WO2018233692A1 - Positioning method, storage medium, and positioning system - Google Patents

Abstract

A positioning method, a storage medium, and a positioning system. The positioning method comprises: performing predictive positioning of a terminal to be positioned according to a wireless local area network WiFi positioning signal sent by the terminal to be positioned, so as to obtain a WiFi-based position of the terminal to be positioned; performing predictive positioning of the terminal to be positioned according to a light-emitting diode (LED) positioning signal sent by the terminal to be positioned, so as to obtain an LED-based position of the terminal to be positioned; and obtaining the final position of the terminal to be positioned according to the WiFi-based position, the LED-based position and the weights of the two positions. The positioning method combines the advantages of the two positioning methods to achieve the complementary advantages of different positioning methods, and accordingly, the WiFi positioning and the visible light positioning are effectively combined, thereby improving the positioning precision and stability.

Description

Positioning method, storage medium and positioning system Technical field

The present disclosure relates to the field of positioning, and in particular to a positioning method, a storage medium, and a positioning system.

Background technique

In recent years, indoor location services, which are increasingly urgent for demand, have developed rapidly in indoor positioning technology. They are research hotspots in the era of mobile internet, and gradually play a role in all walks of life, which has a certain impact on people's daily life. However, compared with the outdoor environment, the indoor environment is limited by the positioning time, positioning accuracy, and indoor complex environment. The positioning effect is still not ideal in practical applications.

At present, the common methods of indoor positioning mainly include the following:

1, infrared positioning

The infrared ray is positioned by the infrared ray emitted by the infrared ray transmitter received by the optical sensor in the room. The Active Badge System, an infrared outdoor positioning system developed by AT&T Labs at the University of Cambridge, is known as the first generation of indoor positioning systems. Ambiplex proposed in 2011 that the IR.Loc system is positioned by measuring thermal radiation, with a positioning accuracy of 20 to 30 cm in the range of 10 m.

The implementation of the infrared indoor positioning system consists of two parts: an infrared emitter and an infrared receiver. Usually, the infrared emitter is a fixed node of the network, and the infrared receiver is mounted on the target to be positioned as a mobile terminal. The advantages of infrared indoor positioning are high positioning accuracy, responsiveness, and low cost for a single device.

2, Bluetooth indoor positioning

The Bluetooth indoor positioning is located by the fingerprint positioning algorithm according to the measurement terminal device signal strength. iBeacon is a protocol technology developed by Apple for Bluetooth positioning. The positioning accuracy is 2~3m. The shopping application Shopkick is deployed in the mall. iBeacon is applied in real life. China's “find deer” and “Guangfa easy go” and other APPs This mode is also used for positioning. Bluetooth positioning technology has high security, low cost, low power consumption and small size. At present, most mobile terminals have their own Bluetooth modules, which is easy to popularize and deploy.

3, radio frequency identification and positioning

Radio Frequency Identification (RFID) positioning technology uses radio frequency signals for non-contact two-way communication to exchange data for identification and positioning purposes. At present, representative RFID positioning systems include the Cricket system developed by the MIT Oxygen project, the SpotON system of the University of Washington, and the RADAR system of Microsoft Corporation. RFID technology has a large transmission range and low cost.

Although there are many mature positioning methods at present, each positioning method has its own significant unavoidable disadvantages. For example, the infrared positioning method is limited to the line of sight positioning, and the infrared information is greatly attenuated in the air, and can only be used for short The distance positioning and positioning accuracy are easily affected by other light sources; the Bluetooth indoor positioning method is susceptible to interference from external noise signals, the signal stability is poor, and the communication range is small; the RF indoor positioning method has a short working distance, and the longest is only tens of meters. Moreover, the radio frequency signal does not have the communication capability, and only the radio frequency identification technology cannot be used for indoor positioning, and must be combined with other auxiliary technologies to complete.

Summary of the invention

The technical problem to be solved by the present disclosure is to provide a positioning method, a storage medium, and a positioning system, which are used to solve the problem that the positioning accuracy in the prior art is not high.

To solve the above technical problem, in one aspect, the present disclosure provides a positioning method, including:

Determining and positioning the positioning terminal according to the wireless local area network WiFi positioning signal sent by the positioning terminal, and obtaining a WiFi positioning position of the positioning terminal;

Determining and positioning the positioning terminal according to the LED LED positioning signal sent by the positioning terminal, and obtaining the LED positioning position of the positioning terminal;

Obtaining a weight value corresponding to the two positioning positions according to the WiFi positioning position and the LED positioning position;

The final positioning position of the positioning terminal is obtained according to the weight position corresponding to the WiFi positioning position, the LED positioning position, and the two positioning positions.

In an embodiment, before the positioning is performed, the positioning area is divided into a plurality of grid points in advance, the WiFi weight value corresponding to the WiFi positioning on each grid point is obtained, and the LED corresponding to the LED positioning on each grid point is obtained. Weights;

In the positioning process, the WiFi weight value corresponding to the grid point closest to the WiFi positioning position is used as the weight value corresponding to the WiFi positioning position; and the grid point closest to the LED positioning position is corresponding to The LED weight value is used as the weight value corresponding to the LED positioning position.

In an embodiment, the WiFi weight value corresponding to the WiFi location on each grid point is obtained, including:

Obtaining a WiFi fingerprint database according to the received signal strength of the set number of times collected at each grid point indicating the RSSI data;

According to the WiFi fingerprint library, a WiFi classifier that uses the RSSI data for location prediction positioning is obtained;

Calculating the WiFi weight value w _r1 corresponding to the WiFi location on each grid point using equations (2) and (3);

ε(x _r (i))=||x _r (i)-x _r || ² Formula (3)

Where argmin represents the minimum obtained w _r1 value of equation (2); ε(x _r (i)) represents the predicted positioning coordinate x _r (i) of the i-th collected data at grid point r and the true position coordinate x _r Square error; N is the number of samples collected at grid point r, r is the grid number, r = 1, ..., R.

