction and the step direction third normal distribution of variance;

Step S32, each particle moves according to the second normal distribution and the third normal distribution.

The particles move according to the obtained second normal distribution and the third normal distribution. In essence, the particles move according to a random step size, which is a preset step size from the second normal distribution. , correspondingly, the moving direction of the particles is moving according to the stepping direction preset by the third normal distribution, but the moving direction and step size of each particle are different, so they move according to the normal distribution. That is to say, the step length of particle movement satisfies the second normal distribution, and the moving direction of particles satisfies the third normal distribution.

Specifically, the pedestrian's step length and step direction are obtained through the pedestrian's dead reckoning PDR model. PDR is called Pedestrian Dead Reckoning, which mainly calculates the position of people walking indoors. This algorithm measures and counts the number of steps, step length, and direction of pedestrians, and calculates the walking trajectory and position of pedestrians. . The algorithm is simple.

Fig. 2 shows a structural block diagram of a geomagnetism-based particle filter positioning device provided by the first embodiment of the present invention. As shown in Figure 2, the present invention provides a particle filter positioning device 10 based on geomagnetism, including:

The data acquisition module 101 is used to obtain rough positioning results and positioning variance, obtain pedestrian step length and step direction, establish a magnetic field map through preset point intervals, and obtain magnetic field values through fixed-point continuous acquisition. The magnetic field value does not count the direction of the magnetic field The value of the magnetic field;

The normal distribution calculation module 102 is used to calculate the first normal distribution according to the rough positioning results and the positioning variance, and to obtain the second normal distribution and the third normal distribution according to the step length and the step direction. distribution, and calculate the magnetic field measurement variance according to the magnetic field value;

The particle movement module 103 is used to arrange a plurality of particles according to the first normal distribution, and according to the second normal distribution and the third normal distribution, each particle is distributed according to the second normal distribution and the third normal distribution. The state distribution is moved, and the magnetic field value of each particle in the magnetic field diagram is obtained after the movement;

The weight value calculation module 104 is used to calculate the probability density of the measured magnetic field value according to the normal distribution according to the magnetic field value of each particle in the position of the magnetic field diagram after the movement, and the probability density is used as the weight of each particle , get the relationship between the position of each particle and its weight;

The new particle acquisition module 105 is used to rearrange new particles according to the relationship between the position of each particle and its weight, and the number of new particles is the same as the number of particles arranged before;

The positioning position acquisition module 106 is used to perform weighted average processing on the position of the new particle in the magnetic field map to obtain the actual positioning position of the pedestrian after taking a step.

The technical solution of the geomagnetism-based particle filter positioning device 10 of the present invention is: firstly, through the data acquisition module 101, obtain the rough measurement positioning result and the positioning variance, and there is no limit to the method used to obtain it. For example, it can be obtained from the magnetometer on the mobile phone of a pedestrian. Pedestrians step length and step direction, establish the magnetic field map through the preset point interval, that is, establish the magnetic field map according to the interval distance in advance, and obtain the magnetic field value through fixed-point continuous acquisition. The magnetic field value is the magnetic field value regardless of the magnetic field direction; it is the magnetic field The value refers to the absolute value of the magnetic field at each fixed point in the magnetic field diagram, that is, regardless of the direction of the corresponding magnetic field, after multiple continuous acquisitions, the obtained magnetic field value is more accurate and stable.

Then, through the normal distribution calculation module 102, according to the rough positioning result and the positioning variance, the rough positioning value and the positioning variance are brought into the normal distribution formula to obtain the first normal distribution; Direction, the second normal distribution and the third normal distribution are calculated correspondingly, and the magnetic field measurement variance is calculated according to the magnetic field value;

Then through the particle moving module 103, a plurality of particles are arranged according to the first normal distribution, that is, a plurality of particles are randomly set within a certain range of the rough positioning results, and according to the second normal distribution and the third normal distribution , each particle moves according to the second normal distribution and the third normal distribution, and obtains the magnetic field value of each particle in the magnetic field diagram after moving; each particle moves according to the second normal distribution and the third normal distribution The steps and directions preset by the three normal distributions are used to move, and the steps and directions of each particle's movement are different.

After moving, each particle is changed to a new magnetic field environment. Through the weight value calculation module 104, the measured magnetic field value is calculated according to the normal distribution according to the magnetic field value of each particle in the magnetic field map after the movement, combined with the variance of the magnetic field measurement The probability density of the probability density is used as the weight of each particle to obtain the relationship between the position of each particle and its weight; then through the new particle acquisition module 105, according to the relationship between the position of each particle and its weight, new particles are rearranged, The number of new particles is the same as the number of particles placed previously.