In an embodiment, the method further includes:

After obtaining the WiFi fingerprint database, the WiFi fingerprint library is divided into a training set and a verification set according to a set ratio;

The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the WiFi classifier.

In an embodiment, the positioning terminal is predicted and located according to the WiFi classifier, and the WiFi positioning position of the positioning terminal is obtained.

In one embodiment, the LED weight values corresponding to the LED positioning on each grid point are obtained, including:

According to the LED positioning data collected at each grid point, the LED fingerprint library is obtained;

According to the LED fingerprint library, an LED classifier for position prediction positioning using LED positioning data is obtained;

Calculate the LED weight value w _r2 corresponding to the LED positioning on each grid point using equations (3) and (4);

ε(x _r (i))=||x _r (i)-x _r || ² Formula (3)

Where argmin represents the minimum obtained w _r2 value of equation (2); ε(x _r (i)) represents the predicted positioning coordinate x _r (i) of the i-th collected data at grid point r and the true position coordinate x _r Square error; N is the number of samples collected at grid point r, r is the grid number, r = 1, ..., R.

In an embodiment, the method further includes:

After obtaining the LED fingerprint library, the LED fingerprint library is divided into a training set and a verification set according to a set ratio;

The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the LED classifier.

In one embodiment, the LED positioning data includes: a heading angle, a pitch angle, and a roll angle of the device that collects the LED positioning data, and position coordinates of one or more LED lights captured by the device on the image. .

In an embodiment, according to the LED classifier, the positioning terminal is predicted and positioned to obtain an LED positioning position of the positioning terminal.

In one embodiment, the final positioning position of the positioning terminal is obtained by formula (1):

Where p denotes the final positioning position of the positioning terminal; j=1, 2; f ₁ (x) denotes a predicted positioning position based on WiFi positioning at the grid point r; f ₂ (x) denotes an LED based on the grid point r The predicted positioning position obtained by the positioning; w _r1 represents the weight value corresponding to the WiFi positioning at the grid point r; w _r2 represents the weight value corresponding to the LED positioning at the grid point r.

In another aspect, the present disclosure also provides a method for acquiring an LED positioning signal, including

Using an positioning terminal to capture an image including an LED light;

Performing image processing to acquire a blinking frequency of each LED light in the image, and identifying the captured LED light according to the blinking frequency;

Obtaining coordinates of the identified LED light in the image, and obtaining a heading angle, a pitch angle, and a roll angle of the positioning terminal;

The heading angle, the pitch angle and the roll angle of the positioning terminal, and the coordinates of the identified LED lamp in the image form a positioning signal at the position and position of the positioning terminal.

In one embodiment, image processing is performed to obtain a blinking frequency of each LED light in the image, including:

Converting a color image to a grayscale image;

Perform image blurring and binarization processing in sequence;

Separating the graphics corresponding to each LED light in the image, extracting the contour information of each graphic, and finding the center position;

Within each contour of the graph, the spacing of the black and white stripes is counted by the radon transform, and the blinking frequency of the LED corresponding to the contour of the graphic is calculated according to the spacing.

In another aspect, the present disclosure also provides a storage medium storing a positioning program for performing positioning, the program comprising:

Determining and positioning the positioning terminal according to the wireless local area network WiFi positioning signal sent by the positioning terminal, and obtaining a WiFi positioning position of the positioning terminal;

Determining and positioning the positioning terminal according to the LED LED positioning signal sent by the positioning terminal, and obtaining the LED positioning position of the positioning terminal;

Obtaining a weight value corresponding to the two positioning positions according to the WiFi positioning position and the LED positioning position;

The final positioning position of the positioning terminal is obtained according to the weight position corresponding to the WiFi positioning position, the LED positioning position, and the two positioning positions.

In an embodiment, before the positioning is performed, the positioning area is divided into a plurality of grid points in advance, the WiFi weight value corresponding to the WiFi positioning on each grid point is obtained, and the LED corresponding to the LED positioning on each grid point is obtained. Weights;

In the positioning process, the WiFi weight value corresponding to the grid point closest to the WiFi positioning position is used as the weight value corresponding to the WiFi positioning position; and the grid point closest to the LED positioning position is corresponding to The LED weight value is used as the weight value corresponding to the LED positioning position.

In an embodiment, the WiFi weight value corresponding to the WiFi location on each grid point is obtained, including:

Obtaining a WiFi fingerprint database according to the received signal strength of the set number of times collected at each grid point indicating the RSSI data;

According to the WiFi fingerprint library, a WiFi classifier that uses the RSSI data for location prediction positioning is obtained;

Calculating the WiFi weight value w _r1 corresponding to the WiFi location on each grid point using equations (2) and (3);

ε(x _r (i))=||x _r (i)-x _r || ² Formula (3)

Where argmin represents the minimum obtained w _r1 value of equation (2); ε(x _r (i)) represents the predicted positioning coordinate x _r (i) of the i-th collected data at grid point r and the true position coordinate x _r Square error; N is the number of samples collected at grid point r, r is the grid number, r = 1, ..., R.

In one embodiment, the program includes:

After obtaining the WiFi fingerprint database, the WiFi fingerprint library is divided into a training set and a verification set according to a set ratio;

The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the WiFi classifier.

In an embodiment, the positioning terminal is predicted and located according to the WiFi classifier, and the WiFi positioning position of the positioning terminal is obtained.

In one embodiment, the LED weight values corresponding to the LED positioning on each grid point are obtained, including:

According to the LED positioning data collected at each grid point, the LED fingerprint library is obtained;

According to the LED fingerprint library, an LED classifier for position prediction positioning using LED positioning data is obtained;

Calculate the LED weight value w _r2 corresponding to the LED positioning on each grid point using equations (3) and (4);

ε(x _r (i))=||x _r (i)-x _r || ² Formula (3)

Where argmin represents the minimum obtained w _r2 value of equation (2); ε(x _r (i)) represents the predicted positioning coordinate x _r (i) of the i-th collected data at grid point r and the true position coordinate x _r Square error; N is the number of samples collected at grid point r, r is the grid number, r = 1, ..., R.

In one embodiment, the program further comprises:

After obtaining the LED fingerprint library, the LED fingerprint library is divided into a training set and a verification set according to a set ratio;

The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the LED classifier.

In one embodiment, the LED positioning data includes: a heading angle, a pitch angle, and a roll angle of the device that collects the LED positioning data, and position coordinates of one or more LED lights captured by the device on the image. .

In an embodiment, according to the LED classifier, the positioning terminal is predicted and positioned to obtain an LED positioning position of the positioning terminal.

In one embodiment, the final positioning position of the positioning terminal is obtained by formula (1):

Where p denotes the final positioning position of the positioning terminal; j=1, 2; f ₁ (x) denotes a predicted positioning position based on WiFi positioning at the grid point r; f ₂ (x) denotes an LED based on the grid point r The predicted positioning position obtained by the positioning; w _r1 represents the weight value corresponding to the WiFi positioning at the grid point r; w _r2 represents the weight value corresponding to the LED positioning at the grid point r.

In another aspect, the present disclosure also provides a storage medium storing a program for acquiring an LED positioning signal, the program comprising:

Using an positioning terminal to capture an image including an LED light;

Performing image processing to acquire a blinking frequency of each LED light in the image, and identifying the captured LED light according to the blinking frequency;

Obtaining coordinates of the identified LED light in the image, and obtaining a heading angle, a pitch angle, and a roll angle of the positioning terminal;

The heading angle, the pitch angle and the roll angle of the positioning terminal, and the coordinates of the identified LED lamp in the image form a positioning signal at the position and position of the positioning terminal.

In one embodiment, image processing is performed to obtain a blinking frequency of each LED light in the image, including:

Converting a color image to a grayscale image;

Perform image blurring and binarization processing in sequence;

Separating the graphics corresponding to each LED light in the image, extracting the contour information of each graphic, and finding the center position;

Within each contour of the graph, the spacing of the black and white stripes is counted by the radon transform, and the blinking frequency of the LED corresponding to the contour of the graphic is calculated according to the spacing.

In still another aspect, the present disclosure also provides a positioning system including a plurality of wireless routers disposed in a positioning area, a plurality of LED lights, and a positioning terminal, a monitoring terminal, and a server:

The positioning terminal collects the WiFi positioning signal and the LED positioning signal of the location grid point, and sends the positioning signal to the server;

The server locates the positioning terminal according to the WiFi positioning signal and the LED positioning signal, and sends the positioning information to the monitoring terminal;

The monitoring terminal receives the positioning information and displays the location of the positioning terminal.

In one embodiment, the server includes a storage medium storing a fusion locator.

In one embodiment, the positioning terminal includes the storage medium stored above for an image processing program.

In one embodiment, the LEDs are modulated to a square wave signal, and the modulation frequencies of the individual LED lamps are different.

The beneficial effects of the disclosure are as follows:

The present disclosure proposes a fusion positioning method and system based on WiFi and visible light, which combines the advantages of the two positioning methods, and realizes the complementary advantages of different positioning methods, thereby effectively combining WiFi positioning and visible light positioning, thereby improving positioning accuracy and stability.

DRAWINGS

In the drawings, which are not necessarily to scale, the Like reference numerals with different letter suffixes may indicate different examples of similar components. The drawings generally illustrate the various embodiments discussed herein by way of example and not limitation.

1 is an effect diagram of an LED lamp actually photographed in an embodiment of the present disclosure;

2 is an effect diagram of photographing LED lamps of different frequencies in an embodiment of the present disclosure;

3 is a schematic structural view of a positioning system in an embodiment of the present disclosure;

4 is a flow chart of positioning in an embodiment of the present disclosure.

Detailed ways

The present disclosure will be further described in detail below in conjunction with the drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the disclosure and are not intended to be limiting.

Embodiment 1

The embodiment of the present disclosure proposes a dynamic fingerprint fusion positioning method based on WiFi and visible light.

Before the positioning, a classifier for performing WiFi positioning and LED positioning is separately constructed, and then a system for performing fusion positioning is constructed. This process is referred to as an "offline phase" in this embodiment. Using the built system to perform the process of fusion positioning, this process is called the “online phase”.

Therefore, the embodiment includes two major parts, and the first part is a fusion positioning system that separately constructs a WiFi classifier and an LED classifier, and fuses the two positioning methods together. The second part is to use the built-in WiFi classifier and LED classifier, as well as the fusion positioning system for specific positioning.

First, the construction steps of the WiFi classifier are described:

(1) Establish a WiFi fingerprint library

Four wireless routers are deployed in the room (positioning area), placed in different corners. In the "offline phase", the room is first divided into different grid points, and the room is first divided into 6*8=48 grid points. In this example, the grid point size is 1m×1m.

Then collect WiFi RSSI data at different grid points, hold the mobile phone (positioning terminal) in a grid point, control the mobile phone to start collecting data, and put the corresponding label (ie grid number: 1, 2, ... 48), save Go to the file. Collected in 48 grid points, 100 sets of RSSI data are collected at each grid point, and each group of data contains RSSI information of 4 routers, thus obtaining a 48*100*4 WiFi fingerprint library.

(2) Constructing a classifier

After getting the WiFi fingerprint library, you can train and build the classifier. We choose the Python language for processing. Based on the powerful machine learning library Scikit-Learn in Python, we first divide the fingerprint database into a training set (Train set) and a validation set (Validation set) according to 9:1 to avoid overfitting. . Since the characteristics of the model are not linearly separable, we choose the nonlinear classifier. Here we choose the Support Vector Machine (SVM) model to train and get the WiFi-based WiFi classifier. Due to the small amount of data, training can be completed in tens of seconds.