Finally, through the positioning position acquisition module 106, according to the new particle, its position in the magnetic field map is weighted and averaged to obtain the actual positioning position of the pedestrian after taking a step. Because the pedestrian movement is continuous, according to this method, when the pedestrian takes another step, the particles of this step are used as the old particles. When the environment (magnetic field environment) around the old particles changes, new particles with the same number as the old particles Particles circulate in this way to get the actual position of the pedestrian in motion. The principle of particle rearrangement is that according to the particle weight obtained by calculation, the higher the weight, the higher the probability of being selected again when re-arranging. Correction, and finally realize the positioning of pedestrians according to the particles resampled each time.

The geomagnetism-based particle filter positioning device 10 of the present invention adopts particle filtering, and does not need to give the starting position of the pedestrian. Every time the pedestrian takes a step, the corresponding particles are collected again, and the position of the pedestrian is corrected according to the change of the magnetic field where the particles are located. The actual location of the pedestrian.

Specifically, in the normal distribution calculation module 102, obtaining the second normal distribution and the third normal distribution also includes:

The mean variance calculation module 1021 is used to calculate the variance of the step length and the variance of the step direction through multiple measurement statistics according to the step length and the step direction, and calculate the mean values of the step length and the step direction respectively;

The normal distribution obtaining module 1022 is used to correspondingly obtain the second normal distribution and the third normal distribution according to the step length variance, the step direction variance, the mean value of the step length and the mean value of the step direction.

To calculate the normal distribution, according to the formula of the normal distribution, the mean value and variance of the corresponding step length are obtained, and the mean value and variance of the corresponding step direction are obtained, so the second normal distribution and the third normal distribution obtain the mean value of the variance The calculation module 1021 and the normal distribution obtaining module 1022 are obtained together; and, in order to make the obtained mean variance more accurate, it is necessary to go through multiple measurement statistics, the more the step length value is obtained, the more stable the mean value and the more accurate the variance, the corresponding , the more the step direction is obtained, the more accurate the mean and variance are.

Specifically, in the particle moving module 103, moving each particle according to the second normal distribution and the third normal distribution also includes:

The particle movement condition module 1031 is used to move each particle according to a specific step size and a specific direction according to the step size and the step direction, the specific step size satisfies the second normal distribution of the step size and the step size variance, and the specific direction satisfies third normal distribution of step direction and step direction variance;

The particle movement sub-module 1032 is configured to move each particle according to the second normal distribution and the third normal distribution.

The particles move according to the obtained second normal distribution and the third normal distribution. In essence, the particles move according to a random step size, which is a preset step size from the second normal distribution. , correspondingly, the moving direction of the particles is moving according to the stepping direction preset by the third normal distribution, but the moving direction and step size of each particle are different, so they move according to the normal distribution. That is to say, the step length of particle movement satisfies the second normal distribution, and the moving direction of particles satisfies the third normal distribution.

To sum up, the geomagnetism-based particle filter positioning method and device of the present invention adopts the particle filter method, and does not need to give the starting position of the pedestrian. Every time the pedestrian takes a step, the corresponding particles are re-collected. According to the change of the magnetic field where the particles are located, the The pedestrian position is corrected to obtain the actual position of the pedestrian. Use geomagnetic positioning combined with particle filter to improve measurement accuracy.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: It is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the various embodiments of the present invention. All of them should be covered by the scope of the claims and description of the present invention.

Claims (10)

1. particle filtering method based on earth magnetism, it is characterised in that including:

Step S1, obtains bigness scale positioning result and positioning variances, obtains pedestrian and takes a step step-length and take a step direction, by default fixed point Magnetic chart is set up at interval, obtains magnetic field value by fixed point continuous acquisition, and described magnetic field value is the magnetic field size disregarding magnetic direction Value；

Step S2, according to described bigness scale positioning result and positioning variances, is calculated first normal distribution, takes a step according to described Step-length and described direction of taking a step, correspondence is calculated second normal distribution and the 3rd normal distribution, and according to described magnetic field Value, is calculated magnetic-field measurement variance；

Step S3, according to described first normal distribution, lays multiple particle, according to described second normal distribution and the 3rd Normal distribution, each described particle is moved by described second normal distribution and the 3rd normal distribution, and is moved Rear each particle magnetic field value of position in described magnetic chart；

Step S4, according to particle each after described movement magnetic field value of position in described magnetic chart, in conjunction with described magnetic field Measuring variance, calculate the probability density of actual measurement magnetic field value according to normal distribution, described probability density is as each described particle Weight, obtain the position of each described particle and the relation of its weight；

Step S5, according to position and the relation of its weight of each described particle, lays new particle, the number of described new particle again Measure identical with the described number of particles laid before；

Step S6, according to described new particle, is weighted average treatment to its position in described magnetic chart, obtains pedestrian and steps Actual location position after one step.