After the WiFi classifier is obtained, each newly acquired RSSI data is input into the WiFi classifier, and a predicted positioning result of the RSSI data collection location is obtained, and the positioning result can locate the target location into a specific grid point.

Second, describe the construction steps of the LED classifier:

(1) Set the LED light

Set a plurality of LED lights in the positioning area, and record the position of each LED light; modulate the LED light into a square wave signal (a square wave higher than 100 Hz, the human eye does not notice the flicker, and does not cause visual discomfort, but only The brightness will be appropriately reduced, and the square wave signal is easy to handle, and the anti-interference ability is strong), usually modulated to 200 Hz-4000 Hz, and each LED lamp is modulated to a different frequency.

It should be noted that, because the positioning position of the LED is required to collect the position information of the LED, and the image processing is required after the photographing, the larger the light-emitting area of the LED, the better the recognition effect. Therefore, the circular panel light is selected in this scheme. If it is a normal lamp, the light-emitting area can be enlarged by adding a lamp cover.

In addition, in order to facilitate image processing, the positioning terminal sets the camera exposure time (shutter) (the normal mobile phone camera can adjust the exposure time), so that the LED lights in the captured image show light and dark stripes; the remaining objects will be presented on the image. It is a black background (due to the short exposure time, the effect of non-illuminating objects such as ceilings and furniture forming a "blackout" in the image), which is more advantageous for image processing. As shown in Figure 1.

(2) Establish a fingerprint library

The structure classifier needs to extract features first. In the "WiFi positioning link", the feature is directly available, that is, the value of RSSI. In the LED positioning, the amount of image data captured is very large, and the really useful information is very small, so it is necessary to pre-process the image, extract the feature, and the fingerprint library stores the feature instead of the entire image.

The frequency of each LED lamp is different, and the performance in the captured image is that the stripe pitch (width corresponds to the modulated frequency) is different, as shown in FIG. 2, therefore, by image processing, the corresponding image of the lamp can be identified. The frequency is equal to the identification of the LED light. The position coordinates of each LED lamp are fixed, so the imaging position of different LED lights in the image can be a feature of positioning.

The process of LED lamp recognition through image processing is as follows:

1. Convert the captured color photos into grayscale images to increase the processing speed while removing interference.

2. Blurring the image (improving the accuracy of LED separation) and performing binarization processing to separate each LED lamp; binarization processing is to set a global threshold T, and use T to image the data. Divided into two parts: a pixel group larger than T and a pixel group smaller than T. The pixel value of the pixel group larger than T is set to white or black, and the pixel value of the pixel group smaller than T is set to black or white.

3. Separate each LED light, extract the contour information of each LED light, and find the center position of each LED (the pixel position in the image). Within each contour, the radon transform can be used to calculate the pitch of the black and white stripes, thereby calculating the blinking frequency of the LED and performing LED lamp recognition based on the blinking frequency.

Due to the limited viewing angle of the camera, the range that can be captured is limited, and it is difficult to capture multiple LEDs at the same time, so the orientation cannot be located. If the direction cannot be determined, the positioning accuracy will be greatly reduced. The angle at which the phone takes a photo also affects the imaging position of the LED in the image. In order to improve the positioning accuracy, you can use the electronic compass and gyroscope of the smartphone to get the three-dimensional tilt angle of the mobile phone, namely the heading angle (Yaw), the pitch angle (Pitch), and the roll angle (Roll). Using the three-dimensional angle as another feature of LED positioning can improve the accuracy of positioning. In smartphones (Android, IOS), the API for 3D tilt is open, so the 3D tilt feature is easy to obtain.

Therefore, to build an LED fingerprint library, the information contained in each feature is shown in Table 1:

Table 1

Yaw

Pitch

Roll

X1

Y1

X2

Y2

X3

Y3

X4

Y4

Wherein, the first three features represent the three-dimensional tilt angle, and Xi and Yi respectively represent the imaging position of the LED lamp labeled i in the image (ie, the pixel coordinates in the image are positive numbers), if the LED labeled i does not Appear in the image, then let Xi = Yi = -1.

For example, when Yaw=10.0, Pitch=20.0, Roll=30.0, only the 3rd LED appears in the image, and the position of the 3rd LED in the image is (500,800). The feature representation is shown in Table 2:

Table 2

Yaw	Pitch	Roll	X1	Y1	X2	Y2	X3	Y3	X4	Y4
10.0	20.0	30.0	-1	-1	-1	-1	500	800	-1	-1

As with WiFi positioning, building a fingerprint library requires collecting data (LED positioning data) in each grid point, extracting the information in the image according to the feature format shown above, and obtaining the three-dimensional tilt angle, storing it, and each grid point. After the collection is completed, the LED fingerprint library can be constructed. Unlike WiFi positioning, only data is collected once at each location because the data obtained by the camera is relatively stable.

(3) Construction of LED classifier

As with WiFi positioning, after obtaining the fingerprint library, training is required to construct the classifier. The method is the same as WiFi positioning. The LED positioning also selects the SVM classifier, but the feature dimension is different (WiFi positioning 4D features, LED positioning 11-dimensional features). Through the regression (regression) algorithm in machine learning, the mapping relationship between the LED positioning data and the predicted positional positioning is obtained, that is, the LED classifier is obtained.

In the actual positioning, the mobile phone automatically takes a photo, recognizes the LED according to the image processing method described above, and then acquires the three-dimensional tilt angle to jointly form the 11-dimensional feature. It is possible to input the LED classifier constructed in (3) and output a predicted positioning result based on the LED positioning.

LED positioning is directly dependent on the position of the LED in the image, and the LED is sometimes invisible, that is, the mobile phone camera can not capture the LED, and therefore can not get the positioning result, then it needs to rely more on WiFi positioning.

Finally, the process of combining the above two positioning methods is as follows:

First, set the fusion indicator ε(x _r (i)):

ε(x _r (i))=||x _r (i)-x _r || ² (1)

ε(x _r (i)) represents the squared error of the predicted value coordinate x _r (i) of the i-th acquired sample at the lattice point r and the true coordinate x _r . Wherein, the predicted value coordinate x _r (i) is predicted by the classifier obtained above, and the real coordinate x _{r is} obtained by actual measurement. For example, in the actual positioning, the grid coordinates predicted by the system are (1, 2), and the actual grid coordinates are (1, 1), then

Therefore, if the index is small, that is, the closer the predicted position and the real position are, the better the positioning accuracy of the system is represented.

Secondly, the corresponding weight w _{rj is} calculated from the fusion index:

Equation (2) is the weight calculation formula, and argmin is the w _rj value obtained by the minimum of the above formula. Where N is the number of samples collected at grid point r, r represents the grid location, j represents the fingerprint library, r=1,...,R;j=1,2; N refers to the number of samples collected at grid point r; i=1, 2..., N.

According to the fingerprint database data, the weight coefficient of the fingerprint library j at the grid point r can be calculated. Specifically, in the offline stage, assuming that we have collected the WiFi location fingerprint database and the LED location fingerprint database, we now take 80% of the data in the fingerprint library for classifier training, and the remaining 20% of the data is used to calculate the weight value. .

For example, now that the classifier has been trained, one of the remaining 20% of the data can be predicted by the classifier, assuming that the predicted grid coordinates are x1=(1,1), but the real grid The point coordinate is x=(0.5,0.5), then the value calculated by the fusion index of (1) is used.

Now we are using the weight formula of (2) to optimize this positioning result. That is, we want ε(w*x1)=||w*x1-x|| ² , 0≤w≤1 to be the smallest. In this example, w = 0.5, that is, when we take this value, we can make ε(w*x1) the smallest, which is 0.

According to this idea, we can find a weight matrix.

w ₁₁ refers to the optimization weight of the WiFI positioning of the grid 1 , and w ₁₂ refers to the optimized weight of the LED positioning of the grid 1 . That is to say, each grid point can calculate the optimization weight of the corresponding WiFi positioning and the optimization weight of the LED positioning by the above method.

After the system is set up, enter the location service phase.

First, the positioning terminal collects data of its location;

Then, according to the similarity, matching the data of the fingerprint library;

Finally, according to the data of the fingerprint database, the positioning result is calculated by using formula (3):

f _j (x) refers to the most similar grid r coordinate obtained by the data collected on the line matching the fingerprint library j, and then multiplied by the weight w _{rj of the} corresponding fingerprint corresponding to the corresponding position obtained under the line to obtain the final positioning position p.

For example, the result of the positioning prediction by the WiFi classifier is (1, 2), and the result of the positioning prediction by the LED classifier is (2, 1), and then according to the similarity matching, the grid point in the fingerprint database that is most similar to the prediction result is found. Find the optimization weight of the grid point in the weight matrix. Specifically, for example, the WiFi (1, 2) is closest to the WiFi fingerprint library point 1, then the weight w ₁₁ is taken, and the result of the LED positioning prediction is closest to the grid point 2 in the LED fingerprint library, then the weight is taken. ₂₂ , such a final output p = w ₁₁ × (1, 2) + w ₂₂ × (2, 1).

Therefore, according to the measured data in the positioning process, the solution in this embodiment dynamically adjusts the positioning result according to the weight, thereby improving the performance of the entire system; combining the advantages of the two positioning methods to achieve complementary advantages of different positioning methods, thereby enabling WiFi The combination of positioning and visible light positioning improves the accuracy and stability of positioning.

Embodiment 2

As shown in FIG. 3, an embodiment of the present disclosure also relates to a positioning system including a wireless local area network and visible light communication. The positioning system comprises a positioning end, a server end and a monitoring end, wherein the positioning end is carried by the monitored target, which is used for locating the location of the target and transmitting the location information to the server end; the server side calculates the positioning result, realizes the positioning fusion, and The positioning result is sent to the monitoring end; the monitoring end is carried by the monitor, and it can view the position information of the monitored target in real time. The communication system includes two parts: a wireless local area network and a visible light communication. The wireless local area network includes a plurality of APs (Access Points), which transmit a WiFi signal as a transmitting end, and the mobile phone serves as a receiving end to perform wireless positioning by using signal strength. The visible light communication system includes a plurality of LED lights of different frequencies, and the mobile phone recognizes different LED lights through visible light communication for positioning. Finally, the positioning method of the first embodiment is used to fuse the results of the two positioning systems to obtain the final positioning.

The positioning system of this embodiment can also be used for multi-target positioning and monitoring, that is, one monitoring terminal mobile phone can monitor a plurality of different positioning terminals.

The process of positioning using the system is as shown in FIG. 4, including:

Step 401, the positioning terminal firstly realizes the collection of the WiFi RSSI and the LED visible light signal, and the LED visible light signal is the image captured by the camera. After the image processing, the frequency and position information contained in the LED image are extracted to form the LED positioning information; The positioning terminal then sends the WiFi positioning information and the LED positioning information to the server.

Step 402: The server calculates a final positioning result, and performs predictive positioning on the positioning terminal according to the wireless local area network WiFi positioning signal sent by the positioning terminal, to obtain a WiFi positioning position of the positioning terminal.

Determining and positioning the positioning terminal according to the LED LED positioning signal sent by the positioning terminal, and obtaining the LED positioning position of the positioning terminal;

Obtaining a weight value corresponding to the two positioning positions according to the WiFi positioning position and the LED positioning position;

The final positioning position of the positioning terminal is obtained according to the weight position corresponding to the WiFi positioning position, the LED positioning position, and the two positioning positions.

Step 403: The server sends the positioning result to the monitoring terminal, and displays the location of the positioning terminal to implement positioning.