Particle filtering method based on earth magnetism the most according to claim 1, it is characterised in that

Described default fixed point is spaced apart 0.01 meter.

Particle filtering method based on earth magnetism the most according to claim 1, it is characterised in that

In described step 2, described second normal distribution and the 3rd normal distribution obtain and also include:

Step S21, according to described step-length and the described direction of taking a step of taking a step, adds up step of taking a step described in calculating respectively by repetitive measurement Long variance and described direction variance of taking a step, take a step described in calculating respectively step-length and the average in described direction of taking a step；

Step S22, according to described step-length variance of taking a step, described in take a step direction variance, described in take a step step-length average and described take a step Direction average, correspondence obtains second normal distribution and the 3rd normal distribution.

4. according to particle filtering method based on earth magnetism described in claim 1 or 3, it is characterised in that

In described step S4, each described particle is moved also wrapped by described second normal distribution and the 3rd normal distribution Include:

Step S41, according to described step-length and the direction of taking a step of taking a step, each described particle moves according to particular step size and specific direction Dynamic, described particular step size meet described in take a step step-length and the second normal distribution of described step-length variance of taking a step, described specific direction Take a step described in Man Zuing direction and the 3rd normal distribution of described direction variance of taking a step；

Step S42, according to described second normal distribution and described 3rd normal distribution, each described particle moves.

Particle filtering method based on earth magnetism the most according to claim 1, it is characterised in that

Described pedestrian takes a step step-length with direction of taking a step by the acquisition of pedestrian's dead reckoning PDR model.

6. particle filtering device based on earth magnetism, it is characterised in that including:

Data acquisition module, is used for obtaining bigness scale positioning result and positioning variances, obtains pedestrian and takes a step step-length and direction of taking a step, to lead to Crossing default fixed point interval and set up magnetic chart, obtain magnetic field value by fixed point continuous acquisition, described magnetic field value is for disregarding magnetic direction Magnetic field sizes values；

Normal distribution computing module, for according to described bigness scale positioning result and positioning variances, is calculated first normal state and divides Cloth, according to described step-length and the described direction of taking a step of taking a step, correspondence is calculated second normal distribution and the 3rd normal distribution, And according to described magnetic field value, it is calculated magnetic-field measurement variance；

Particle mobile module, for according to described first normal distribution, lays multiple particle, divides according to described second normal state Cloth and the 3rd normal distribution, each described particle is moved by described second normal distribution and the 3rd normal distribution, And each particle magnetic field value of position in described magnetic chart after being moved；

Weight value calculation module, is used for according to particle each after described movement magnetic field value of position in described magnetic chart, In conjunction with described magnetic-field measurement variance, calculate the probability density of actual measurement magnetic field value, described probability density conduct according to normal distribution The weight of each described particle, obtains the position of each described particle and the relation of its weight；

New particle acquisition module, for the position according to each described particle and the relation of its weight, lays new particle, institute again The quantity stating new particle is identical with the described number of particles laid before；

Position location acquisition module, for according to described new particle, is weighted averagely its position in described magnetic chart Process, obtain the actual location position after pedestrian steps a step.

Particle filtering device based on earth magnetism the most according to claim 6, it is characterised in that

Described default fixed point is spaced apart 0.01 meter.

Particle filtering device based on earth magnetism the most according to claim 6, it is characterised in that

In described normal distribution computing module, described second normal distribution and the 3rd normal distribution obtain and also include:

Mean variance computing module, for take a step described in basis step-length and described direction of taking a step, by repetitive measurement statistics respectively Step-length of taking a step described in calculating variance and described direction variance of taking a step, calculate respectively described in take a step the equal of step-length and described direction of taking a step Value；

Normal distribution obtains module, for according to described in take a step step-length variance, described in take a step direction variance, described in step-length of taking a step equal Value and described direction average of taking a step, correspondence obtains second normal distribution and the 3rd normal distribution.

9. according to particle filtering device based on earth magnetism described in claim 6 or 8, it is characterised in that

In described particle mobile module, each described particle is moved by described second normal distribution and the 3rd normal distribution Move and also include:

Particle mobile condition module, for step-length and the direction of taking a step of taking a step described in basis, each described particle is according to particular step size Move with specific direction, described particular step size meet described in take a step step-length and the second normal distribution of described step-length variance of taking a step, Described specific direction is taken a step direction and the 3rd normal distribution of described direction variance of taking a step described in meeting；

Particle mover module, for according to described second normal distribution and described 3rd normal distribution, each described particle enters Row is mobile.

Particle filtering device based on earth magnetism the most according to claim 6, it is characterised in that

Described pedestrian takes a step step-length with direction of taking a step by the acquisition of pedestrian's dead reckoning PDR model.