The embodiment also relates to a storage medium disposed on the server side, and the storage medium may be a server, or a storage, calling, and processing device of a data installed on the server, used to establish a WiFi fingerprint database, and construct a WiFi classification. And perform WiFi positioning, establish LED fingerprint library, LED classifier and LED positioning, and use to establish weight matrix, and integrate WiFi positioning and LED positioning to determine the final positioning result.

In one embodiment, the storage medium stores a positioning program for performing positioning, the program comprising:

Determining and positioning the positioning terminal according to the wireless local area network WiFi positioning signal sent by the positioning terminal, and obtaining a WiFi positioning position of the positioning terminal;

Determining and positioning the positioning terminal according to the LED LED positioning signal sent by the positioning terminal, and obtaining the LED positioning position of the positioning terminal;

Obtaining a weight value corresponding to the two positioning positions according to the WiFi positioning position and the LED positioning position;

The final positioning position of the positioning terminal is obtained according to the weight position corresponding to the WiFi positioning position, the LED positioning position, and the two positioning positions.

Before performing the positioning, the positioning area is divided into a plurality of grid points in advance, the WiFi weight values corresponding to the WiFi positioning on each grid point are obtained, and the LED weight values corresponding to the LED positioning on each grid point are obtained;

In the positioning process, the WiFi weight value corresponding to the grid point closest to the WiFi positioning position is used as the weight value corresponding to the WiFi positioning position; and the grid point closest to the LED positioning position is corresponding to The LED weight value is used as the weight value corresponding to the LED positioning position.

The WiFi weight value corresponding to the WiFi location on each grid point is obtained, including:

Obtaining a WiFi fingerprint database according to the received signal strength of the set number of times collected at each grid point indicating the RSSI data;

According to the WiFi fingerprint library, a WiFi classifier that uses the RSSI data for location prediction positioning is obtained;

Calculating the WiFi weight value w _r1 corresponding to the WiFi location on each grid point using equations (2) and (3);

ε(x _r (i))=||x _r (i)-x _r || ² Formula (3)

Where argmin represents the minimum obtained w _r1 value of equation (2); ε(x _r (i)) represents the predicted positioning coordinate x _r (i) of the i-th collected data at grid point r and the true position coordinate x _r Square error; N is the number of samples collected at grid point r, r is the grid number, r = 1, ..., R.

The method further includes:

After obtaining the WiFi fingerprint database, the WiFi fingerprint library is divided into a training set and a verification set according to a set ratio;

The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the WiFi classifier.

According to the WiFi classifier, the positioning terminal is predicted and located, and the WiFi positioning position of the positioning terminal is obtained.

Wherein, obtaining LED weight values corresponding to LED positioning on each grid point, including:

According to the LED positioning data collected at each grid point, the LED fingerprint library is obtained;

According to the LED fingerprint library, an LED classifier for position prediction positioning using LED positioning data is obtained;

Calculate the LED weight value w _r2 corresponding to the LED positioning on each grid point using equations (3) and (4);

ε(x _r (i))=||x _r (i)-x _r || ² Formula (3)

Where argmin represents the minimum obtained w _r2 value of equation (2); ε(x _r (i)) represents the predicted positioning coordinate x _r (i) of the i-th collected data at grid point r and the true position coordinate x _r Square error; N is the number of samples collected at grid point r, r is the grid number, r = 1, ..., R.

The method further includes:

After obtaining the LED fingerprint library, the LED fingerprint library is divided into a training set and a verification set according to a set ratio;

The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the LED classifier.

The LED positioning data includes: a heading angle, a pitch angle, and a roll angle of the device that collects the LED positioning data, and position coordinates of the one or more LED lights captured by the device on the image.

The predictive positioning of the positioning terminal is performed according to the LED classifier, and the LED positioning position of the positioning terminal is obtained.

Wherein, the final positioning position of the positioning terminal is obtained by using formula (1):

Where p denotes the final positioning position of the positioning terminal; j=1, 2; f ₁ (x) denotes a predicted positioning position based on WiFi positioning at the grid point r; f ₂ (x) denotes an LED based on the grid point r The predicted positioning position obtained by the positioning; w _r1 represents the weight value corresponding to the WiFi positioning at the grid point r; w _r2 represents the weight value corresponding to the LED positioning at the grid point r.

The embodiment of the present disclosure further relates to a storage medium, which is located at the positioning terminal side, and may be a positioning terminal, or a hardware device for data storage and processing of the positioning terminal. The storage medium stores a program for acquiring an LED positioning signal, the program comprising:

Using an positioning terminal to capture an image including an LED light;

Performing image processing to acquire a blinking frequency of each LED light in the image, and identifying the captured LED light according to the blinking frequency;

Obtaining coordinates of the identified LED light in the image, and obtaining a heading angle, a pitch angle, and a roll angle of the positioning terminal;

The heading angle, the pitch angle and the roll angle of the positioning terminal, and the coordinates of the identified LED lamp in the image form a positioning signal at the position and position of the positioning terminal.

In the above steps, image processing is performed to obtain the flicker frequency of each LED lamp in the image, including:

Converting a color image to a grayscale image;

Perform image blurring and binarization processing in sequence;

Separating the graphics corresponding to each LED light in the image, extracting the contour information of each graphic, and finding the center position;

Within each contour of the graph, the spacing of the black and white stripes is counted by the radon transform, and the blinking frequency of the LED corresponding to the contour of the graphic is calculated according to the spacing.

The present disclosure proposes a fusion positioning method and system based on WiFi and visible light, which combines the advantages of the two positioning methods, and realizes the complementary advantages of different positioning methods, thereby effectively combining WiFi positioning and visible light positioning, thereby improving positioning accuracy and stability.

While the preferred embodiment of the present disclosure has been disclosed for purposes of illustration, those skilled in the art will recognize that various modifications, additions and substitutions are possible, and the scope of the present disclosure should not be limited to the embodiments described above.

Industrial applicability

Through the technical solution provided by the present disclosure, the advantages of the two positioning methods based on WiFi and visible light are integrated, and the advantages of different positioning methods are complemented, thereby effectively combining the WiFi positioning and the visible light positioning, thereby improving the positioning accuracy and stability.

Claims (28)

1. A positioning method comprising:

   Determining and positioning the positioning terminal according to the wireless local area network WiFi positioning signal sent by the positioning terminal, and obtaining a WiFi positioning position of the positioning terminal;

   Determining and positioning the positioning terminal according to the LED LED positioning signal sent by the positioning terminal, and obtaining the LED positioning position of the positioning terminal;

   Obtaining a weight value corresponding to the two positioning positions according to the WiFi positioning position and the LED positioning position;

   The final positioning position of the positioning terminal is obtained according to the weight position corresponding to the WiFi positioning position, the LED positioning position, and the two positioning positions.
2. The positioning method according to claim 1, wherein before the positioning, the positioning area is divided into a plurality of grid points in advance, the WiFi weight values corresponding to the WiFi positioning on each grid point are obtained, and each grid point is obtained. LED weight value corresponding to LED positioning;

   In the positioning process, the WiFi weight value corresponding to the grid point closest to the WiFi positioning position is used as the weight value corresponding to the WiFi positioning position; and the grid point closest to the LED positioning position is corresponding to The LED weight value is used as the weight value corresponding to the LED positioning position.
3. The positioning method of claim 2, wherein obtaining a WiFi weight value corresponding to the WiFi location on each of the grid points comprises:

   Obtaining a WiFi fingerprint database according to the received signal strength of the set number of times collected at each grid point indicating the RSSI data;

   According to the WiFi fingerprint library, a WiFi classifier that uses the RSSI data for location prediction positioning is obtained;

   Calculating the WiFi weight value w _r1 corresponding to the WiFi location on each grid point using equations (2) and (3);

   

   ε(x _r (i))=||x _r (i)-x _r || ² Equation (3) where argmin represents the minimum value of w _r1 obtained by equation (2); ε(x _r (i)) positioned in the i th prediction data collection grid coordinates X r _r (i) with the real position coordinates X _r squared error; N is the number of samples collected at the lattice point r, r the representative grid point number, r = 1, ..., R.
4. The positioning method of claim 3, wherein the method further comprises:

   After obtaining the WiFi fingerprint database, the WiFi fingerprint library is divided into a training set and a verification set according to a set ratio;

   The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the WiFi classifier.
5. The positioning method according to claim 3 or 4, wherein the positioning terminal is predicted and located according to the WiFi classifier, and the WiFi positioning position of the positioning terminal is obtained.
6. The positioning method according to claim 2, wherein the LED weight value corresponding to the LED positioning on each grid point is obtained, including:

   According to the LED positioning data collected at each grid point, the LED fingerprint library is obtained;

   According to the LED fingerprint library, an LED classifier for position prediction positioning using LED positioning data is obtained;

   Calculate the LED weight value w _r2 corresponding to the LED positioning on each grid point using equations (3) and (4);

   

   ε(x _r (i))=||x _r (i)-x _r || ² Equation (3) where argmin represents the minimum value of w _r2 obtained by equation (2); ε(x _r (i)) positioned in the i th prediction data collection grid coordinates X r _r (i) with the real position coordinates X _r squared error; N is the number of samples collected at the lattice point r, r the representative grid point number, r = 1, ..., R.
7. The positioning method of claim 6, wherein the method further comprises:

   After obtaining the LED fingerprint library, the LED fingerprint library is divided into a training set and a verification set according to a set ratio;

   The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the LED classifier.
8. The positioning method of claim 6, wherein the LED positioning data comprises: a heading angle, a pitch angle, and a roll angle of the device that collects the LED positioning data, and one or more LED lights captured by the device The position coordinates on the image.
9. The positioning method according to claim 6 or 7 or 8, wherein the positioning terminal is predicted and positioned according to the LED classifier, and the LED positioning position of the positioning terminal is obtained.
10. The positioning method according to claim 3 or 6, wherein the final positioning position of the positioning terminal is obtained by the formula (1):

    

    Where p denotes the final positioning position of the positioning terminal; j=1, 2; f ₁ (x) denotes a predicted positioning position based on WiFi positioning at the grid point r; f ₂ (x) denotes an LED based on the grid point r The predicted positioning position obtained by the positioning; w _r1 represents the weight value corresponding to the WiFi positioning at the grid point r; w _r2 represents the weight value corresponding to the LED positioning at the grid point r.
11. A method for acquiring an LED positioning signal includes:

    Using an positioning terminal to capture an image including an LED light;

    Performing image processing to acquire a blinking frequency of each LED light in the image, and identifying the captured LED light according to the blinking frequency;

    Obtaining coordinates of the identified LED light in the image, and obtaining a heading angle, a pitch angle, and a roll angle of the positioning terminal;

    The heading angle, the pitch angle and the roll angle of the positioning terminal, and the coordinates of the identified LED lamp in the image form a positioning signal at the position and position of the positioning terminal.
12. The method for acquiring an LED positioning signal according to claim 11, wherein performing image processing to acquire a blinking frequency of each LED light in the image comprises:

    Converting a color image to a grayscale image;

    Perform image blurring and binarization processing in sequence;

    Separating the graphics corresponding to each LED light in the image, extracting the contour information of each graphic, and finding the center position;

    Within each contour of the graph, the spacing of the black and white stripes is counted by the radon transform, and the blinking frequency of the LED corresponding to the contour of the graphic is calculated according to the spacing.
13. A storage medium storing a positioning program for performing positioning, the program comprising:

    Determining and positioning the positioning terminal according to the wireless local area network WiFi positioning signal sent by the positioning terminal, and obtaining a WiFi positioning position of the positioning terminal;

    Determining and positioning the positioning terminal according to the LED LED positioning signal sent by the positioning terminal, and obtaining the LED positioning position of the positioning terminal;

    Obtaining a weight value corresponding to the two positioning positions according to the WiFi positioning position and the LED positioning position;

    The final positioning position of the positioning terminal is obtained according to the weight position corresponding to the WiFi positioning position, the LED positioning position, and the two positioning positions.
14. The storage medium of claim 13, wherein the positioning area is divided into a plurality of grid points in advance, the WiFi weight values corresponding to the WiFi positioning on each grid point are obtained, and each grid point is obtained. LED weight value corresponding to LED positioning;

    In the positioning process, the WiFi weight value corresponding to the grid point closest to the WiFi positioning position is used as the weight value corresponding to the WiFi positioning position; and the grid point closest to the LED positioning position is corresponding to The LED weight value is used as the weight value corresponding to the LED positioning position.
15. The storage medium of claim 14, wherein the obtaining a WiFi weight value corresponding to the WiFi location on each of the grid points comprises:

    Obtaining a WiFi fingerprint database according to the received signal strength of the set number of times collected at each grid point indicating the RSSI data;

    According to the WiFi fingerprint library, a WiFi classifier that uses the RSSI data for location prediction positioning is obtained;

    Calculating the WiFi weight value w _r1 corresponding to the WiFi location on each grid point using equations (2) and (3);

    

    ε(x _r (i))=||x _r (i)-x _r || ² Equation (3) where argmin represents the minimum value of w _r1 obtained by equation (2); ε(x _r (i)) positioned in the i th prediction data collection grid coordinates X r _r (i) with the real position coordinates X _r squared error; N is the number of samples collected at the lattice point r, r the representative grid point number, r = 1, ..., R.
16. The storage medium of claim 15 wherein said program comprises:

    After obtaining the WiFi fingerprint database, the WiFi fingerprint library is divided into a training set and a verification set according to a set ratio;

    The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the WiFi classifier.
17. The storage medium according to claim 15 or 16, wherein the positioning terminal is predicted and located according to the WiFi classifier, and the WiFi positioning position of the positioning terminal is obtained.
18. The storage medium of claim 14, wherein the LED weight values corresponding to the LED positioning on each grid point are obtained, including:

    According to the LED positioning data collected at each grid point, the LED fingerprint library is obtained;

    According to the LED fingerprint library, an LED classifier for position prediction positioning using LED positioning data is obtained;

    Calculate the LED weight value w _r2 corresponding to the LED positioning on each grid point using equations (3) and (4);

    

    ε(x _r (i))=||x _r (i)-x _r || ² Equation (3) where argmin represents the minimum value of w _r2 obtained by equation (2); ε(x _r (i)) positioned in the i th prediction data collection grid coordinates X r _r (i) with the real position coordinates X _r squared error; N is the number of samples collected at the lattice point r, r the representative grid point number, r = 1, ..., R.
19. The storage medium of claim 18, wherein the program further comprises:

    After obtaining the LED fingerprint library, the LED fingerprint library is divided into a training set and a verification set according to a set ratio;

    The support vector machine algorithm SVM nonlinear classifier is selected and trained in the Python language to obtain the LED classifier.
20. The storage medium of claim 19, wherein the LED positioning data comprises: a heading angle, a pitch angle, and a roll angle of the device that collects the LED positioning data, and one or more LED lights captured by the device The position coordinates on the image.
21. The storage medium according to claim 18 or 19 or 20, wherein the positioning terminal is predicted and positioned according to the LED classifier to obtain an LED positioning position of the positioning terminal.
22. The storage medium according to claim 15 or 18, wherein the final positioning position of the positioning terminal is obtained by the formula (1):

    

    Where p denotes the final positioning position of the positioning terminal; j=1, 2; f ₁ (x) denotes a predicted positioning position based on WiFi positioning at the grid point r; f ₂ (x) denotes an LED based on the grid point r The predicted positioning position obtained by the positioning; w _r1 represents the weight value corresponding to the WiFi positioning at the grid point r; w _r2 represents the weight value corresponding to the LED positioning at the grid point r.
23. A storage medium storing a program for acquiring an LED positioning signal, the program comprising:

    Using an positioning terminal to capture an image including an LED light;

    Performing image processing to acquire a blinking frequency of each LED light in the image, and identifying the captured LED light according to the blinking frequency;

    Obtaining coordinates of the identified LED light in the image, and obtaining a heading angle, a pitch angle, and a roll angle of the positioning terminal;

    The heading angle, the pitch angle and the roll angle of the positioning terminal, and the coordinates of the identified LED lamp in the image form a positioning signal at the position and position of the positioning terminal.
24. The storage medium of claim 23, wherein performing image processing to acquire a blinking frequency of each of the LED lights in the image comprises:

    Converting a color image to a grayscale image;

    Perform image blurring and binarization processing in sequence;

    Separating the graphics corresponding to each LED light in the image, extracting the contour information of each graphic, and finding the center position;

    Within each contour of the graph, the spacing of the black and white stripes is counted by the radon transform, and the blinking frequency of the LED corresponding to the contour of the graphic is calculated according to the spacing.
25. A positioning system includes a plurality of wireless routers disposed in a positioning area, a plurality of LED lights, and a positioning terminal, a monitoring terminal, and a server:

    The positioning terminal collects the WiFi positioning signal and the LED positioning signal of the location grid point, and sends the positioning signal to the server;

    The server locates the positioning terminal according to the WiFi positioning signal and the LED positioning signal, and sends the positioning information to the monitoring terminal;

    The monitoring terminal receives the positioning information and displays the location of the positioning terminal.
26. The positioning system according to claim 25, wherein the server comprises the storage medium according to any one of claims 13 to 22.
27. A positioning system according to claim 25, wherein said positioning terminal comprises the storage medium of claim 23 or 24.
28. A positioning system according to claim 25, wherein the light modulation of said LEDs is a square wave signal, and the modulation frequencies of the respective LED lamps are different.

Images